![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 14, Issue 6, 2262-2276, June 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Oncology, University of Alberta/Cross Cancer Institute, Edmonton
Alberta Canada T6G 1Z2;
Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia,
Pennsylvania 19111; and
Department of Cell Biology, University of Alberta, Edmonton Alberta Canada T6G
1Z2
Submitted July 3, 2002;
Revised February 8, 2002;
Accepted March 4, 2003
Monitoring Editor: Tim Stearns
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
;
Turley and Torrance, 1985;
Turley et al., 1987,
1991), is a multifaceted
protein with both intracellular and extracellular functions
(Hardwick et al.,
1992; Assmann et al.,
1999; Pilarski et
al., 2001). RHAMM mediates the HA-specific motility of a
variety of cell types, including transformed cell lines and ex vivo malignant
leukocytes (Masellis-Smith et
al., 1996; Pilarski
et al., 2001). RHAMM is the founding member of the
B(X)7B hyaladherin protein family and interacts with HA through
basic (B) 9–11 amino acid (aa) B(X)7B motifs
(Yang et al., 1994).
Encoded by 18 exons, RHAMM localizes to chromosome 5q33.2-qter
(Spicer et al., 1995)
and encodes an 85-kDa protein with extensive coiled coil structure and a basic
amino-terminal globular domain (Hardwick
et al., 1992; Assmann
et al., 1999). In addition to its role in cell migration,
RHAMM expression, and overexpression, have been linked to ras transformation,
tumor progression, and metastasis (Hall
et al., 1995). Moreover, RHAMM has been identified as a
microtubule-associated protein that also interacts with the actin cytoskeleton
(Assmann et al.,
1999). Overexpression of RHAMM in B lymphoid malignancies, and
other cancers, suggests that it may play a role in human oncogenesis and
disease progression (Wang et al.,
1998; Crainie et al.,
1999). Two RHAMM splice variants, RHAMM-exon 4 and
RHAMM-exon13, have been identified in myeloma patients and cancer
cell lines (Assmann et al.,
1999; Crainie et al.,
1999); all variants localize to the mitotic spindle, although loss
of exon 4 inhibits interaction with interphase microtubules
(Assmann et al.,
1999). The localization of RHAMM isoforms at the mitotic spindle
provides a putative mechanism through which RHAMM expression regulates mitotic
integrity, ploidy, and possibly carcinogenesis.
In animal cells, the duplication of centrosomes, and their subsequent polar
separation, initiate the establishment of a bipolar microtubule array. Many
centrosomal and noncentrosomal proteins, including microtubule motor
complexes, coordinate early mitotic events. Kinesin-like protein 2 family
members (Xklp2, Hklp2, and KRP180) are essential for centrosomal separation
and the maintenance of spindle bipolarity
(Boleti et al., 1996).
The carboxy-terminal leucine zipper of Xklp2 is required for its centrosomal
and spindle pole localization, via an interaction with targeting protein for
Xklp2 (TPX2) and the dynein/dynactin motor complex
(Wittmann et al.,
1998). TPX2 has recently been shown to mediate
Ran-GTP–dependent microtubule assembly and to target Aurora A kinase to
the spindle pole (Gruss et al.,
2002; Kufer et al.,
2002). A recurring mechanism in spindle pole formation and
stability is dynein/dynactin-mediated recruitment of structural proteins to
the spindle pole (reviewed in Zimmerman
and Doxsey, 2000). Pericentrin, a 220-kDa coiled coil protein,
plays an integral role in centrosomal stability by acting as a molecular
scaffold and interacting with numerous proteins and protein complexes,
including dynein (Purohit et al.,
1999; Zimmerman and Doxsey,
2000). Although centrosomal localization of pericentrin is
microtubule independent (Gillingham and
Munro, 2000), based on the observations that nocodazole disruption
fails to abolish pericentrin centrosomal localization, recruitment of
pericentrin to centrosomes requires an association with the dynein motor
complex (Young et al.,
2000; Zimmerman and Doxsey,
2000). Nuclear mitotic apparatus protein (NuMA), a highly coiled
coil 240-kDa protein, is also concentrated at spindle poles, where it
cross-links microtubule minus ends, through an association with the
dynein/dynactin complex (Gaglio et
al., 1996). Recently, a direct microtubule interaction has
been localized to the carboxy terminal tail domain of NuMA
(Haren and Merdes, 2002).
Thus, NuMA may function in microtubule sliding and spindle stability by
directly binding with microtubules at the carboxy terminus, complexing with
the dynein/dynactin motor complex and anchoring one microtubule relative to
another sliding microtubule (Gaglio et
al., 1996).
The transforming acidic coiled coil (TACC) proteins, a family of proteins
that concentrate to centrosomes through a conserved coiled coil
carboxy-terminal domain called the TACC domain, also play a role in organizing
centrosomal microtubules (Gergely et al.,
2000a,b;
Lee et al., 2001).
The TACC domain of Drosophila-TACC, D-TACC, has been shown to
interact with minispindles, msps, the Drosophila homolog of human
colonic and hepatic tumor overexpressed (TOGp) protein
(Cullen and Ohkura, 2001;
Lee et al., 2001).
The TOG family of proteins (TOGp, XMAP215, and Msps) can bind directly to
microtubules and promote their polymerization. In Drosophila, Msps is
transported to microtubule minus ends by the kinesin-like protein Ncd and
anchored to centrosomes by an association with D-TACC
(Cullen and Ohkura, 2001;
Lee et al., 2001);
centrosomal targeting of D-TACC is regulated by Aurora A kinase, which, in
turn, is dependent on TPX2-mediated targeting
(Giet et al., 2002;
Kufer et al., 2002).
Like RHAMM, TACC proteins have been intimately linked to carcinogenesis. TACC1
overexpression transforms mouse fibroblasts
(Still et al.,
1999a). A TACC2 isoform has been identified as the tumor
suppressor protein Azu-1 in breast carcinoma lines and TACC3 is up-regulated
in multiple cancer lines (Still et
al., 1999b; Chen et
al., 2000). Moreover, TACC3 is essential for hematopoietic
stem cell function and may play a primary role in the regulation of p53
function within hematopoietic cells
(Piekorz et al.,
2002).
This article investigates the involvement of RHAMM at the mitotic spindle
with emphasis on its localization in nonadherent cell lines derived from human
lymphocytes. We demonstrate that RHAMM shares structural and functional
similarity with proteins that are essential for the maintenance of the mitotic
spindle. RHAMM localizes to the centrosome during interphase and to the
spindle poles during mitosis. Consistent with sequence similarity to the Klp2
family, we demonstrate that RHAMM interacts with the dynein complex in vivo.
The centrosomal targeting domain of RHAMM localizes to the previously
characterized hyaluronan binding domain
(Yang et al., 1994);
we show this motif is phylogenetically related to the TACC and Klp2
centrosomal targeting motifs. We find that overexpression of GFP-RHAMM leads
to mitotic delays and apoptosis. We also find that disruption of RHAMM mitotic
function, through microinjection of purified anti-RHAMM antibodies, affects
spindle integrity and results in the formation of tripolar and tetrapolar
spindles. We have shown that in addition to its other functions, RHAMM is a
centrosomal protein that interacts with the dynein microtubule motor complex,
functions in the maintenance of spindle integrity, and is chromosomally
located proximal to the putative FGFR4-TACC4 gene cluster on
chromosome 5qter.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Primer Sequences. The following primer sequences were used:
Primer sequences are shown 5' to 3', lowercase letters denote mispriming from de novo mRNA, bold sequences denote restriction endonuclease recognition sites, and italic letters denote start and stop codons.
Antibodies. 
-Tubulin (clone B-5-1-2),
-tubulin (clone GTU-88), and dynein (clone 70.1) were purchased from
Sigma-Aldrich (St. Louis, MO); pericentrin was from Babco (Richmond, CA); and
the mouse monoclonal NuMA antibodies were identified in a monoclonal antibody
(mAb) screen for mitotic chromosome scaffold proteins
(Compton et al.,
1991). The polyclonal RHAMM antibody CM1 was produced by
Washington Biotechnology (Baltimore, MD) to the following carboxy-terminal
peptide sequence: G681IKHFDPSKAFHHESK696. The polyclonal
serum was affinity purified over a peptide-loaded NHS-activated Sepharose 4B
column (Amersham Biosciences, Piscataway, NJ), washed, eluted with glycine pH
2.5, neutralized, and quantified by OD280. Fractions were pooled
and concentrated with Ultrafree-MC (Millipore, Bedford, MA) to >10 mg/ml.
The specificity of this serum was tested by immunoblot, immunoprecipitation,
and immunofluorescent analysis (Figure
5). A second anti-RHAMM serum was as described previously
(Assmann et al.,
1999); this antiserum is a pan-specific anti-RHAMM sera raised
against a bacterially expressed GST-RHAMM fusion protein (RHAMM exons
10–13, aa307–498)
(Assmann et al.,
1999). The anti-TOGp serum was as described previously
(Dionne et al.,
2000). Secondary antibodies were from Molecular Probes (Eugene,
OR).
|
Cell Culture, Transient Transfection, Nocodazole Treatment, and
Immunofluorescence
Cells from RPMI 8226, a human lymphoblastic cell line derived from the
peripheral blood of a multiple myeloma patient, and Raji, a human
lymphoblastic cell line derived from a Burkitts lymphoma patient, were grown
in suspension in RPMI 1640 medium supplemented with 10% fetal bovine serum
(Invitrogen), at 37°C in 5% CO2. HeLa cells, a human adherent
epithelial cell line derived from a patient with cervical adenocarcinoma, were
grown in DMEM medium supplemented with 10% fetal bovine serum (Invitrogen), at
37°C in 5% CO2. Cells were passaged 24 h before transfection.
Suspension cells were transfected by electroporation (270 mV, 960 µF,
47–53 ms), and stable GFP--tubulin transfectants were selected
for in 600 µg/ml G418. HeLa cells were transfected with LipofectAMINE 2000
(Invitrogen) following manufacturers' protocols. For nocodazole experiments,
HeLa cells were incubated for 90 min in 25 µM nocodazole (Sigma-Aldrich)
before immunofluorescence. Suspension cells and HeLa were fixed and
permeabilized in cold MeOH at defined time points posttransfection. Cells were
washed with phosphate-buffered saline (PBS)-0.5% Triton X-100 (Sigma-Aldrich)
before immunofluorescence. Primary and secondary antibodies were diluted in
PBS-0.1% Tween (Sigma-Aldrich) + skim milk powder (blocking buffer), and all
antibody incubations were for 30 min at room temperature. For double-staining
experiments, antibodies were added sequentially. Cells were washed three times
in PBS-0.5% Tween before and after incubations. Cells were mounted in 90%
glycerol/PBS + 4,6-diamidino-2-phenylindole (DAPI) and images were acquired
using a confocal LSM 510 or multiphoton microscope (Carl Zeiss, Thornwood,
NY). Images were processed using MetaMorph software (Universal Imaging,
Downingtown, PA) and Photoshop 5.02 software (Adobe Systems, Mountain View,
CA).
Immunoprecipitations
HeLa cells were transfected using LipofectAMINE 2000 (Invitrogen) and the
manufacturer's protocol. After transfection, cells were incubated for 12 to 16
h, in the presence of OptiMEM (Invitrogen), whereas untransfected cells were
incubated in fresh DMEM-10% fetal bovine serum. After incubation, cells were
released from plates with 1x trypsin, washed three times with PBS, and
lysed at 5 x 106–107 cells/ml in 1%
3-[(3-cholamidopropy-l)dimethylammonio]propanesulfonate (CHAPS) plus 10
µg/ml leupeptin, 10 µg/ml antipain, and 1 mM phenylmethylsulfonyl
fluoride (all from Sigma-Aldrich). For some experiments, cells were washed and
lysed in situ with lysis buffer. All immunoprecipitation procedures were
performed at 4°C. Lysates were precleared with protein A-Sepharose beads
(Amersham Biosciences) with rotation for 30 min. Precleared lysates were
incubated with antibodies (3 µl/150 ml of lysate) for 2 h at 4°C with
rotation, and then with protein A beads (40 µl of a 1:1 slurry in CHAPS+)
for 1 h at 4°C with rotation. Beads were collected with centrifugation and
washed four times with CHAPS+ buffer. Immunoprecipitated proteins were eluted
with boiling SDS buffer and analyzed by a 5% stacking/8% separating SDS-PAGE.
Precleared lysates (25 µl) and postimmunoprecipitation fractions were
analyzed to determine efficiency and relative quantity of the
immunoprecipitations.
Microinjection
HeLa cells were plated onto coverslips and synchronized at the
G1/S boundary by double thymidine block (2.5 mM thymidine). Cells
were injected 1 h after release from thymidine block. Microinjection was
performed with a semiautomatic microinjector (model 5412; Brinkman Eppendorf,
Westbury, NY). For each coverslip, 200–300 cells were injected with each
antibody mix over the course of 
30 min. Affinity-purified antibodies were
kept in calcium- and magnesium-free PBS and were filtered through a 0.22-µm
microfiltration cup (Millipore) before microinjection. Cells were released
from the G1/S block by washing with PBS and replaced with fresh
media before microinjection. Cells were fixed with 3.5% paraformaldehyde
12 h after release from the G1/S boundary. Cells were
permeabilized with 0.2% Triton X-100 in KB (10 mM Tris, pH 7.5, 0.15 M NaCl,
0.1% bovine serum albumin) for 5 min and washed with KB before antibody
incubation. To minimize loss of loosely attached mitotic cells, the coverslips
were centrifuged at 200 x g for 2 min in a clinical centrifuge
(GPKR; Beckman Coulter, Fullerton, CA). The injected antibodies were detected
by Cy5-conjugated anti-rabbit secondary antibodies (Jackson Immunoresearch
Laboratories, West Grove, PA).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
). Because RHAMM overexpression and function also
characterizes human lymphocytes and B lymphoid malignancies
(Masellis-Smith et al.,
1996; Gares et al.,
1998; Crainie et al.,
1999), we investigated RHAMM localization in nonadherent,
suspension cell lines derived from lymphocyte malignancies (RPMI 8226 and
Raji). The subcellular localization of endogenous RHAMM was investigated by
indirect immunofluorescence by using two polyclonal antisera. The first,
termed IHABP, was previously characterized as a pan-specific anti-RHAMM
antiserum (Assmann et al.,
1999), whereas the second, termed CM1, is characterized below
(Figure 5A). The adherent line
HeLa was used to confirm RHAMM localization in cells derived from solid tumors
(our unpublished data). Examination of RHAMM localization within interphase, nonadherent cell lines revealed pronounced centrosomal localization in addition to microtubule association (Figure 1A). Although less pronounced in adherent HeLa cells, centrosomal localization was identified within interphase and prophase cells (our unpublished data). We speculate that the more polymerized, filamentous microtubule arrays of adherent lines has prevented previous observation of RHAMM centrosomal localization. During prophase, in both adherent and suspension lines, RHAMM concentrated at the center of microtubule asters and associated with microtubules growing between the two asters (Figure 1B). During prometaphase and metaphase, RHAMM intensified at the spindle pole and extended along spindle microtubules during metaphase. Additionally, RHAMM localized to the spindle midzone during anaphase and telophase. During telophase, RHAMM redistributed from the centrosomes and concentrated to the spindle midzone (Figure 1B). To compare the localization of RHAMM in fixed and live cells and to control for possible fixation and permeabilization artifacts, the distribution of full-length RHAMM (RHAMMFL)-GFP was examined in transiently transfected RPMI 8226, Raji, and HeLa cells. Consistent with the localization of endogenous RHAMM in fixed samples, RHAMMFL-GFP proteins localized to centrosomes, interphase microtubules, the mitotic spindle pole, and midzone microtubules in suspension and in adherent cells. Although microtubules are suboptimally visualized in nonadherent cells, RHAMM antibodies colocalized with microtubules in all interphase cells examined, including methanol fixed suspension cells (our unpublished data).
|
Previous analysis of RHAMM established the existence of three major
structural domains: the amino-terminal head (aa 1–69; pI = 10.72), which
interacts with microtubules; an extensive coil coiled stalk (aa 70–680;
pI = 5.06); and a short carboxy-terminal tail (aa 681–724; pI = 8.12)
(Assmann et al.,
1999). Sequence alignments and structural prediction reveal a
relationship between RHAMM and the Klp2 protein family
(Figure 1C). Structure
prediction for RHAMM, like Xklp2, predicts extensive coiled coil structure
within the carboxy-terminal 600 amino acids
(Figure 1C)
(Boleti et al., 1996;
Assmann et al., 1999).
Moreover, BLAST 2 sequence analysis
(Tatusova and Madden, 1999) of
RHAMM primary structure against the Klp2 family reveals significant homology
(e-28 for Hklp2), including conservation of the carboxy-terminal
leucine zipper, which exhibits 72% identity
(Figure 1C, i); this domain is
vital to Klp2 centrosomal localization and function through an indirect
interaction with the dynein/dynactin complex
(Wittmann et al.,
1998). The coiled coil terminus of RHAMM, however, bears little
sequence identity to the centrosomal targeting TACC domain of TACC1. The
carboxy-terminal leucine zipper of RHAMM (boxed in
Figure 1C, i) overlaps the
defined HA-binding domains (underlined in
Figure 1C, i); this region is
conserved among RHAMM proteins in human, mouse and rat
(Lynn et al., 2001).
RHAMM lacks the highly conserved amino-terminal kinesin motor domain but
contains a microtubule-binding domain in its place (aa 1–70;
Figure 1C). Based on these
similarities, we tested whether the conserved carboxy-terminal leucine zipper
directed RHAMM to centrosomes and spindle poles.
The Amino Terminus of RHAMM Is Required for Interaction with
Interphase Microtubules, whereas the Carboxy-terminal Leucine Zipper Targets
RHAMM to the Centrosome
To define the domains in RHAMM responsible for centrosomal targeting, we
constructed GFP fusion proteins that correspond to known RHAMM splice variants
(RHAMM-exon4 and RHAMM-exon13) as well as RHAMM
carboxy-terminal deletion mutants (RHAMM1–525,
RHAMM1–623, and RHAMM1–679) and
carboxy-terminal fragments (RHAMM625–725 and
RHAMM500–725) (Figure
2A).
|
To investigate the centrosomal localization of RHAMM in nonadherent cells,
GFP-RHAMM fusion constructs were transiently transfected into RPMI 8226 and
Raji lines. Consistent with previous observations in HeLa cells
(Assmann et al.,
1999), RHAMM variants that contain exon 4 (i.e.,
RHAMMFL and RHAMM-exon13) colocalized with interphase
microtubules (Figure 2B; our
unpublished data). The degree of colocalization is illustrated by the
intensity profile along the inset line within
Figure 2B, c. Although the low
cytoplasm to nucleus ratio of the suspension cells makes microtubule
colocalization experiments difficult, occasional cells with well-defined
microtubule networks were observed. Within these cells, GFP-RHAMM variants,
containing exon 4, demonstrated precise colocalization with microtubules (our
unpublished data). Loss of exon 4 (GFP-RHAMM-exon4) disrupts
interphase microtubule interactions as demonstrated by the lack of
colocalization along the inset line 1 within
Figure 2B, i. In fact,
GFP-RHAMM-exon4 demonstrated diffuse, relatively invariant,
cytoplasmic localization (Figure
2B, line 1). However, GFP-RHAMM-exon4 localized to
centrosomes (see line 2 profile) and mitotic spindle poles
(Figure 2B). Interestingly,
whereas GFP-RHAMMFL localized to the spindle pole and along
chromosome-contacting microtubules, GFP-RHAMM-exon4 localized
primarily to the spindle pole. Nocodazole inhibition of microtubule
polymerization was used to determine whether polymerized microtubules were
required for the association of RHAMM with centrosomes and the mitotic spindle
pole. Pericentrin was used as a positive control because it is known to
localize at the centrosome in a microtubule-independent manner
(Gillingham and Munro, 2000).
Centrosomal localization was analyzed by -tubulin staining, and DAPI
staining was used to determine the mitotic status of the cells. In the
presence of polymerized microtubules (-nocodazole), RHAMM localizes at
centrosomes and the mitotic spindle (Figure
2C). As demonstrated by the RHAMM staining within metaphase cells
(Figure 2C, e), nocodazole
treatment affected cellular architecture and disrupted metaphase mitotic
spindles. Consistent with previous reports, pericentrin localization at
centrosomes was microtubule independent (our unpublished data); in the absence
of polymerized microtubules, RHAMM maintained an association with interphase
centrosomes and the mitotic spindle pole
(Figure 2C). It is important to
note that microtubule-independent centrosomal localization is distinct from
microtubule-independent centrosomal targeting. Although RHAMM and other
centrosomal proteins such as pericentrin do not require microtubules to remain
localized at the centrosome, it is likely that RHAMM centrosomal targeting is
facilitated through an interaction, either direct or indirect, with a
minus-end directed microtubule molecular motor (see below).
The carboxy-terminal coiled coil domain of RHAMM shares 72% identity with
the basic leucine zipper (BZIP) of the Klp2 family
(Figure 1C). To test the
importance of the conserved leucine zipper in centrosomal targeting, deletion
constructs of RHAMM, lacking the carboxy-terminal 200, 102, and 47 aa, were
constructed (Figure 2A).
Transient transfection of empty vector EGFPC1 was used as a negative control
for centrosomal and spindle pole localization; both the centrosome and spindle
pole showed slight amplification of transfected EGFP-C1 (our unpublished
data). Deletion of the carboxy-terminal 200 aa (RHAMM525; our
unpublished data) or 102 aa (RHAMM623;
Figure 2D), inhibited the
centrosomal localization of the fusion proteins. However, RHAMM proteins that
contained the leucine zipper, RHAMMFL
(Figure 2D),
RHAMM-exon4 (Figure
2B), RHAMM-exon13 (our unpublished data), and
RHAMM679 (Figure
2D), localized to the centrosome. The BZIP motif was also
essential for localization of RHAMM constructs to the mitotic spindle pole;
after examination of five transfection experiments, we were unable to identify
a mitotic figure within GFP-RHAMM679-transfected cells. This
observation was tested and quantitated in HeLa cells. Transient transfectants
of GFP-RHAMM679 and GFP-RHAMMFL were examined for
mitotic stage at 12, 13, 14 and 20 h posttransfection and compared with
neighboring untransfected cells. GFP-RHAMM679 transfection resulted
in only prophase transfectants (n = 47; our unpublished data), whereas
GFP-RHAMMFL transfection led to accumulation of prometaphase and
metaphase cells (58.6% [34/58] of transfected cells vs. 26.2% [101/386] within
the untransfected population; Figure
4). Moreover, the majority of GFP-RHAMM679
transfectants demonstrated large aggregation of GFP fluorescence (our
unpublished data). Microtubule associations and centrosomal localization were
maintained in transfectants with low levels of fluorescence (our unpublished
data). Interestingly, 704TPLK707, a consensus cdc2
phosphorylation site conserved in mouse and rat, falls within the deleted
region of RHAMM679. To test the sufficiency of the BZIP motif for
centrosomal targeting, we constructed GFP-tagged, carboxy-terminal RHAMM
fragments consisting of the terminal 100 and 225 aa, respectively. When
transiently transfected into RPMI 8226, these constructs localized to the
interphase nucleus and mitotic spindle pole. Although the carboxy terminus
does not include a predicted (predictNLS analysis) simple or bipartite nuclear
localization signal, characterized by a short stretch of basic aa or two
interdependent positively charged clusters separated by a short linker region,
it is highly basic (Cokol et al.,
2000). Given their basic nature, the nuclear localization of the
small carboxy-terminal fragments may be the result of transport, and not
simple diffusion, into the nucleus. Both fragments also showed slight
localization to the centrosome although not greater than GFP alone. Therefore,
the carboxy-terminal basic leucine zipper of RHAMM, which overlaps the defined
B(X)7B HA binding domains and is homologous to the Klp2 family, is
essential for centrosomal targeting and sufficient for spindle pole
localization.
|
RHAMM Interacts with the Dynein Motor Complex In Vivo
Because the carboxy-terminal leucine zipper is conserved between RHAMM and
the Klp2 family and this domain mediates the indirect interaction of Xklp2 and
dynein (Wittmann et al.,
1998), we investigated whether RHAMM interacts with the dynein
motor complex in cells. We first determined the spatial relationship of
endogenous RHAMM and dynein IC in HeLa cells. Double immunofluorescence by
confocal microscopy demonstrated colocalization of dynein and RHAMM at the
spindle pole within mitotic HeLa cells
(Figure 3A). Because a large
fraction of intracellular RHAMM, and dynein, does not localize to the spindle
pole, it is likely that only a subset of the total RHAMM protein interacts
with the dynein/dynactin motor complex, and vice versa.
|
To extend the confocal colocalization data, we investigated the ability of
RHAMM antibodies to coimmunoprecipitate dynein IC from interphase and mitotic
Xenopus extracts. A cell cycle-dependent
NuMA–dynein–dynactin complex has been identified in
immunoprecipitates by using either anti-dynein or anti-NuMA antibody within
the Xenopus system (Merdes et
al., 1996; Merdes et
al., 2000). However, the NuMA–dynein–dynactin
complex seems to be less stable in higher vertebrate cells with some, but not
others, able to coimmunoprecipitate these protein complexes
(Zeng, 2000). This raised the
possibility that it may also be difficult to detect a RHAMM–dynein
complex in HeLa cells. Thus, we analyzed RHAMM–dynein interactions in
Xenopus extracts. Immunoprecipitation of mitotic and interphase
Xenopus extracts with RHAMM antibodies revealed a major species at 95
kDa as well as minor bands around 150, 113, and 85 kDa (our unpublished data).
These sizes are consistent with those published for RHAMM species in various
mammalian tissues (Hall et al.,
1995; Assmann et al.,
1999). RHAMM antibodies coimmunoprecipitated dynein, with slight
amplification within M phase extracts, but some dynein was found to interact
with RHAMM during interphase (our unpublished data). Interestingly, the
monoclonal dynein IC antibody was unable to coimmunoprecipitate RHAMM protein,
consistent with our observation that only a fraction of dynein was associated
with RHAMM, and vice versa.
To confirm this association in mammalian cells, HeLa extracts were
transiently transfected with GFP-RHAMMFL. CHAPS-soluble lysates
from transfected, or untransfected, populations were separately
immunoprecipitated with an anti-intermediate chain dynein mAb, IgG2a control
antibody, a polyclonal RHAMM antibody, a polyclonal TOGp antibody, a
monoclonal NuMA antibody, and a nonimmune serum. As demonstrated in
Figure 3C, lanes 1 and 2,
transfected and untransfected lysates contained comparable amounts of
endogenous proteins with the transfected populations overexpressing a
GFP-RHAMM species of 120 kDa (lane 2, RHAMM immunoblot). Recently, dynein
has been coimmunoprecipitated from HeLa extracts with NuMA antiserum; for this
reason, we included a NuMA immunoprecipitation as a positive control for
association with dynein (Bhattacharya
et al., 2002). Also, because the drosophila homolog of
TOGp, msps, uses the Ncd motor, and not the dynein complex, to localize to
centrosomes (Cullen and Ohkura,
2001), TOGp immunoprecipitation of dynein served as a negative
control. RHAMM antibodies coimmunoprecipitated dynein from transfected and
untransfected lysates. Comparison of the amount of coprecipitated dynein with
the Western blot signal seen in the total lysates indicates that, under the
conditions used herein, only a fraction of dynein is recovered in a complex
with RHAMM. Although the level of coprecipitated dynein was low, the amount
was similar to that precipitated by NuMA antibodies. The specificity of the
RHAMM–dynein interaction is demonstrated by the inability of TOGp
antibodies, control IgG2a, and nonimmune serum to coimmunoprecipitate dynein.
The reciprocal coimmunoprecipitation of RHAMM with dynein IC antibodies was
achieved in transfected and untransfected HeLa lysates; again, comparison with
the western blot signal indicates that only a fraction of endogenous RHAMM is
precipitated by dynein. TOG antibodies also coimmunoprecipitated
GFP-RHAMMFL from transfected lysates; this result is consistent
with live cell observation, within RPMI 8226, of accumulation of
GFP-RHAMMFL aggregates at the cell periphery and vectoral movement
of GFP-RHAMMFL toward the cell body (our unpublished data). Thus,
the dynamic redistribution of GFP-RHAMMFL in live suspension cells
is consistent with minus-end directed motion, mediated by the dynein motor
complex, and plus-end association, likely mediated through an interaction with
TOG.
Overexpression of GFP-RHAMM Fusions Initiates a Mitotic Block in RPMI
8226 and HeLa Cells
Because GFP-RHAMMFL does not generate stable transfectants
within RPMI 8226 and HeLa cells, time-lapse confocal microscopy was used to
follow the kinetics of mitosis within these transfectants. Transiently
transfected GFP-RHAMMFL cells arrested in mitosis
(Figure 4A, i). These
transfected cells failed to divide, mitotic spindles broke down, and the cells
underwent apoptosis (Figure 4A, i). To demonstrate that the visualization techniques were not causing the
mitotic block and that the GFP tag was not influencing the kinetics of
mitosis, tubulin-GFP fusions were stably transfected into 8226 cells and
followed through mitosis. These cells demonstrated normal division
(Figure 4A, ii).
To quantitate the level of GFP-RHAMMFL overexpression and its
effect on mitosis, RPMI 8226 cells were transiently transfected with
GFP-RHAMMFL or EGFP-C1 vector and sorted 8 h later. An aliquot of
transfected cells was examined by immunoblot for overexpression at 16 h
posttransfection (Figure 5A).
The 8226 cells transfected with EGFPC1
(Figure 5A, lane 1) express an
endogenous RHAMM species of 90 kDa, whereas the 8226 cells transiently
transfected with GFP-RHAMMFL
(Figure 5A, lane 2) express
endogenous RHAMM and a GFP-RHAMM protein at 120 kDa. Band intensities,
measured by digital imaging quantitation, indicated that the GFP-RHAMM
intensity was approximately five times that of the endogenous RHAMM. The level
of overexpression in these sorted populations of transiently transfected RPMI
8226 cells is greater than that in the unsorted population of HeLa cells used
in the immunoprecipitation experiments. This difference is due to the sorting
of populations (i.e., many unsorted, lysed HeLa cells may not be expressing
GFP-RHAMM) and likely gives a more definitive representation of the level of
GFP-RHAMM, relative to endogenous RHAMM, within transfected cells. Transfected
cells were fixed at 12, 16, 20 and 36 h posttransfection and stained with
-tubulin, to identify the spindle, and with DAPI, to determine mitotic
status. Untransfected RPMI 8226 and EGFPC1 (control vector)-transfected RPMI
8226 served as controls. GFP-RHAMMFL overexpression resulted in an
increase in the frequency of prometaphase cells; 70% of mitotic
GFP-RHAMMFL cells were prometaphase arrested 20 h posttransfection
(Figure 4, B and C). Interestingly, true metaphase alignment of chromosomes was not achieved in
GFP-RHAMMFL-transfected cells. Probably due to the prometaphase
arrest, few GFP-RHAMMFL-transfected cells were observed in anaphase
or cytokinesis as demonstrated by the increasing metaphase: anaphase +
cytokinesis ratio (Figure 4C). In combination with the prometaphase arrest, GFP-RHAMMFL
overexpression resulted in a dramatic increase in apoptotic cells as well as
the presence of cells with large GFP-RHAMMFL clusters (our
unpublished data). It is unclear whether these large GFP-RHAMMFL
clusters precede apoptosis; these clusters did not colocalize with
-tubulin and were phenotypically similar to those identified by
overexpression of GFP-TACC domain fusions
(Gergely et al.,
2000a).
Inhibition of Endogenous RHAMM Results in Abnormal Mitotic
Figures
To further investigate the mitotic functions of RHAMM, a polyclonal
peptide-specific antibody was created against the carboxy-terminal peptide
sequence G682-K697, of RHAMM
(Figure 5A). This sequence is
outside the conserved basic leucine zipper motif to avoid inhibiting Hklp2
function. This serum was affinity purified and the specificity of the antibody
was tested by immunoblotting endogenous RHAMM and GFP-RHAMMFL,
immunoprecipitating endogenous RHAMM from HeLa and 8226, and colocalization
with GFP-RHAMMFL in HeLa and 8226
(Figure 5A).
To determine whether spindle integrity was affected by disruption of RHAMM
function, RHAMM peptide antibodies were microinjected into the cytoplasm of
cells arrested at the G1/S boundary. The microinjected cells were
released from the G1/S boundary and allowed to proceed to mitosis
before fixation and analysis by immunofluorescence. In four separate
experiments, microinjection of anti-RHAMM antibodies into HeLa cells led to
the formation of tripolar and tetrapolar spindles within 11 ± 3.4 and
14 ± 4.3% of injected cells, respectively
(Figure 5B). Immunoblocking
RHAMM function resulted in symmetrical and asymmetrical tripolar and
tetrapolar spindles. Asymmetrical tetrapolar spindles
(Figure 5B) were phenotypically
similar to those observed within NuMA-immunoblocked CFPAC-1 cells
(Gordon et al.,
2001); however, microinjection of RHAMM antibodies did not
significantly affect spindle focusing. Thus, RHAMM does not seem to be
essential for spindle focusing but rather for maintenance of spindle
integrity. The fact that roughly three-quarters of injected cells have
phenotypically normal spindles suggests that other proteins can compensate for
loss of RHAMM function. With all abnormal spindles, the injected RHAMM
antibodies and NuMA colocalized to the aberrant spindle poles
(Figure 5B).
Chromosomal Location and Phylogenetic Analysis of RHAMM Primary
Sequence Suggests Membership in the TACC Family
The TACC protein family members are centrosomal proteins that associate
with microtubules minus ends through a conserved coiled coil carboxy-terminal
TACC domain and are putative regulators of the mitotic apparatus. Despite the
low sequence identity within the TACC domain
(Figure 1C), RHAMM shares
structural and functional similarity to the TACC family. A defining feature of
the TACC proteins is an evolutionary conserved relationship with the
fibroblast growth factor receptor (FGFR) gene family
(Still et al.,
1999b). TACC1, 2, and 3 genes map proximal to
FGFR1, 2, and 3 genes on chromosomes 8p11, 10q26, and
4p16.3, respectively. Currently, no TACC gene has been identified proximal to
the FGFR4 locus on chromosome 5q35.1-qter. RHAMM maps to a
chromosomal region near to the FGFR4 gene locus at 5q33.2-qter. Given
the proximity of RHAMM to FGFR4, we used phylogenetic
analysis to compare the carboxy-terminal 240 aa of RHAMM to its current
protein family members, the B(X)7B hyaladherins, the TACC family,
and the Klp2 family (Figure
6).
|
Phylogenetic analysis was performed with the MEGA software program by using
carboxy-terminal sequences previously aligned by BLAST and PEPTOOL software
(Wishart et al.,
1997; Tatusova and Madden,
1999; Kumar et al.,
2000). Phylogeny inference was performed using the Unweighted Pair
Group Method with Arithmetic Mean and Neighbor-Joining (our unpublished data)
distance methods and branch lengths are shown. Theoretically, the evolutionary
distance separating two sequences can be defined as the number of mutational
events per site underlying the evolutionary history separating these sequences
(Brocchieri, 2001). We used the
Poisson correction method to approximate these distances.
BLAST analysis of the B(X)7B family members revealed no significant similarity in the B(X)7B-containing domains. CD44, which binds HA with a classical extracellular link domain, contains an intracellular, carboxy-terminal B(X)7B domain and this motif was included in the phylogenetic analysis. As demonstrated by their pairwise evolutionary distances, the B(X)7B family members maintain little evolutionary convergence in their functional domains. The carboxy terminus of RHAMM, currently defined as a B(X)7B domain, shares no significant sequence similarity, and a minimal evolutionary relationship, with other B(X)7B family members.
Comparison of the carboxy terminus of RHAMM with the TACC family reveals
that RHAMM diverged before the formation of the TACC3 and TACC1/2 ancestor.
This is consistent with the divergence of FGFR4 in the phylogenetic tree of
the FGFR family. The carboxy terminus of RHAMM shares minimal identity
(20%) with TACC1. Therefore, we included the PACT domain of human
pericentrin in the phylogenetic analysis to control for the possibility that
the RHAMM/TACC relationship is wholly due to RHAMM's coiled coil structure.
The PACT domain is an alternative conserved centrosomal targeting domain
(Gillingham and Munro, 2000).
Pericentrin occupied a separate branch that was 1.10 evolutionary distance
units from TACC1 (our unpublished data). Thus, the carboxy terminus of RHAMM
is more closely related to the TACC domain than is the PACT domain of human
pericentrin.
BLAST analysis of the carboxy-terminal 240 aa of RHAMM reveals significant homology to the Klp2 family (e-13). As there is currently only one human protein member of the Klp2 family (Hklp2), we compared the carboxy-terminal domain of human RHAMM and the sea urchin Klp2-related KRP-180 with those of Hklp2 and Xklp2 to estimate the evolutionary distance between these motifs. As shown in Figure 6, the evolutionary distance between RHAMM (0.73) and Hklp2/Xklp2 approximates that of KRP-180 (0.64). Therefore, the centrosomal targeting motif of RHAMM is more closely related to the Klp2 family than the TACC family. However, the fact that RHAMM lacks the highly conserved molecular motor domain precludes it from membership in the Klp2 family.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
); functionally related to other TACC family members; and,
like TACC 1–3, the RHAMM gene is located near the
putative FGFR4-TACC4 gene locus.
To date, no human TACC4 protein has been identified. A TACC4 protein has
been identified in rabbits (rTACC4) (accession no. AAK54244
[GenBank]
, 454 aa) and shown
to directly interact with AKAP350
(Steadman et al.,
2002). However, BLAST analysis of the corresponding cDNA
(AF372837
[GenBank]
) against the human genome localized this gene to chromosome 4p16.3
(up to e-29) with exon boundaries that mirror TACC3. This gene
product may be the result of a secondary start site in rTACC3 rather than a
bone fide rTACC4; in fact, alignment of rTACC4 protein sequence (AAK54244
[GenBank]
)
with an hTACC3 variant (Q9Y6A5) reveals higher identity (60%) than alignment
of mTACC3 (XP_132002) with hTACC3 (57%)
(Figure 7). Perhaps more
convincingly, examination of rTACC4, hTACC3, and mTACC3 cDNA demonstrates that
regions of lowest identity between rTACC4 and hTACC3 are also less conserved
in mTACC3 (Figure 7). Moreover,
the tissue expression of rTACC4 is similar to that defined for mTACC3. MTACC3,
but not mTACC2, is highly expressed in the testis, spleen, and lung and absent
in the brain, liver, kidney, and muscle
(Piekorz et al.,
2002); similarly, rTACC4 is highly expressed in the spleen and gut
and absent in the brain and liver
(Steadman et al.,
2002). Interestingly, RHAMM, like TACC3 and rTACC4, is highly
expressed in the testis, colon, and stomach but absent in the brain, liver,
heart, kidney, and lung (Line et
al., 2002).
|
Currently, RHAMM is characterized by its HA binding potential through the
carboxy-terminal B(X)7B domains (Yang et al.,
1993,
1994). RHAMM has two
additional B(X)7B domains within its NH2-terminal
globular head [K40(X)7K48 and
R67(X)7K75], which, like the carboxy-terminal
domains, contain two internal basic residues. These domains have been
identified as binding motifs for interphase and mitotic microtubules
(Assmann et al.,
1999). We have also shown that the Klp2 family contains putative
B(X)7B HA binding domains that are virtually identical to those in
RHAMM. For the Klp2 family, these domains are defined as a BZIP motif that
allows for centrosomal localization through an indirect interaction with the
dynein/dynactin motor complex. We suggest that in addition to their
participation in HA binding (Yang et
al., 1993), the carboxy-terminal B(X)7B domains of
RHAMM are vital components of a consensus BZIP domain that targets this
protein to the centrosome and allows for an interaction with the
dynein–dynactin complex.
As an acidic (pI = 5.64), extensively coil coiled centrosomal protein that
maintains mitotic spindle integrity and is chromosomally located proximal to
the FGFR4 locus, RHAMM is a good candidate to be the fourth member of
the TACC family. However, RHAMM does not contain the highly conserved, carboxy
terminal, centrosomal targeting TACC domain. In fact, RHAMM uses a
carboxy-terminal BZIP motif (L-X6-L-X6-L-X6),
preceded by lengthy coil coiled structure, to localize to centrosomes.
Interestingly, many centrosomal targeting domains may be defined by coiled
coil structure terminating in a BZIP motif. Examination of the alignments of
two defined centrosomal targeting domains, the TACC and PACT domain protein,
reveals carboxy-terminal conservation of leucine residues (our unpublished
data). Interestingly, Xenopus and human NuMA, which also interact
with the dynein motor complex, contain a stretch of conserved leucines,
L209-X9-L-X3-L-X9-L-X3-L-X6-L,
that precedes an extensive predicted coiled coil structure (our unpublished
data). Recent sequence analysis predicts that NuMA may interact with dynactin
through a predicted CH domain (aa 110–210) interaction with Arp 1
(Novatchkova and Eisenhaber,
2002); interestingly, the conserved BZIP motif outlined above lies
at the terminus of the predicted CH domain. Moreover, a direct interaction
between EB1 and the dynactin subunit p150glued has recently been
localized to the carboxy terminus of EB1, which contains a BZIP motif
(Askham et al., 2002).
Thus, evolutionary pressures may have allowed for divergence among centrosomal
targeting domains, and dictated conservation of coiled coil structure
terminating, or initiating, in a BZIP motif. RHAMM is unique to the human TACC
family in its ability to interact with the dynein motor complex. To date, no
association has been demonstrated between TACC1, 2, or 3 and the dynein motor
complex, although it has been suggested that D-TACC and Msps may be maintained
in the vicinity of the spindle poles by the activity of
microtubule–motor complexes such as NuMA–dynein–dynactin
(Lee et al., 2001).
The divergence of the carboxy-terminal coiled coil domain of RHAMM from the
TACC domain may have enabled a RHAMM–dynein interaction, allowing for
RHAMM-specific mitotic functions. Thus, RHAMM may be a new member of the TACC
family or there may be another, yet to be identified, TACC protein located at
5q35.1. Currently, we conclude that RHAMM is a TACC-like protein and may, in
fact, be TACC4.
Because RHAMM can interact with microtubules both directly and indirectly, through an interaction with dynein, RHAMM may function in the maintenance of mitotic spindles by cross-linking centrosomal microtubules. A similar function has been proposed for NuMA, although NuMA has also been shown to affect spindle focusing as well as integrity. Because RHAMM (85 kDa) is much smaller than NuMA (240 kDa), RHAMM–dynein complexes may function proximal to the centrosome compared with NuMA–dynein complexes. Thus, in RHAMM immunoblocked cells, NuMA–dynein complexes may maintain spindle focusing but, in a quarter of injected cells, not spindle integrity. In fact, NuMA is not displaced from the RHAMM-immunoblocked abnormal spindle poles. Endogenous RHAMM function may assist in spindle integrity and microtubule focusing, along with the minus-end directed KIN C motor HSET, in the absence of NuMA but be inadequate to maintain spindle focusing.
To date, it is unclear whether the interaction between RHAMM and the dynein
motor complex is direct or indirect; the structural similarity of the carboxy
terminus of RHAMM to Xklp2 would suggest that RHAMM, like Xklp2, is targeted
to microtubule minus ends through an association with TPX2. The localization
of RHAMM, as determined by immunofluorescence and GFP-RHAMMFL,
mirrors that of TPX2 throughout mitosis
(Gruss et al., 2002).
GFP-hTPX2 overexpression, like GFP-RHAMMFL, initiates a
prometaphase block and inhibition of TPX2 function, through microinjection of
anti-TPX2 antibodies, resulted in abnormal mitoses in 37.8% of injected cells
(Gruss et al., 2002).
Interestingly, TPX2 has been recently shown to be essential for the spindle
targeting of Aurora A kinase (Kufer et
al., 2002). Thus, it is possible that RHAMM, like Aurora A
kinase and Klp2, is dependent on TPX2 for spindle pole localization.
This work outlines an essential role for RHAMM in the organization and
integrity of a bipolar spindle. The role of RHAMM in mitotic stability may
partially explain its relationship to cancer. Overexpression of RHAMM, and its
variants, characterizes both hematological malignancies and solid tumors
(Crainie et al.,
1999; Wang et al.,
1998); additionally, RHAMM has been identified by SEREX
(serological identification of antigens by recombinant expression cloning)
analysis to be a tumor-associated antigen in colon cancer, acute myelogenous
leukemia, and chronic myeloid leukemia
(Greiner et al.,
2002; Line et al.,
2002). Autoantibodies to centrosomal proteins, such as
pericentrin, are a hallmark of autoimmune disease, such as scleroderma
(Doxsey et al., 1994;
Rattner et al., 1998;
Gavanescu et al.,
1999). Moreover, NuMA has been investigated as a biomarker for
colorectal cancer (Briggman et
al., 1999). It is possible that the tumor-specific anti-RHAMM
autoantibodies may be indicative of overexpression of RHAMM, or tumor-specific
expression of RHAMM variants. Such RHAMM disregulation could potentially
affect centrosomal and spindle pole dynamics with consequent dramatic, and
possibly oncogenic, effects on cell shape, motility, and ploidy.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
Footnotes |
|---|
Corresponding author. E-mail address:
lpilarsk{at}gpu.srv.ualberta.ca.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Assmann, V., Jenkinson, D., Marshall, J.F., and Hart, I.R. (1999). The intracellular hyaluronan receptor RHAMM/IHABP interacts with microtubules and actin filaments. J. Cell Sci. 112, 3943-3954.[Abstract]
Bhattacharya, N., Wang, Z., Davitt, C., McKenzie, I.F., Xing, P.X., and Magnuson, N.S. (2002). Pim-1 associates with protein complexes necessary for mitosis. Chromosoma 111, 80-95.[Medline]
Boleti, H., Karsenti, E., and Vernos, I. (1996). Xklp2, a novel Xenopus cent
