Proc. Nati. Acad. Sci. USA
Vol. 86, pp. 1885-1889, March 1989
Cell Biology
Phosphorylation and disassembly of intermediate filaments
in mitotic cells
(vimentin/desmin/kinase)
YING-HAO CHOU, ELLEN ROSEVEAR, AND ROBERT D. GOLDMAN
Department of Cell Biology and Anatomy, Northwestern University Medical School, 303 East Chicago Avenue, Chicago, IL 60611
Communicated by Hans Ris, December 19, 1988
ABSTRACT
As baby hamster kidney (BHK-21) cells enter
mitosis, networks of intermediate filaments (IFs) are transformed
into cytoplasmic aggregates of protofilaments. Coincident with
this morphological change, the phosphate content of vimentin
increases from 0.3 mol of Pi per mol of protein in interphase to 1.9
mol of Pi per mol of protein in mitosis. A similar increase in
phosphate content is observed with desmin, from 0.5 mol of P1 per
mol of protein to 1.5 mol of Pi per mol of protein. Fractionation
of mitotic cell lysates by hydroxylapatite column chromatography
reveals the presence of two IF protein kinase activities, designated
as IF protein kinase I and IF protein kinase H. Comparison of
two-dimensional 32P-labeled phosphopeptide maps of vimentin
and desmin phosphorylated in vivo in mitosis, and in vitro using
partially purified kinase fractions, reveals extensive similarity in
the two sets of phosphorylation sites. Phosphorylation of in vitro
polymerized IFs by IF protein kinase II induces complete disassembly as determined by negative-stain electron microscopy. The
results support the idea that the disassembly of IFs in mitosis is
regulated by the phosphorylation of its subunit proteins.
culture medium {[2P]phosphate at 5 uCi/ml (1 Ci = 37 GBq)
in phosphate-free DMEM buffered with 25 mM Hepes (pH
7.4), supplemented with 20% dialyzed fetal calf serum and
200 jLM sodium phosphate} for 30 hr to allow the internal and
external [32P]phosphate pool to reach equilibrium. Nocodazole (0.4 ,ug/ml) was included in the last 4 hr of incubation to
obtain sufficient numbers of mitotically arrested cells.
Rounded up mitotic cells were obtained by mechanical
shake-off. Interphase cells used were obtained from the cells
left over after the mitotic cells were shaken off. Both mitotic
and interphase cells were rinsed immediately with ice-cold
phosphate-buffered saline and lysed in a buffer of 50 mM
Tris HCI, pH 7.4/5 mM EGTA/5 mM EDTA/10 mM sodium
pyrophosphate/0.2% NaDodSO4. Immunoabsorption of vimentin and desmin was carried out by using protein ASepharose beads as described elsewhere (17). The rabbit
anti-vimentin/desmin antiserum has been characterized elsewhere (18). 32P-labeled vimentin and desmin were separated
on NaDodSO4/polyacrylamide gels (19). Concentrations of
individual proteins on these gels were determined by measuring the amount of Coomassie blue bound using bovine
serum albumin as a standard (20). The final concentration of
phosphate in the medium was quantitated according to Chen
et al. (21) and the specific activity of the [32P]phosphate in the
medium at the end of the incubation period was used to
calculate mol of phosphate per mol of immunoabsorbed
vimentin and desmin.
Partial Purification of IF Kinases. Logarithmic-phase cultures of BHK cells grown in 150-cm2 flasks were mitotically
arrested in medium containing nocodazole (0.4 ,ug/ml) for 8
hr. The arrested rounded up cells were removed mechanically by shaking the flasks rigorously. This resulted in
preparations containing >95% mitotic cells as determined by
direct observation of condensed chromosomes and by immunofluorescence observation, demonstrating that they contained the bright vimentin/desmin spots, which typify mitotic
BHK cells (see Fig. 1B). The cells were pelleted by centrifugation at 500 x g and lysed in a buffer containing 20 mM
Pipes (pH 7), 5 mM EDTA, 2 mM dithiothreitol, 0.5%
Nonidet P-40, and a mixture of protease inhibitors (1 mM
phenylmethylsulfonyl fluoride, and 20 ,ug/ml each of leupeptin, aprotinin, and pepstatin). The mitotic cell lysate was
centrifuged at 200,000 x g for 30 min. The resulting pellet was
suspended by homogenization in the hydroxylapatite column
buffer consisting of 10 mM potassium phosphate, pH 6.9/400
mM KCl/2 mM dithiothreitol/1 mM EDTA/5% (vol/voi)
glycerol. This extract was centrifuged at 200,000 x g for 30
min and the supernatant was fractionated on a hydroxylapatite column with the column buffer containing a linear
potassium phosphate gradient from 10 to 300 mM. Fractions
were assayed for kinase activity as described below. The
flow-through fractions that have the IF protein kinase I
activity (see Fig. 2) were pooled, dialyzed against a solution
The intermediate filament (IF) network of certain types of
cultured cells undergoes reorganization during mitosis (1-5) and
differentiation (6). In other cell types, an increase in phosphorylation of vimentin and desmin has been observed when cells
enter mitosis (7-9) or during myogenesis (10-12). The possibility that phosphorylation might regulate the in vivo polymerization state of IFs has been supported by recent in vitro studies
using purified cyclic AMP-dependent kinase (A kinase), protein
kinase C, and protein phosphatase 1B (13-16). However, a
detailed correlation of the state of IF phosphorylation with the
level of phosphorylation within the same cell has not been
carried out. Furthermore, the identification of the kinase(s) and
phosphatase(s) involved in the organizational changes of cytoplasmic IFs in vivo remains unknown.
As a first step to determine whether or not there is a causal
relationship between increased phosphorylation and the
disassembly of IFs in vivo, we have carried out both morphological and biochemical studies using a baby hamster
kidney (BHK-21) cell line and have searched for endogenous
protein kinase(s) that cause the hyperphosphorylation of IF
proteins during mitosis. To this end, we have found two IF
protein kinase activities in mitotic cell lysates. In vitro
preparations of BHK IFs can be induced to depolymerize
during phosphorylation by one of these endogenous kinases.
EXPERIMENTAL METHODS
Quantitation of Phosphate Associated with Vimentin and
Desmin in Vivo. Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal
calf serum, penicillin (50 units/ml), and streptomycin (50
,g/ml). Radiolabeled cells were obtained by incubating in
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: IF, intermediate filament; A kinase, cAMP-dependent protein kinase.
1885
1886
Proc. Natl. Acad. Sci. USA 86 (1989)
Cell Biology: Chou et A
FIG. 1. Indirect immunofluorescence micrographs of BHK-21 cells stained with rabbit anti-BHK vimentin/desmin antiserum. Cells were
fixed and stained as described (18). (A) Interphase cell. (B) Metaphase cell. (Bar = 10 tum.)
of 2 mM Hepes, pH 7.4/2 mM dithiothreitol, and used in the
phosphorylation assay without further purification.
The IF protein kinase 11-containing fractions were pooled and
dialyzed against 5 mM potassium phosphate (pH 6.9). The
resulting solution was made 0.1% with Nonidet P-40 and further
fractionated on a DEAE-cellulose column with a linear salt
gradient of 20-300 mM KCl in 5 mM potassium phosphate, pH
6.9/1 mM EDTA/1 mM dithiothreitol. The fractions with IF
A
0
protein kinase II activity were concentrated and dialyzed
against a buffer of 2 mM Hepes, pH 7.4/2 mM dithiothreitol.
Protein Phosphorylation and Two-Dimensional 32P-Labeled
Peptide Mapping. IF subunits were prepared from ureatreated IF preparations by dialysis against 2 mM Hepes (pH
7.4) (22). The soluble subunits were polymerized in 40 ;LI of
15 mM Hepes, pH 7.4/1.5 mM MgSO4/1.5 mM EGTA/45
mM NaCl for 60 min at 250C. Phosphorylation was initiated
52 55 58 61
5 8 11 14 17 20 23 26 29 32 34 36 38 40 42 44464850
4 6
70 73 76
B
T 1.0
I
0
e.8
0
I
.6
kinase I
<.4.
/
0
Fraction No.
of
FIG. 2. Hydroxylapatite column chromatographic separation of IF protein kinases. (A) Autoradiograph of NaDodSO4/PAGE analysis
shows
0
in
B.
Lane
to
those
numbers
fraction
are
lane
each
of
the
corresponding
at
Numbers
fractions.
column
top
kinase activities from different
Coomassie blue-stained vimentin (V) and desmin (D). (B) Protein elution profile of the column. Arrow indicates the start of the potassium
phosphate gradient.
Cell Biology: Chou
et
al.
by the addition of kinase fraction (14 jul) and [32P]ATP (6 Al,
1 mM). The reactions were stopped with 2x NaDodSO4/
polyacrylamide gel sample buffer. Vimentin and desmin
bands separated on NaDodSO4/polyacrylamide gel were cut
out and extracted overnight with 300 Al of 90%o Protosol, and
the radioactivity was counted in 10 ml of Aquasol-2 (New
England Nuclear). For 32P-labeled peptide mapping, cells
were labeled in 32P-containing medium for 4 hr in the
presence of nocodazole (0.4 Ag/ml). 32P-labeled mitotic cells
were shaken off, lysed, and immunoabsorbed with rabbit
anti-vimentin/desmin antiserum as described above. 32plabeled vimentin and desmin were separated on a NaDodS04/polyacrylamide gel. Trypsinization of proteins and twodimensional peptide mapping were performed as described
elsewhere (12). 32P-labeled peptides were identified by autoradiography. Similar experiments were carried out using the
catalytic subunit of bovine heart cAMP-dependent kinase
(P2645; Sigma).
RESULTS
During interphase, BHK cells contain extensive networks of
IFs, which course from the juxtanuclear region to the cell
surface as determined by indirect immunofluorescence (Fig.
lA). During the transition from late prophase to metaphase, this
filamentous pattern disappears and only cytoplasmic spots are
seen (Fig. 1B). This spotty staining can be observed in both
normal and nocodazole-arrested mitotic cells. Ultrastructural
Proc. Natl. Acad. Sci. USA 86 (1989)
1887
observations of cells in the same mitotic stages reveal no
obvious IFs (unpublished data). Coincident with this reduction
in polymerized IFs, the amount of phosphate associated with
the IF structural proteins, vimentin and desmin, increases
significantly. The phosphate content of vimentin changes from
the interphase level of 0.3 to 1.9 mol of Pi per mol of protein in
mitosis. For desmin, the increase is from 0.5 to 1.5 mol of Pi per
mol of protein. Similar increases in IF protein phosphorylation
have been reported in other systems during mitosis (7-9). To
identify the endogenous kinases responsible for the phosphorylation ofIFs in mitosis and to determine whether such kinases
are responsible for the disassembly process seen in vivo, we
have fractionated lysates of mitotic cells by hydroxylapatite
column chromatography (Fig. 2). As a result of this procedure,
we have identified an IF protein kinase activity that phosphorylates preferentially vimentin (referred to as IF protein kinase
I) in the flow-through fractions and a second IF protein kinase
activity that phosphorylates both vimentin and desmin (referred
to as IF protein kinase II) in the retarded fractions (Fig. 2).
The authenticity ofthese two kinases has been determined by
comparing two-dimensional 32P-labeled phosphopeptide maps
of vimentin and desmin phosphorylated in vivo in mitotic cells
and in vitro using the partially purified kinases. IF protein kinase
I and II phosphorylate vimentin in vitro at different sites. In each
case, the phosphorylated peptides are subsets of those labeled
in vivo (Fig. 3 A-E), except for peptide 5, which is not found in
in vitro labeled samples (Fig. 3 A and D). Furthermore, IF
FIG. 3. Autoradiograph of 32P-labeled phosphopeptide maps of vimentin and desmin. Proteins were labeled and processed for
two-dimensional peptide maps. Different combinations of in vivo and in vitro labeled samples were run to confirm the identity of individual
peptides. (A-G) Vimentin samples. (H-L) Desmin samples. Direction of electrophoresis (+, -) and ascending chromatography (arrow) are
shown in A.
1888
Cell Biology: Chou et al.
Proc. Natl. Acad. Sci. USA 86 (1989)
c
40
0.
E
-6
IN
0.
.5
E
min
FIG. 4. Time course of phosphorylation of BHK IFs by IF protein kinase II and electron microscopic images of morphological changes during
phosphorylation. In a parallel experiment, [32P]ATP was replaced by unlabeled ATP and electron microscopic samples were taken at different
time points and negatively stained with 1% uranyl acetate. (Bar = 0.15 gum.)
protein kinase II phosphorylates desmin at sites similar to those
phosphorylated in vivo with two minor variations; peptide 1 is
not phosphorylated in vitro and peptide 6 is not seen in vivo (Fig.
3 H-J). The similarity of sites phosphorylated in vitro and in
vivo suggests to us that these two endogenous kinases are
authentic IF protein kinases.
Since A kinase and protein kinase C are known to phosphorylate vimentin and desmin (13-16), it is of interest to
determine whether the two endogenous kinases are the same
as these two known kinases. Comparison of 32P-labeled
phosphopeptide maps of vimentin phosphorylated by either
A kinase or IF protein kinase II shows that these two kinases
are not identical. A kinase phosphorylates vimentin at two
unique sites (Fig. 3 F and G, asterisks), in addition to a subset
of in vivo sites. Desmin is also phosphorylated by both A
kinase and IF protein kinase II. A kinase does not phosphorylate peptide 2, which is phosphorylated by IF protein
kinase II, and the preference for phosphorylation sites
appears different between the two kinases (Fig. 3 I, K, and L).
IF protein kinase I and A kinase phosphorylate vimentin at
different sites (Fig. 3 B and F). Based on these data, we
conclude that the two endogenous kinases are not A kinases.
In addition, IF protein kinases I and II do not appear to be
protein kinase C due to the fact that their activities are not
enhanced by the substitution of EGTA in the assay buffer
with either Ca2+/calmodulin (0.2 mM, 1 ,uM) or Ca2W/phospholipid (0.2 mM, 100 ,ug of phosphatidylserine per ml and 4
,ug of diolein per ml). Since Ca2+ sensitivity could be lost due
to proteolysis during the fractionation steps, these observations are not conclusive at the present time.
BHK IFs contain vimentin and desmin, both of which are
phosphorylated by IF protein kinase II. Therefore we have
studied the effect of kinase II on polymerized IFs (Fig. 4).
Using the conditions specified in Experimental Methods,
soluble IF subunits are induced to polymerize into IFs for 60
min (Fig. 4, 0 min). Addition of the kinase II and [32P]ATP
results in the incorporation of 4.0 and 5.0 mol of phosphate
into each mol of vimentin and desmin, respectively (Fig. 4).
As the phosphate content increases, IFs undergo dramatic
structural changes. Within 20 min, typical 10-nm IFs appear
significantly shorter. After 40 min, no IFs are seen and are
apparently replaced by short filaments 3-5 nm in diameter
(Fig. 4, 40 min). Similar mixtures, lacking either ATP or IF
protein kinase II, have no effect on the morphology of IFs.
DISCUSSION
We have shown by fluorescence microscopy that there is a
dramatic reorganization of IFs from a filamentous pattern in
interphase to an apparent nonfilamentous (spotty fluorescence) state in normal or drug-arrested mitotic BHK cells..
Coincident with this morphological change during mitosis,
there is an increase in the phosphorylation of vimentin and
desmin. The observed increase of phosphate associated with
vimentin and desmin during mitosis does not appear to be the
result of a secondary effect of nocodazole treatment since in
interphase cells that have been exposed to the drugs vimentin
and desmin do not become hyperphosphorylated.
We have also shown that in vitro polymerized BHK IFs can
be disassembled after phosphorylation by a partially purified
endogenous kinase activity (IF protein kinase II) found in
mitotic cell lysates. Furthermore, it has been known that the
disassembly of type V IFs, the nuclear lamins, is accompanied by hyperphosphorylation in mitosis (17, 23). Therefore,
our in vitro and in vivo data, in conjunction with the in vitro
results from other laboratories using A kinase and protein
kinase C (13-16), suggest that phosphorylation is important
in regulating the organizational changes of both nuclear and
cytoplasmic IF systems.
At mitosis, not only is the amount of phosphate associated
with vimentin and desmin increased, but the sites phosphorylated in vivo at mitosis are also very different from those
phosphorylated during interphase (ref. 8; unpublished obser-
Cell
Biology: Chou et A
vations). Furthermore, vimentin is phosphorylated in vivo
exclusively at the N-terminal non-a-helical domain and
desmin is phosphorylated at both the N- and C-terminal
non-a-helical domains (24). During mitosis, the increase in
phosphate appears to be restricted to the N-terminal nona-helical domain (24). Although our preliminary observations
indicate that the IF protein kinase activity in mitotic cell
lysates is much higher than that in interphase cell lysates, to
date we have not specifically characterized the kinase activity in interphase cells. We do not know whether the same two
mitotic cell kinases are present, but in some fashion inhibited,
in interphase cells, or whether phosphorylation of IF proteins
in interphase cells is mediated by a different set of protein
kinases.
There is evidence that cAMP might be involved in the
reorganization of IF networks during myogenesis (11, 12) and
that A kinase disassembles IF in vitro (13-16). However, the
two endogenous kinases identified in this study do not appear
to be A kinases. At the present time, we have not yet
determined conclusively whether or not either of the BHK
kinases is a protein kinase C. Since the head and tail domains
of vimentin and desmin are very rich in serine and thteonine
residues (25-28), it is possible that several different kinases
may be involved in phosphorylating IFs at different sites,
each specific for different stages of the cell cycle and for
different developmentally regulated events. In BHK cells,
one of the two endogenous IF protein kinases (IF protein
kinase II) is able to induce BHK IF disassembly in vitro. We
have also found that the phosphorylation of purified bovine
lens vimentin and chicken gizzard desmin by BHK IF protein
kinase II induces their disassembly (data not shown). The
effects of the other endogenous kinase (IF protein kinase I)
on IF structure are not known, and it is not yet clear what role
this kinase plays in regulating IF organization during mitosis.
Our finding that the vimentin/desmin-containing IF network is altered in BHK cells during mitosis is not unique.
Similar alterations in keratin-containing IFs have been reported in several types of cells (1-5). However, the morphological fate of IF networks during mitosis varies in different
cell types (3, 5, 9) and even in different culture conditions
(29). Since vimentin and desmin can be phosphorylated at
multiple sites, the level of increased phosphorylation and the
different spectra of phosphorylation sites might modulate the
conformational changes of IF to different extents. Dramatic
morphological changes, like the IF network alteration into
vimentin/desmin-rich spots, can be easily observed by light
microscopy. However, the detection of subtle changes in IF
structure, which may result from lesser degrees of phosphorylation at different sites, will require more sensitive experimental approaches. The further purification and characterization of endogenous kinases (or other additional factors)
Proc. Natl. Acad. Sci. USA 86 (1989)
1889
responsible for the modulation of IF structure should in the
future permit the elucidation of the differential effects of
phosphorylation on the organization of IF networks.
We would like to thank Manette McReynolds and Laura Davis for
their help in preparing this manuscript. This research was supported
by a National Cancer Institute grant to R.D.G.
1. Franke, W. W., Schmid, E. & Grund, C. (1982) Cell 30, 103113.
2. Aubin, J. E., Osborn, M., Franke, W. W. & Weber, K. (1980)
Exp. Cell Res. 129, 149-165.
3. Jones, J. C. R., Goldman, A. E., Yang, H.-Y. & Goldman,
R. D. (1985) J. Cell Biol. 100, 93-102.
4. Horwitz, B., Kupfer, H., Eshar, Z. & Geigef, B. (1981) Exp.
Cell Res. 134, 281-290.
5. Lane, E. B., Goodman, S. L. & Trejdosiewicz, L. K. (1982)
EMBO J. 1, 1365-1372.
6. Gard, D. L. & Lazarides, E. (1980) Cell 19, 263-275.
7. Evans, R. M. & Fink, L. M. (1982) Cell 29, 43-52.
8. Evans, R. M. (1984) J. Biol. Chem. 259, 5372-5375.
9. Celis, J. E., Larsen, P. M., Fey, S. J. & Celis, A. (1983) J. Cell
Biol. 97, 1429-1434.
10. Gard, D. L. & Lazarides, E. (1982) Proc. Natl. Acad. Sci. USA
79, 6912-6916.
11. Gard, D. L. & Lazarides, E. (1982) Mol. Cell. Biol. 2, 11041114.
12. O'Connor, C. M., Gard, D. L. & Lazarides, E. (1981) Cell 23,
135-143.
13. Inagaki, M., Nishi, Y., Nishizawa, K., Matsuyama, M. & Sato,
C. (1987) Nature (London) 328, 649-652.
14. Inagaki, M., Gonda, Y., Matsuyama, M., Nishizawa, K.,
Nishi, Y. & Sato, C. (1988) J. Biol. Chem. 263, 5970-5978.
15. Geisler, N. & Weber, K. (1988) EMBO J. 7, 15-20.
16. Evans, R. M. (1988) FEBS Lett. 234, 73-78.
17. Ottaviano, Y. & Gerace, L. (1985) J. Biol. Chem. 260, 624-632.
18. Starger, J., Brown, W. E., Goldman, A. & Goldman, R. (1978)
J. Cell Biol. 78, 93-109.
19. Laemmli, U. K. (1970) Nature (London) 227, 680-685.
20. Fenner, C., Trant, R. R., Mason, D. T. & Wikman-Coffelt, J.
(1975) Anal. Biochem. 63, 595-602.
21. Chen, P. S., Toribara, T. Y. & Warner, H. (1956) Anal. Chem.
28, 1756-1758.
22. Zackroff, R. V. & Goldman, R. D. (1979) Proc. Natl. Acad.
Sci. USA 76, 6226-6230.
23. Gerace, L. & Blobel, G. (1980) Cell 19, 277-287.
24. Evans, R. M. (1988) Eur. J. Cell Biol. 46, 152-160.
25. Geisler, N. & Weber, K. (1982) EMBO J. 1, 1649-1656.
26. Quax, W., Egberts, W. V., Hendriks, W., Quax-Jenken, H. &
Bloemendal, H. (1983) Cell 35, 215-223.
27. Quax, W., van den Broek, L., Egbert, W. V., Ramaekers, F.
& Bloemendal, H. (1985) Cell 43, 327-338.
28. Zehner, Z. E., Li, Y., Roe, B. A., Paterson, B. M. & Sax,
C. M. (1987) J. Biol. Chem. 262, 8112-8120.
29. Tolle, H.-G., Weber, K. & Osborn, M. (1987) Eur. J. Cell Biol.
43, 35-47.