Desmosome assembly may be induced in simple epithelial (MDBK and MDCK) cells maintained in low calcium medium (LCM: [Ca2+] <0·05mM) by raising [Ca2+] to that of standard culture medium (SM: [Ca2+] = l·8mM). Here it is shown that if cells in SM are simply returned to LCM, their desmosomes split in the intercellular region within 15 min and the desmosomal halves are internalized within 30 min. This is the first time that desmosome splitting has been shown to occur in response to a reduction in [Ca2+] rather than Ca2+ chelation. Fluorescent antibody staining shows that the desmosomal glycoproteins as well as the plaque constituents are internalized, although a pool of the glycoproteins known as desmocollins remains at the cell surface, apparently unassociated with other desmosomal components. Desmosomal halves that have been recently internalized in response to LCM treatment do not return to the cell surface to participate in new desmosome formation. MDCK cells are able to form new desmosomes rapidly (15—30 min) while old desmosomes continue to be internalized.

The desmosomes of MDBK cells remain sensitive to splitting and internalization in response to reduction in [Ca2+] for up to 14 days of culture in SM. In contrast, the desmosomes of MDCK cells become resistant to reduction in [Ca2+], as well as Ca2+ chelation by EGTA, after 4–5 days in SM. When treated with LCM or EGTA, MDCK cells with ‘stabilized’ desmosomes partially separate but remain attached to each other at some points. Regions of attachment stain brightly with anti-demosomal antibodies and are characterized by ‘giant’ desmosomes, up to 4/tm long, roughly 20 times larger than those formed in cells in SM. These giant desmosomes may form by lateral fusion of small desmosomes.

The purpose of this paper is to present some new observations on the breakdown of desmosomes in response to reduction in [Ca2+]. Previous work has shown that the chelating agent EDTA causes the desmosomes of some intact tissues to split in the intercellular region. This is the case with desmosomes in simple epithelia (Borysenko & Revel, 1973). Desmosomes in stratified epithelia, on the other hand, require trypsin digestion in order to split them (Borysenko & Revel, 1973; Overton, 1962). Many cultured cell types are routinely dissociated by treatment with a mixture of trypsin and EDTA, although the desmosomes (and zonulae adhaerentes) of MDBK cells may be split by treatment with the calcium-specific chelating agent EGTA (Kartenbeck et al. 1982; Cowin et al. 1984). However, simple epithelial cells in culture may be induced to form desmosomes by raising the [Ca2+] from 0·05 mM to 1 ·8 mM. It is of interest, therefore, to determine whether desmosome splitting may be induced simply by placing the cells at a [Ca2+] of 0’05 mM, or whether actual calcium chelation is required, as implied by previous studies. Calcium-induced desmosomes of human kératinocytes have been shown to remain susceptible to splitting by EDTA treatment for 2h after their formation, but thereafter to become resistant to splitting by this reagent (Watt et al. 1984).

A consequence of desmosomal splitting is that the cells internalize the half-desmosomes, which are no longer linked by intercellular bonding (Overton, 1968, 1973; Overton & Culver, 1973; Fukuyama et al. 1974; Shimono & Clementi, 1977; Kartenbeck et al. 1982). Internalized desmosomes of MDBK cells have been shown to react with antibodies to desmoplakins, the high molecular weight desmosomal proteins (Kartenbeck et al. 1982). However, Cowin et al. (1984) showed that the desmocollins, which are desmosomal glycoproteins, remain on the surface of MDBK cells after EGTA treatment. Clearly, a fuller study of the behaviour of desmosomal components during internalization is required, especially since the fate of the other major desmosomal components is unknown.

In this paper we show that the desmosomes of MDBK and MDCK cells may be split simply by reducing [Ca2+] from 1·8mM to 0·05 mM. This remains true for MDBK cells for at least 14 days after desmosome formation. However, the desmosomes of MDCK cells become resistant to reduction in [Ca2+] and even to divalent cation chelation, after approximately 4 days in culture. Fluorescent antibody staining is used to study the behaviour of various desmosomal components during internalization of half-desmosomes. We also show that cells can internalize ‘old’ desmosomes and form ‘new’ ones simultaneously.

The cells, antibodies and techniques used in the present paper are identical to those used in the accompanying paper (Mattey & Garrod, 1986).

Internalization of desmosomes

Electron microscopy. MDBK and MDCK cells were placed in standard medium (SM: [Ca2+] = 1·8mM) for 24—48h, ample time for desmosome formation (Mattey & Garrod, 1986). They were then transferred into low calcium medium (LCM: [Ca2+] = 0·05 mM) and fixed for electron microscopy at various times. In both cell types desmosome splitting occurred within 15 min (Fig. 1A). After 30min many of these half-desmosomes had become internalized in vesicular structures (Fig. IB). Plaque material and tonofilaments remained attached to the cytoplasmic side of the internalized membranes. With further time in LCM these internalized structures became detectable deep within the cytoplasm of the cells in close association with bundles of intermediate filaments.

Fig. 1.

Electron micrographs showing splitting and internalization of desmosomes of MDCK cells transferred from SM to LCM for 15 min (A), 30 min (B) and 3h (C,D). A. Note fibrillar intercellular material at splitting desmosomes. The desmosomal intercellular space is much wider than that of mature desmosomes of cells in SM (Mattey & Garrod, 1986, fig. 6C). B. Note close association of internalized desmosomes (arrows) with intermediate filaments. C,D. Higher-power photographs showing typical examples of internalized desmosomes. Bars: A, 0·3·m; B, 0·4·m; C, 0·2·m; D, 0·2·m.

Fig. 1.

Electron micrographs showing splitting and internalization of desmosomes of MDCK cells transferred from SM to LCM for 15 min (A), 30 min (B) and 3h (C,D). A. Note fibrillar intercellular material at splitting desmosomes. The desmosomal intercellular space is much wider than that of mature desmosomes of cells in SM (Mattey & Garrod, 1986, fig. 6C). B. Note close association of internalized desmosomes (arrows) with intermediate filaments. C,D. Higher-power photographs showing typical examples of internalized desmosomes. Bars: A, 0·3·m; B, 0·4·m; C, 0·2·m; D, 0·2·m.

Internalized desmosomes showed a variety of forms. These included circular and crescent-shaped segments of membrane associated with tonofilaments and aggregates of electron-dense material (Fig. 1C,D) or, more rarely, paired plaque structures (see Mattey & Garrod, 1986, fig. 4).

Immunofluorescence microscopy

The staining patterns produced by the various anti-desmosomal antibodies on MDBK and MDCK cells maintained in SM are described in the accompanying paper (Mattey & Garrod, 1986). In both cell types the boundaries between cells in contact are stained by the antibodies. This staining may appear punctate with anti-desmoplakin (figs IC, 6C, Mattey & Garrod, 1986) and anti-175—164 K, or linear with anti-desmocollin (fig. 6L, Mattey & Garrod, 1986) and anti-83/75 K (K represents 103Mr).

The sequence of changes that occurred when MDCK cells that had been maintained in SM for less than 3 days were transferred into LCM, as revealed by anti-desmoplakin staining, is shown in Fig. 2. At 5 min some gaps appeared between the cells and the separated surfaces stained brightly, the staining presumably representing separated desmosomal halves (Fig. 2A). By 15 min the bright ring of desmoplakin staining in each cell was separated from that of its neighbours (Fig. 2B). One or two regions of contact persisted. By 30min each fluorescent ring became much smaller in diameter, in some cases apparently surrounding the cell nucleus (Fig. 2C). The rings also began to fragment into dots of staining. Electron microscopy showed that internalization of desmosomal remnants had occurred by this time. By 4 h staining was rarely present in the form of a ring. Instead there were many small dots together with one or two larger bright spots adjacent to the nucleus (Fig. 2D). The cells at this stage resembled cells that had been plated into LCM directly (Mattey & Garrod, 1986). Similar patterns of invagination were seen with other desmosomal components (Fig. 2E,F,G). However, as well as an internalized ring, anti-desmocollin staining of living or formaldehyde-fixed cells revealed that some of these antigens persisted on the surface (Fig. 2H). The internalized ring of desmoplakin staining was co-localized with a ring of anti-cytokeratin staining (Fig. 3A,B).

Fig. 2.

Fluorescence photomicrographs showing changes in patterns of staining with anti-desmosomal antibodies in MDCK cells transferred from SM to LCM for: A, 5 min; B, 15 min; C,E,F,G,H, 30min; and D, 4h. A-G. Methanol fixation showing cytoplasmic staining; H, formaldehyde fixation showing surface staining. Antibodies used were: A-D, anti-desmoplakin; E, anti-175–164 K; F, anti-83/75 K; G,H, anti-desmocollin. Bar, 20μm.

Fig. 2.

Fluorescence photomicrographs showing changes in patterns of staining with anti-desmosomal antibodies in MDCK cells transferred from SM to LCM for: A, 5 min; B, 15 min; C,E,F,G,H, 30min; and D, 4h. A-G. Methanol fixation showing cytoplasmic staining; H, formaldehyde fixation showing surface staining. Antibodies used were: A-D, anti-desmoplakin; E, anti-175–164 K; F, anti-83/75 K; G,H, anti-desmocollin. Bar, 20μm.

Fig. 3.

Fluorescence micrographs showing co-localization of internalized desmoplakin staining with a ring of cytokeratin staining in an MDCK cell transferred from SM to LCM for 30 min. A. Stained with monoclonal antibody to desmoplakin I; B, stained with guinea-pig anti-cytokeratin. Methanol fixation. Bar, 20/.μm.

Fig. 3.

Fluorescence micrographs showing co-localization of internalized desmoplakin staining with a ring of cytokeratin staining in an MDCK cell transferred from SM to LCM for 30 min. A. Stained with monoclonal antibody to desmoplakin I; B, stained with guinea-pig anti-cytokeratin. Methanol fixation. Bar, 20/.μm.

In MDBK cells a similar sequence of events occurred, but was rather less striking because there were fewer desmosomes. Peripheral rows of anti-desmoplakin or anti-175–164 staining dots were present on separated cells after 10-15 min, presumably representing half-desmosomes (Fig. 4A). Such dots were found within the cytoplasm of cells that had been treated with LCM for longer periods (Fig. 4B). However, they did not accumulate in a juxtanuclear spot. Internalization of anti-83/75 and anti-desmocollin staining occurred initially as a continuous ring (Fig. 4C,D). Some punctate cytoplasmic staining was also seen with anti-83/75 K (Fig. 4D) but not with anti-desmocollin. Cytoplasmic staining for desmocollins became undetectable after 4–6 h. However, desmocollin staining persisted on the cell surface in a diffuse distribution, similar to that shown by MDBK cells that have been cultured in LCM (fig. 7A, Mattey & Garrod, 1986).

Fig. 4.

Fluorescence micrographs showing changes in patterns of staining with anti-desmosomal antibodies in MDBK cells transferred from SM to LCM for: A, 15 min; B, 24h; C,D, 45 min. Antibodies used are: A,B, anti-desmoplakin; C, anti-desmocollin; D, anti-83/75 K. Methanol fixation. Bar, 20μm.

Fig. 4.

Fluorescence micrographs showing changes in patterns of staining with anti-desmosomal antibodies in MDBK cells transferred from SM to LCM for: A, 15 min; B, 24h; C,D, 45 min. Antibodies used are: A,B, anti-desmoplakin; C, anti-desmocollin; D, anti-83/75 K. Methanol fixation. Bar, 20μm.

Re-formation of desmosomes

MDBK and MDCK cells cultured for 48 h in SM were permitted to internalize their desmosomes by switching the cells to LCM for 45 min. Cells were then switched back to SM and fixed in methanol at various times before staining with anti-desmoplakin antibody.

In MDCK cells desmoplakins became detectable at cell boundaries within 20–30 min of switching back into SM. The previously internalized desmoplakins remained as a perinuclear ring in the cytoplasm. The desmoplakins thus appeared as a double ring with staining at the cell boundaries due to newly formed desmosomes and intra-cytoplasmic staining due to internalized desmosomes (Fig. 5A). The internalized components accumulated in a juxtanuclear region (Fig. 5B) and eventually disappeared after 4 or 5h (Fig. 5C). In contrast, desmoplakin staining in MDBK cells was first seen at cell boundaries only after 3–4 h of re-switching into SM. By 6h the cell-boundary staining was more pronounced although dotted staining of internalized desmosomal components was still present in the cytoplasm well away from the cell periphery (Fig. 5D).

Fig. 5.

Fluorescence micrographs showing simultaneous formation of new desmosomes and internalization of old desmosomes in MDCK cells (A,B,C) and MDBK cells (D) transferred from SM to LCM for 45min and then returned to SM for: A, 1 h; B, 3h; C, 5h; D, 6h. Staining of methanol-fixed cells with guinea-pig anti-desmoplakin. Internalized material marked with arrowheads and staining at cell periphery with arrows. The punctate staining areas in C (arrowheads) are regions of intercellular contact viewed obliquely, not internalized material. Bar, 20μm.

Fig. 5.

Fluorescence micrographs showing simultaneous formation of new desmosomes and internalization of old desmosomes in MDCK cells (A,B,C) and MDBK cells (D) transferred from SM to LCM for 45min and then returned to SM for: A, 1 h; B, 3h; C, 5h; D, 6h. Staining of methanol-fixed cells with guinea-pig anti-desmoplakin. Internalized material marked with arrowheads and staining at cell periphery with arrows. The punctate staining areas in C (arrowheads) are regions of intercellular contact viewed obliquely, not internalized material. Bar, 20μm.

Stabilization of desmosomes

MDBK cells cultured for up to 14 days in SM always showed internalization of desmoplakins within 2h of switching from SM to LCM. However, the same was true for MDCK cells for only 3 days after plating. After 4–5 days of culture cells switched into LCM pulled apart from each other to some extent, but remained attached at regions around the periphery. There was some internalization of desmosomal material as demonstrated by desmoplakin staining, but desmoplakin staining persisted at the regions of cell contact, even after 24h in LCM (Fig. 6A). The cytokeratin filaments became organized into large bundles at those regions of contact (Fig. 6C,D). After 7 days in culture there was little or no internalization of desmoplakin-staining material, but instead very brightly staining concentrations of desmosomal material were found at persistent regions of cell contact (Fig. 6B). Examination of the regions of contact between these cells by electron microscopy revealed that they consist of ‘giant’ desmosomes (Fig. 6E) up to 4μm long, and therefore about 20 times larger than those found in SM.

Fig. 6.

A-D. Fluorescence micrographs showing desmosomal material in MDCK cells maintained in SM for 4 days (A,C,D) or 7 days (B) and transferred into LCM for 24 h. Methanol fixation staining with guinea-pig anti-desmoplakin (A,B), monoclonal anti-desmoplakin I (C) and guinea-pig anti-cytokeratin (D). C,D. Showing double staining of the same cells. E. Electron micrograph showing ‘giant’ desmosome in region of contact between MDCK cells maintained in SM for 4 days and transferred to LCM for 24 h. Bars: A-D, 20μm; E, 0·3μm

Fig. 6.

A-D. Fluorescence micrographs showing desmosomal material in MDCK cells maintained in SM for 4 days (A,C,D) or 7 days (B) and transferred into LCM for 24 h. Methanol fixation staining with guinea-pig anti-desmoplakin (A,B), monoclonal anti-desmoplakin I (C) and guinea-pig anti-cytokeratin (D). C,D. Showing double staining of the same cells. E. Electron micrograph showing ‘giant’ desmosome in region of contact between MDCK cells maintained in SM for 4 days and transferred to LCM for 24 h. Bars: A-D, 20μm; E, 0·3μm

The time of onset of desmosomal stability was dependent upon cell seeding density. The above remarks apply to cells seeded at a density of 104cm-2 Cells seeded at lower density took about 24 h longer to acquire desmosomal stability. However, cells seeded at confluent density initially still took about 4–5 days to acquire stability.

The stabilized desmosomes were not only resistant to reduction in [Ca2+], but also not broken down by treatment with ImM-EGTA for lh. EGTA treatment eventually caused the cells to detach from the substratum as a sheet.

Desmosome splitting and desmosome stability

Our results show that desmosomes formed by MDBK and MDCK cells in the presence of 1·8 mM-Ca2+ may be split and internalized simply as a result of reduction in [Ca2+] to 0·05 mM. Splitting does not require treatment with chelating agents such as EDTA or EGTA. The desmosomes of MDBK cells remain susceptible to splitting by reduced [Ca2+] for at least 14 days in culture. However, the desmosomes of MDCK cells acquire resistance to splitting, even by treatment with EGTA, from about 4 days of culture in SM. The desmosomes are then susceptible to splitting by trypsin plus EDTA as used in routine passaging.

We have previously shown that kératinocytes acquire resistance to desmosomal splitting by EDTA within 3 h of transfer into high [Ca2+], and consequently we have suggested that kératinocyte desmosomes undergo stabilization (Watt et al. 1984). MDCK desmosomes may also undergo stabilization, but initiated later and taking a considerably longer time to occur. We have no indication of the biochemical nature of the stabilizing event.

The effect of reduced [Ca2+] on MDBK cells and on MDCK cells cultured for less than 4 days appears to be twofold. First the desmosomes split, and second the cells round up. The observation that desmosomes are split at [Ca2+] of 0·05 mM may indicate that the adhesive bond in the desmosomes is extremely sensitive to [Ca2+]. Whatever the nature of the stabilization that occurs in MDCK cells, it appears to overcome this sensitivity to [Ca2+] : in cells with stable desmosomes, transfer to LCM still appears to have the same general effect on cell rounding, but the desmosomes do not split. The alteration in cell shape probably involves both the cytoskeleton and the splitting of intercellular junctions other than desmosomes. However, we have not studied these in detail.

The most striking feature of MDCK cells with stable desmosomes in LCM is that as the cells round up desmosomal staining accumulates in remaining regions of the intercellular contact. These regions are found to consist of massive desmosomes on electron-microscopical examination. We suggest that as the cells round up and pull apart from each other, the desmosomes, which must be mobile within the cell membrane, are drawn together and fuse laterally. The cytokeratin filament bundles that are attached to the desmosomal plaques are thereby also drawn together into larger bundles. Mobility of desmosomes within the cell membrane has been suggested previously on the basis of indirect evidence by Klymkowsky et al. (1983) who showed that if the cytokeratin system of one of an attached pair of cells was disrupted by injection of anti-cytokeratin antibody, the cytokeratin filaments in the other cell became bundled together at a point in the region of contact between the two. They inferred that the desmosomes had accumulated at this point. Our present results suggest that the distribution of desmosomes is determined by some aspect of overall cellular shape organization, possibly involving the cytoskeleton. Furthermore, the size of desmosomes appears to be determined in relation to their distribution on the cell surface rather than by some size-determining mechanism intrinsic to them. Thus, when cellular shape organization is disrupted, desmosomes are able to accumulate and to fuse, generating giant desmosomes. It may be significant that the desmosomes of carcinoma cells are sometimes found to be enlarged compared with those of their normal counterparts (Pauli et al. 1978; Hand, Garrod & Parry, unpublished observations). This may reflect lateral desmosomal fusion as a result of cell shape disorganization.

Desmosome internalization and re-formation

Internalization of desmosomes in MDBK cells has been shown previously to involve internalization of desmoplakins (Kartenbeck et al. 1982), while some des-mocollin staining remains at the cell surface (Cowin et al. 1984). Here we confirm that desmocollin staining persists at the cell surface, but that some internalization of desmocollin staining also occurs, together with staining for all other desmosomal antigens. This is the case for both MDBK and MDCK cells.

Apart from the greater number of desmosomes and therefore greater quantity of staining in MDCK cells, there are two principal differences between the two cell types. First, the tendency was for internalized components to collect in one or more juxtanuclear spots in MDCK cells, whereas in MDBK cells they were dispersed as discrete dots within the cytoplasm. Second, the internalized components persisted in the cytoplasm of MDCK cells even when they were cultured for long periods in LCM, whereas with MDBK cells, staining for internalized components disappeared after about 4 days in LCM.

In the accompanying paper (Mattey & Garrod, 1986) we suggested that calcium-induced desmosome formation in MDCK cells did not seem to involve rapid movement of cytoplasmic desmosomal particles from the nuclear region to the cell periphery as has been suggested by Jones & Goldman (1985) for mouse kératinocytes. This view is reinforced by the observation that ‘old’ desmosomes continue to be internalized while ‘new’ desmosomes form. This raises the question of the origin of the components that form the new desmosomes. Presumably, pools of unassembled desmosomal proteins are available within these cells. However, they are apparently not detectable by fluorescent antibody staining until they reach the cell periphery to participate in desmosome formation.

Lastly, the appearance of MDBK cells shortly after being transferred from SM to LCM is different from that of cells cultured for longer periods in LCM (Mattey & Garrod, 1986). The latter have peripheral rings of anti-desmocollin and anti-83/75 staining where cells are in contact. We assume that the appearance of these antigens at the cell periphery involves a ‘recovery’ phase, a type of adaptation to culture in LCM in which cell contact is re-established. Zonulae adhaerentes junctions, but not desmosomes, are able to form under these conditions (Mattey & Garrod, 1986). As discussed in the accompanying paper, further investigations are needed into the possibility of an association between the desmocollins, the 83/75 K protein and the zonulae adhaerentes in these cells.

We thank Dr Terry Kenny for suggesting improvements to the manuscript. This work was supported by the Cancer Research Campaign.

Borysenko
,
J. Z.
&
Revel
,
J.-P.
(
1973
).
Experimental manipulation of desmosome structure
.
J. Anat
.
137
,
403
422
.
Cowin
,
P.
,
Mattey
,
D. L.
&
Garrod
,
D. R.
(
1984
).
Identification of desmosomal surface components (desmocollins) and inhibition of desmosome formation by specific Fab’
.
J. Cell Sci
.
70
,
41
60
.
Fukuyama
,
K.
,
Black
,
M. M.
&
Epstein
,
U. L.
(
1974
).
Ultrastructural studies of newborn rat epidermis after trypsinization.y
.
Ultrastruct. Res
.
46
,
219
229
.
Jones
,
J. C. R.
&
Goldman
,
R. D.
(
1985
).
Intermediate filaments and the initiation of desmosome assembly. J’
.
Cell Biol
.
101
,
506
517
.
Kartenbeck
,
J.
,
Schmid
,
E.
,
Franke
,
W. W.
&
Geiger
,
B. M.
(
1982
).
Different modes of internalization of proteins associated with adhaerens junctions and of desmosomal plaque material
.
EMBOJ
.
1
,
725
732
.
Klymkowsky
,
M. W.
,
Miller
,
R. H.
&
Lane
,
E. B.
(
1983
).
Morphology, behaviour and interaction of cultured epithelial cells after the antibody-induced disruption of keratin filament organization.J
.
Cell Biol
.
96,-
495
509
.
Mattey
,
D. L.
&
Garrod
,
D. R.
(
1986
).
Calcium-induced desmosome formation in cultured kidney epithelial cells
.
J. Cell Sci
.
85
,
95
111
.
Overton
,
J.
(
1962
).
Desmosome development in normal and reassociating cells of the early chick blastoderm
.
Devi Biol
.
4
,
532
548
.
Overton
,
J.
(
1968
).
The fate of desmosomes in trypsinized tissue. J’
,
exp. Zool
.
168
,
203
213
.
Overton
,
J.
(
1973
).
Experimental manipulation of desmosome formation
.
J. Cell Biol
.
56
,
636
646
.
Overton
,
J.
&
Culver
,
N.
(
1973
).
Desmosomes and their components after cell dissociation and reaggregation in the presence of cytochalasin B
.
J. exp. Zool
.
185
,
341
356
.
Pauli
,
B. U.
,
Cohen
,
S. M.
,
Alroy
,
J.
&
Weinstein
,
R. S.
(
1978
).
Desmosome ultrastructure and the biological behaviour of chemical carcinogen-induced urinary bladder carcinomas
.
Cancer Res
.
38
,
3276
3285
.
Shimono
,
M.
&
Clementi
,
F.
(
1977
).
Intercellular junctions of oral epithelium. II. Ultrastructural changes in rat buccal epithelium induced by trypsin digestion
.
J. Ultrastruct. Res
.
59
,
101
112
.
Watt
,
F. M.
,
Mattey
,
D. L.
&
Garrod
,
D. R.
(
1984
).
Calcium-induced reorganization of desmosomal components in cultured human kératinocytes
.
J. Cell Biol
.
99
,
2211
2215
.