Nature Structural Biology
9, 560 - 562 (2002)
doi:10.1038/nsb0802-560
A new fold on an old story: attachment of intermediate filaments to
desmosomesPierre A. CoulombePierre A. Coulombe is in the Department of
Biological Chemistry, Johns Hopkins University School of Medicine,
Baltimore, Maryland 21205, USA.
coulombe@jhmi.edu. Desmoplakin is a major and essential constituent of the interface
between the transmembrane glycoproteins mediating adhesion and the intermediate
filament cytoskeleton in the cell interior. The structures of two desmoplakin
domains that participate in intermediate filament binding reveal that a unique
38-amino acid repeat-containing domain shared with related proteins packs in a
novel globular fold.Desmosomes are prominent cell−cell adhesion devices that make an
essential contribution to the cohesiveness, resilience and integrity of
numerous tissues. Accordingly, they are especially frequent in tissues
subjected to significant physical trauma, including skin, muscle and vascular
endothelium1. In addition to this structural role, desmosomes
also promote the maturation of adherens junctions, influence the dynamic
properties of the cytoskeleton2 and the morphogenesis of
glandular epithelia3. Three types of proteins are required to
make up a functional desmosome: (i) type I transmembrane glycoproteins of the
cadherin superfamily, the desmogleins and desmocollins, which mediate
cell−cell attachment on the extracellular side; (ii) cytoplasmic
intermediate filaments (IFs), to which desmosomes are attched on the
intracellular side (Fig. 1a,b); and (iii) a
complex macromolecular interface connecting desmosomal cadherins and
intermediate filaments1. One of the key, invariable components of
this interface is desmoplakin (DP)4,
5, a protein essential for
the assembly and structure of desmosomes2,
6,
7,
8 and their
attachment to cytoplasmic IFs6,
8,
9,
10. On
page 612 of this issue of Nature Structural
Biology, Choi et al.11 present the high-resolution
structures of domains that are responsible for DP−IF binding at the
desmosomal plaque (Fig. 1b,c); each of
these domains consist of four and a half copies of a unique 38-amino acid
repeat motif 5. At least three features make this an important
advance: (i) this domain exhibits a new fold with only distant structural
homology to ankyrin repeats; (ii) it provides the first high-resolution
structure of an IF-binding domain; and (iii) it offers new insight into how
desmoplakin and related proteins have evolved the ability to bind IFs.
 | | Figure 1. Domain structure and subcellular localization of
desmoplakin. | | |  | a, Fluorescence micrograph of epidermal
keratinocytes in culture that were fixed and stained for DNA (blue), keratin
filaments (red), and desmoplakin (green) using standard reagents and methods in
the context of a triple-fluorescence labeling experiment. The two boxed areas
highlight desmoplakin-containing desmosomes, which appear bright yellow due to
the significant overlap of the signals emitted by the red (keratin IFs) and
green (desmoplakin) fluorochromes. Adapted from ref. 1, with permission. b, Cross-section through a
single desmosome contact made by two neighboring epithelial cells, as revealed
by transmission electron microscopy. Three distinct zones can be recognized in
a desmosome prepared in this manner: a thickening of the plasma membrane (PM),
which reflects in part the lateral packing of desmosomal cadherins; a plaque
region, which contains desmoplakin and other key proteins and serves to anchor
the third element, intermediate filaments. c, Domain organization of
desmoplakin I and II. Of particular interest here are the plakin repeat domains
A, B and C at the C-terminus of these two proteins, which are capable of
intermediating filament binding. See refs 15,18 for additional
information.
 Full Figure and legend (69K) |
| |
Desmoplakin: a fascinating protein
Along with plectin12 and the bullous pemphigoid antigen 1
(BPAG1)13, desmoplakin is a founding member of the plakin family
of proteins14. Plakins are generally involved in tethering major
cytoskeletal fibers, including intermediate filaments, F-actin and
microtubules, to one another and/or to adhesion complexes beneath the plasma
membrane15. The name 'plakin' was coined16 to
reflect the presence of many plakin proteins — at least as they were
known early on — within the plaque domain of desmosomes and
hemidesmosomes. The oldest plakin gene, from an evolutionary standpoint,
encodes kakapo/SHORTSTOP, a protein essential for the
integrin-based attachment of epidermis to the underlying muscle in
Drosophila15,
17,
18. For each of DP, plectin and BPAG1,
alternative splicing of a single mRNA precursor generates two or more protein
isoforms with distinct domains and properties. Constant elements in all splice
isoforms of these three proteins include a coiled-coil dimerization interface
and a common type of IF-binding domain within the C-terminus12,
15,
18. In addition to this multidomain organization and a related
genomic structure, nearly all plakin proteins contain one or several copies of
a homologous domain, the plakin repeat domain (PRD), consisting of four and a
half copies of a unique 38-amino acid repeat motif. This signature domain
likely makes a significant contribution to IF binding, and three of them are
found in the C-terminus of desmoplakin5.
A crucial role for plakins toward various aspects of cytoarchitecture,
cell adhesion and cytoskeletal dynamics in vivo has been inferred from
studies of mouse models whose genomes carry a null allele and the
characterization of patients with desmoplakin defects resulting from inherited
and autoimmune mechanisms1,
19. How plakin proteins function is
thus fairly well understood from molecular genetic and cell biological
perspectives. What has been lacking is the deeper understanding that comes with
high-resolution structural data.
Alternative splicing of the desmoplakin precursor mRNA generates the two
known DPs that differ in size (2,871 residues in DP I versus 2,232
residues in DP II) and cell type distribution. DP I and II have identical N-
and C-termini (1,056 and 926 amino acids, respectively; ref. 18) but their central  -helical domains are of
different lengths (Fig. 1c). Inactivation of the
mouse DP gene is lethal at an early stage of embryogenesis, likely due to
defects in extra-embryonic tissues8. Sophisticated follow-up
studies in transgenic mice revealed the crucial role of DP in tethering IFs to
desmosomes in embryonic heart, neuroepithelium, and vasculature, as well as in
embryonic and adult skin2,
20. In humans, desmoplakin
haploinsufficiency underlies a dominantly inherited disorder known as striate
plamoplantar keratoderma, in which an odd, linear pattern of skin thickening
occurs on palms and soles21. Moreover, recessive inheritance of a
frameshift mutation causing a DP truncation just N-terminal to the third PRD
disrupts IF−desmosome interactions and results in dilated cardiomyopathy,
wooly hair and keratoderma22. These observations extended earlier
studies involving experimental mutagenesis9,
10, which indicated
that IF binding involves the C-terminus of DP and related plakin family
members13,
23,
24. Altogether, these various lines of evidence
establish that the presence of intact desmoplakin in sufficient amounts is
crucial for the assembly, structure and/or function of desmosomes.
The signature repeat domain has a new fold
Given this knowledge base, the stage was set to undertake the
characterization of desmoplakin at a structural level. The team led by Bill
Weis at Stanford University and Kathleen Green at Northwestern University
Medical School successfully determined the structure of two of the three PRDs
within DP's C-terminus. The authors first established the physical boundaries
of PRD A, B and C by subjecting the purified recombinant full-length DP
C-terminus to limited proteolysis. These boundaries conformed reasonably well
to predictions made at an earlier time based on sequence analysis5,
14. Next, the authors went on to successfully crystallize PRD B and
C, and solve their structure at 3.0 Å and 1.8 Å resolution,
respectively.
The structure of the 38-amino acid repeat consists of an 11-residue
 -hairpin followed by two antiparallel -helices that are typically
8 and 14 residues long (Fig. 2a,b). This
represents a new fold but exhibits some features in common with the structure
adopted by ankyrin repeats. Each of the repeats within a PRD adopts a highly
analogous fold, although there are important differences in the spatial
orientation of specific elements, notably, the second -helix of repeats
2 and 4 (Fig. 2b). Repeats 1 and 2 are roughly
parallel to each other, whereas repeat 2 is packed against repeat 3 with a
120° angle (Fig. 2a). The second 14-residue
-helix plays a key role in defining the overall structure of the PRD:
its first half contributes to the formation of a hydrophobic core within each
38-residue repeat, and its second half makes contacts with the next repeat.
Moreover, the half repeat 5 has an atypical C-terminal moiety that is followed
by two antiparallel -strands, the second of which mediates an interaction
with repeat 1 (Fig. 2a). Several of the conserved
hydrophobic residues lie on the surface of the folded 38-residue repeat, and
participate in contacts with neighboring repeats. These unusual features
promote the adoption of a globular, rather than linear, shape by the entire
PRD. They also make it unlikely that a repeat would be stable as a single
entity. As expected from sequence gazing, the overall fold of the two PRDs, B
and C, is conserved. The authors speculate that the multiple PRDs present
within the C-terminus of DP and in other plakin family members form independent
domain structures, with a 'beads-on-a-string' arrangement similar to that seen
in fibronectin and other proteins with tandemly repeated domains. A complete
description of the remarkably interesting features of this structure can be
found in their paper11.
| | Figure 2. Structure of a plakin repeat domain within desmoplakin. | | |  | a, Ribbon diagram of plakin repeat domain C. The
38-residue plakin repeats 1−5 contained in this domain are colored red,
magenta, blue, green and yellow, respectively, and non-repeat regions are shown
in grey. The N- and C-termini, and the breaks in the polypeptide backbone are
indicated. b, Overlay of the four full repeats (1−4) of plakin
repeat domain B, colored as in (a), showing their structural similarity.
The most striking differences are the different spatial orientation adopted by
helix 2 in repeats 2 and 4. c, Surface representation of
plakin repeat domain C. The surface is colored to reflect electrostatic
potential, with blue and red representing positive and negative regions,
respectively. White surfaces are electrostatically neutral. See Choi et
al.11 in this issue for additional details.
Full Figure and legend (147K) |
| |
Implications on IF binding
Unlike the conventional filament-forming actins, which are highly
conserved in sequence and interact with a vast number of associated proteins,
there is considerable sequence heterogeneity within the IF superfamily and
comparatively few IF-associated proteins (IFAPs)25. Yet, both
F-actin and IFs are believed to function largely as bundles or crosslinked
networks in the cytoplasm. Relatively little is actually known about the
specifics defining the interactions between IFAPs, including DP and other
plakins, and their IF targets. Many 'structural IFAPs', here defined as
proteins effecting IF organization25, can bind several types of
IFs in vitro and in vivo. For instance, DP itself binds vimentin,
desmin and various types of keratin filaments, and some of these interactions
involve distinct regions within each protein partner9,
10,
26. It
seems unlikely, therefore, that the binding determinants correspond to discrete
elements contained within linear sequences. In the specific cases of filaggrin
(an IFAP found in terminally differentiating keratinocytes in epidermis) and
plectin, studies have shown that electrostatic interactions influence IF
binding23,
27. Moreover, phosphorylation of a specific Ser
residue at the extreme C-terminus of DP (not included in the constructs used by
Choi et al.11 for crystallization) modulates binding to
IFs28. Electrostatic interactions seemingly play a role,
therefore, and in that regard it is worth noting that vimentin, keratin and
other IFs can undergo self-induced bundling in vitro depending on ionic
and pH conditions29.
The information reported in this issue by Choi et al.11 extends these principles and, perhaps more importantly, provides a
structural framework that will be useful in furthering our understanding of
these interactions. Prior to crystallization, the authors had shown that
individual subdomains A, B and C can each bind reconstituted vimentin IFs with
specificity. The affinities observed are very low, however, and stronger
binding is seen when using fragments containing subdomains B and C, or the
entire C-terminus of DP. This is consistent with the view that a sizable number
of weak but simultaneous interactions are required for stable binding, as is so
well established for the calcium-dependent formation of stable adhesions at the
cell surface. This type of binding modality can be promoted by the duplication
of a low-affinity binding repeat and/or domain within monomeric proteins, along
with further multimerization through self-assembly. Not surprisingly, nearly
all plakin family members exhibit multiple copies of this globular subdomain
and also undergo assembly into higher order structures. The weak binding
between individual PRDs and intermediate filaments observed by the authors may
also reflect an important role for the sequences linking PRDs to one another in
DP, as has been shown to be the case for plectin24.
A groove of  10−15 Å wide and 22−26 Å
long, located near the center of a specific face of these folded globular PRDs,
is lined with conserved basic residues (Fig. 2c).
Choi et al.11 argue that this groove could accommodate up
to four turns of the -helical rod domain of vimentin, known to contain
binding determinants for DP26 and parts of which were recently
crystallized (ref. 30 and refs therein). This
positively charged groove is surrounded by acidic and neutral pockets (Fig. 2c). While these topological features are conserved
at a broad level between PRDs B and C in desmoplakin, there are significant
local differences that likely translate into different binding potentials11. Overall, these structural features support a role for electrostatic
interactions, as discussed above. This said, the demonstration that binding of
vimentin to the C-terminus of DP is relatively resistant to salt indicates
that, as expected, these interactions are likely chemically complex.
Future directions
The mechanistic inferences made by Choi et al.11
for DP−IF binding can now be directly tested through site-directed
mutagenesis within the plakin repeat domain. There is a need as well to
determine the structure of PRDs in other plakin family members, of larger
fragments of protein containing multiple PRDs, and even of PRDs interacting
with their binding sites on IF proteins. Such information may also help resolve
the issue of whether there are IF-binding determinants outside of PRDs, as
suggested from studies of plectin and other plakin family members23,
31, and will provide further insights into what makes a good
IF-binding domain. Also much anticipated is the acquisition of comparable
structural information for the N-terminus of DP, which contains crucial
information for its targeting to desmosomes6,
7. The contribution
of Choi et al.11 is hopefully the first of a string of
advances in our understanding of the function of these proteins at a structural
level.
REFERENCES
- Green, K.J. & Gaudry, C.A. Nature Rev. Mol. Cell Biol. 1, 208−216 (2000). | Article | PubMed | ISI | ChemPort |
- Vasioukhin, V., Bowers, E., Bauer, C., Degenstein, L. & Fuchs, E. Nature Cell Biol. 3, 1076−1085 (2001). | Article | PubMed | ISI | ChemPort |
- Runswick, S.K., O'Hare, M.J., Jones, L., Streuli, C.H. & Garrod, D.R. Nature Cell Biol. 3, 823−830 (2001). | Article | PubMed | ISI | ChemPort |
- Mueller, H. & Franke, W.W. J. Mol. Biol. 163, 647−671 (1983). | Article | PubMed | ISI | ChemPort |
- Green, K.J. et al. J. Biol. Chem. 265, 11406−11407 (1990). | PubMed | ISI | ChemPort |
- Bornslaeger, E.A., Corcoran, C.M., Stappenbeck, T.S. & Green, K.J. J. Cell Biol. 134, 985−1001 (1996). | Article | PubMed | ISI | ChemPort |
- Smith, E.A. & Fuchs, E. J. Cell Biol. 141, 1229−1241 (1998). | Article | PubMed | ISI | ChemPort |
- Gallicano, G.I. et al. J. Cell Biol. 143, 2009−2022 (1998). | Article | PubMed | ISI | ChemPort |
- Stappenbeck, T.S. et al. J. Cell Biol. 123, 691−705 (1993). | Article | PubMed | ISI | ChemPort |
- Kouklis, P.D., Hutton, E., Fuchs, E. J. Cell Biol. 127, 1049−1060 (1994). | Article | PubMed | ISI | ChemPort |
- Choi, H.J., Park-Snyder, S., Pascoe, L., Green, K.J. & Weis, W.I. Nature Struct. Biol. 9, 612−620 (2002). | Article | PubMed | ISI | ChemPort |
- Wiche, G. J. Cell Sci. 111, 2477−2486 (1998). | PubMed | ISI | ChemPort |
- Yang, Y. et al. Cell 86, 655−665 (1996). | Article | PubMed | ISI | ChemPort |
- Green, K.J., Virata, M.L., Elgart, G.W., Stanley, J.R. & Parry, D.A. Int. J. Biol. Macromol. 14, 145−153 (1992). | PubMed | ISI | ChemPort |
- Fuchs, E. & Yang, Y. Cell 98, 547−550 (1999). | Article | PubMed | ISI | ChemPort |
- Ruhrberg, C., Nasser Hajibagheri, M.A., Parry, D.A. & Watt, F.M. J. Cell Biol. 139, 1835−1849 (1997). | Article | PubMed | ISI | ChemPort |
- Gregory, S.L. & Brown, N.H. J. Cell Biol. 143, 1271−1282 (1998). | Article | PubMed | ISI | ChemPort |
- Leung, C.L., Green, K.J. & Liem, R.K. Trends Cell Biol. 12, 37−45 (2002). | Article | PubMed | ISI | ChemPort |
- McMillan, J.R. & Shimizu, H. J. Dermatol. 28, 291−298 (2001). | PubMed | ISI | ChemPort |
- Gallicano, G.I., Bauer, C. & Fuchs, E. Development 128, 929−941 (2001). | PubMed | ISI | ChemPort |
- Armstrong, D.K. et al. Hum. Mol. Genet. 8, 143−148 (1999). | Article | PubMed | ISI | ChemPort |
- Norgett, E.E. et al. Hum. Mol. Genet. 9, 2761−2766 (2000). | Article | PubMed | ISI | ChemPort |
- Nikolic, B., MacNulty, E., Mir, B. & Wiche, G. J. Cell Biol. 134, 1455−1467 (1996). | Article | PubMed | ISI | ChemPort |
- Steinbock, F.A. et al. J. Cell Sci. 113, 483−491 (2000). | PubMed | ISI | ChemPort |
- Coulombe, P.A., Bousquet, O., Ma, L., Yamada, S. & Wirtz, D. Trends Cell Biol. 10, 420−428 (2000). | Article | PubMed | ISI | ChemPort |
- Meng, J.J., Bornslaeger, E.A., Green, K.J., Steinert, P.M. & Ip, W. J. Biol. Chem. 272, 21495−21503 (1997). | Article | PubMed | ISI | ChemPort |
- Mack, J.W., Steven, A.C. & Steinert, P.M. J. Mol. Biol. 232, 50−66 (1993). | Article | PubMed | ISI | ChemPort |
- Stappenbeck, T.S., Lamb, J.A., Corcoran, C.M. & Green, K.J. J. Biol. Chem. 269, 29351−29354 (1994). | PubMed | ISI | ChemPort |
- Yamada, S., Wirtz, D. & Coulombe, P.A. Mol. Biol. Cell 13, 382−391 (2002). | Article | PubMed | ISI | ChemPort |
- Strelkov, S.V. et al. EMBO J. 21, 1255−1266 (2002). | Article | PubMed | ChemPort |
- DiColandrea, T., Karashima, T., Maatta, A. & Watt, F.M. J. Cell Biol. 151, 573−586 (2000). | Article | PubMed | ISI | ChemPort |
|
