JBC Papers in Press.

Published on November 10, 2006 as Manuscript M609692200

The latest version is at http://www.jbc.org/cgi/doi/10.1074/jbc.M609692200

MICROFIBRIL-ASSOCIATED MAGP-2 STIMULATES ELASTIC FIBER ASSEMBLY

Raphael Lemaire1, Julie Bayle1, Robert P. Mecham2 and Robert Lafyatis1
From 1the Department of Medicine, The Arthritis Center, Boston University School of
Medicine Boston, MA 02118
and 2the Department of Cell Biology & Physiology Washington University School of
Medicine, St. Louis, MO 63110
Running title: MAGP-2 stimulates elastic fiber assembly.
Address correspondence to: Raphael Lemaire, Boston University School of Medicine, The
Arthritis Center Bldg. E-5, 715 Albany Street, Boston, MA 02118, USA; Tel.: (617) 638 6821;
Fax: (617) 638 5226; E-mail: rlemaire@bu.edu
Elastic fibers are complex structures composed
of a tropoelastin inner core and microfibril
outer mantle guiding tropoelastin deposition.
Microfibrillar proteins mainly include fibrillins
and
microfibrilassociated
glycoproteins
(MAGPs). MAGP-2 exhibits developmental
expression peaking at elastic fiber onset,
suggesting that MAGP-2 mediates elastic fiber
assembly. To determine whether MAGP-2
regulates elastic fiber assembly, we used an in
vitro model featuring doxycycline-regulated
cells conditionally overexpressing exogenous
MAGP-2, and constitutively expressing EGFPtagged
tropoelastin.
Analysis
by
immunofluorescent staining showed that
MAGP-2
overexpression
dramatically
increased elastic fibers levels, independently of
extracellular levels of soluble tropoelastin,
indicating that MAGP-2 stimulates elastic fiber
assembly. This was associated with increased
levels of matrix-associated MAGP-2. Electron
microscopy showed that MAGP-2 specifically
associates with microfibrils and that elastin
globules primarily colocalize with MAGP-2associated
microfibrils,
suggesting
that
microfibril-associated
MAGP-2
facilitates
elastic fiber assembly. MAGP-2 overexpression
did not change levels of matrix-associated
fibrillin-1, MAGP-1, fibulin-2, fibulin-5, or
emilin-1, suggesting that microfibrils and other
elastic fiber-associated proteins known to
regulate elastogenesis do not mediate MAGP-2
induced elastic fiber assembly. Moreover,
mutation analysis showed that MAGP-2 does
not stimulate elastic fiber assembly through its
RGD motif, suggesting that integrin receptor
binding does not mediate MAGP-2-induced
elastic fiber assembly. Since MAGP-2 interacts
with Jagged-1 that controls cell-matrix

interaction and cell motility, two key factors in

elastic fiber macroassembly, microfibrilassociated MAGP-2 may stimulate elastic fiber
macroassembly by targeting the release of
elastin globules from the cell membrane onto
developing elastic fibers.
Elastic fibers are major insoluble extracellular
matrix structures that provide connective tissues
with flexibility and extensibility. These properties
are critical to the functions of arteries, lung, skin
and all other dynamic tissues and have been
essential requirements in the evolution of multicellular organisms. The importance of elastic
fibers is highlighted by severe heritable connective
tissue diseases caused by mutations in components
of elastic fibers (1). Elastic fibers are complex
structures composed of two major components: an
inner core of amorphous cross-linked tropoelastin
and an outer microfibril mantle composed of
multiple proteins. Various other components of the
elastic fiber localize at the elastin-microfibril or
cell-elastic fiber interface (see (2) for review),
including fibulin-2, fibulin-5 and emilin-1. These
elastic fiber associated molecules are important
mediators of elastic fiber assembly as deletion of
these genes induces elastic fiber defects in mice
(3-5).
Microfibrils play an essential role in elastic
fiber assembly by orchestrating deposition of
tropoelastin onto the developing elastic fiber.
Recent data have shown that a microfibril scaffold
is required for deposition of tropoelastin into the
extracellular matrix (6). This confirms previous
ultrastructural studies showing that microfibrillar
proteins are generally expressed prior to
tropoelastin during development (7) and that
elastin in early stages of assembly is always
associated with microfibrils in developing elastic

1
Copyright 2006 by The American Society for Biochemistry and Molecular Biology, Inc.

tissues (8,9). However, due to the structural and

functional heterogeneity of microfibrils, the
molecular basis of tropoelastin alignment by
microfibrils remains unclear. In this respect, each
microfibrillar
component
may
selectively
determine the structure of microfibrils and/or their
function in tropoelastin alignment. The biological
role of these microfibrillar molecules is still under
investigation. Fibrillin-1 and -2 are the principal
structural components of microfibrils in elastic
fibers (10,11). Mutation of fibrillin-1 in the Tightskin (Tsk) mice induces major alterations in
microfibril structures (12,13). They also have a
function in tropoelastin deposition as they regulate
tropoelastin coacervation (14). They are large
glycoproteins that have distinct but overlapping
pattern of expression (15). Fibrillin-2 is generally
expressed earlier in development than fibrillin-1
and may be particularly important in elastic fiber
formation (16).
Apart from the fibrillins, microfibrilassociated glycoprotein 1 (MAGP-1) is possibly
the best candidate for an integral microfibril
molecule important for structural integrity. It is
associated with virtually all microfibrils and
widely expressed in mesenchymal and connective
tissue cells throughout the development. MAGP-1
binds to amino-terminal exons 1-10 of fibrillin-1
(17) through its carboxy-terminal region
containing seven common cysteine residues (18).
MAGP-1 forms ternary complexes with fibrillin-1
and decorin (19) as well as with biglycan and
tropoelastin (20). MAGP-1 also binds to the 3
chain of type VI collagen (21).
MAGP-2, the other member of this small
microfibrillar protein family, is a 170-173 residue
protein structurally related to MAGP-1 mainly in
its central region. MAGP-2 contains an RGD
motif through which it specifically binds to the
v3 integrin (22), and interacts with Jagged1, an
activating ligand for Notch receptor signaling (23).
MAGP-2 is specifically associated with fibrillincontaining microfibrils. It binds to a carboxyterminal 7 tandem cbEGF-like repeat-containing
region of fibrillin-1 through its carboxy-terminal
region (18), as well as to amino-terminal and
central regions of fibrillin-1 (24). The role of
MAGP-2 in microfibril structure remains unclear
since it exhibits restricted patterns of tissue
localization and developmental expression.

However, it may have a specialized function in

matrix homeostasis. We have recently shown that
fibrosis in skin of scleroderma patients and Tsk
mice is associated with increased MAGP-2 levels
(13) and that MAGP-2 stimulates expression of
type I procollagen (25). Other evidence suggests
that MAGP-2 may also modulate microfibril
function in elastogenesis, as MAGP-2 is
associated with elastin associated microfibrils in
developing nuchal ligaments (26), MAGP-2
mRNA expression peaks at the period of onset of
elastogenesis in this tissue (27), and increased
matrix-associated MAGP-2 in the hypodermis of
Tsk mice is associated with increased elastic fiber
assembly. Consistent with MAGP-2 function in
elastogenesis, several fibrillin-1 mutations giving
rise to Marfan syndrome have been mapped to the
7 cbEGF repeats-containing region that binds
MAGP-2 (28). In the present study, using cell
lines conditionally overexpressing MAGP-2 and
constitutively
expressing
EGFP-tagged
tropoelastin, we show that microfibril-associated
MAGP-2 stimulates elastic fiber assembly
independently of its RGD motif and without
affecting microfibril assembly or matrixassociated levels of elastic fiber associated
molecules, fibulin-2, fibulin-5 and emilin-1.
EXPERIMENTAL PROCEDURES
Cells and plasmids - All cells used in this study
were developed from a mouse embryonic
fibroblast line (MEF 3T3, Clontech, Palo Alto,
CA) that harbors the pTet-Off regulator, plasmid
expressing a tet-controlled transactivator (tTA)
protein that binds to promoter tet-responsive
element (CMVmin + TRE). MEF-TE cells
constitutively overexpressing tropoelastin were
created by transfecting MEF cells with pEGFPbTE-FL, a vector supporting CMV-driven
expression of bovine tropoelastin tagged with
EGFP (B. Kozel). MG-TE cells conditionally
overexpressing mouse MAGP-2 and constitutively
tropoelastin were developed by transfecting MEFTE cells with pTRE-MAGP-2, vector supporting
tet-regulated TRE-driven expression of MAGP-2
(25). MGV5-TE and MGRGD-TE cells
conditionally overexpressing V5-tagged MAGP-2
and RGD MAGP-2, respectively, were
developed by transfecting MEF-TE cells with
pTRE2-MAGP2-V5 and pTRE2-MAGP2RGD

(25).
MEF-cFbn
cells
conditionally
overexpressing C-terminal truncated fibrillin-1
mutant were developed by transfecting MEF cells
with pTRE-cFbn-EGFP, vector supporting tetregulated TRE-driven expression of the truncated
fibrillin-1 mutant. pTRE-cFbn-EGFP was
constructed, first, by creating pTRE-Fbn-1, vector
supporting conditional expression of wild-type
fibrillin-1. pTRE-Fbn-1 was constructed by
deleting the 3.0-kb Sph-1 fragment from pTRE2Tsk fibrillin (13), corresponding to the duplicated
region within Tsk fibrillin-1, followed by
religation. Second, a 792-bp EGFP cDNAcontaining fragment from pEGFP-N1 (Clontech)
was inserted in frame into pTRE-Fbn-1 in place of
a 300-bp Sac I / Xba I fragment, corresponding the
91 C-terminal amino acid of fibrillin-1.
Cell transfection - Cell lines were transfected
using lipofectamine plus reagent (Invitrogen,
Carlsbad, CA). After transfection, cells were
selected for 12-14 days using G418 (100 g/ml)
and Hygromycin (60 g/ml). 40 to 50 selected cell
colonies were then ring-cloned, expanded, and
analyzed for the transfected gene by northern blot
and western blot.
SDS-polyacrylamide
gel
electrophoresis
(PAGE) and immunoblotting - Secreted proteins
were precipitated from culture media by
trichloracetic acid. Proteins from cell and matrix
layers were directly lysed in SDS-PAGE buffer /
2% betamercaptoethanol. Proteins were resolved
on an 8% SDS-PAGE gel, transferred to
nitrocellulose, and then incubated with primary
polyclonal rabbit antisera directed against mouse
MAGP-2 (1/1000 dilution) (13), mouse fibrillin-1
(Ab9543, gift from L. Sakai), mouse fibulin-5
(1/2000, gift from B. Schiemann), bovine
fibronectin (1/2000, Calbiochem, San Diego, CA),
EGFP (1/1000, Abcam, Cambridge, MA), or
monoclonal mouse antisera to bovine alphatropoelastin (1/400, clone BA4, RDI, Flanders,
NJ). The secondary antibody consisted of a
secondary horseradish peroxidase-conjugated
donkey anti-mouse IgG antibody (1/10000). Signal
was detected using Supersignal West Pico or
Femto Chemiluminescent Reagent (Pierce,
Rockford, IL) followed by autoradiography.
RNA analysis - Total RNA was prepared using
RNeasy Total RNA kit (Qiagen, Valencia, CA),
separated by electrophoresis through a 1M

formaldehyde/1% agarose gel and blotted onto a

nylon
membrane
(Hybond-H,
Amersham
Biosciences, Piscataway, NJ). Blotted RNAs were
then hybridized to random primed 32P-labeled
cDNA probes for either mouse MAGP-2,
tropoelastin or 18S ribosome, as previously
described (29).
Immunofluorescence - Cells were grown in
Lab-Tek 8-chamber culture slides (Nalge Nunc
International, Naperville, IL), fixed for 10 min in
paraformaldehyde at room temperature, blocked
for 30 min with 3% BSA in TBS, and then
incubated for 2 h with polyclonal rabbit antisera
directed against either mouse MAGP-2 (gift from
JM. Shipley), mouse fibrillin-1 (Ab9543), mouse
MAGP-1 (R. Mecham), mouse fibulin-2 (gift from
R. Timpl), mouse emilin-1 (gift form A.
Colombatti),
mouse
tropoelastin
(EPC,
Owensville, MO) (30), or monoclonal mouse
antisera to bovine alpha-tropoelastin (clone BA4)
or V5 epitope (Invitrogen, Carlsbad California) at
a 1/200 dilution in 3% BSA/TBS. Cells were
washed 3 times with TBS for 5 min, incubated for
1 h with a rhodamine-conjugated donkey antirabbit IgG or FITC-conjugated donkey anti-mouse
IgG
antibody
(Jackson
ImmunoResearch
laboratories, West Grove, PA) at a 1/200 dilution.
Specific fluorescence signals were then visualized
by an Olympus BH-2 microscope under
fluorescent light associated with high resolution
Olympus DP70 camera. Exposure times were set
manually and were identical between dox + and
dox - conditions.
Immuno-electron microscopy of elastic fibers
MGV5-TE cells overexpressing MAGP-2 were
plated 1:3 in 35mm tissue culture dishes and
cultured without dox in the presence of TGF-.
After 2 weeks, cells were fixed in 2%
paraformaldehyde for 30 min, washed three times
5 min with PBS, and aldehydes quenched with
0.1M ammonium chloride for 10 min. Cells were
then blocked for 30 min with 1% BSA/PBS,
incubated with primary antibodies directed to
MAGP-2-V5 and/or elastin-EGFP in 1%
BSA/PBS for 90 min, washed three times 5 min
with PBS, and incubated with secondary goldlabeled anti IgG antibodies in 1% BSA/PBS for 60
min. Immunolocalization of MAGP-2-V5 was
performed using a mouse anti-V5 antibody in
combination with a 6 nm-gold labeled anti-mouse
IgG antibody (1:30 dilution - Jackson

immunofluorescence. Cells were cultured in the

presence of TGF- in order to maximize the
assembly of fibrillin-1-containing microfibrils in
the matrix (29). At 1-week post-confluence, MEF
cells constitutively overexpressing tropoelastin
(MEF-TE cells) showed a sparse and restricted
network of thin developing elastic fibers that
stretched between a discontinuous network of
dense elastin globules (Fig. 1B). Importantly,
MEF-TE cells cultured for longer time, up to a
month, eventually showed a much more robust
elastic fiber network as elastic fibers built up in
diameter and length (data not shown).
To investigate the role of MAGP-2 in elastic
fiber assembly, we developed MEF-TE cells
conditionally overexpressing exogenous mouse
MAGP-2 (MG-TE cells) under the control of
tetracycline (tet) via the tet-off system (31). METE cells, originally designed to constitutively
express a tet-controlled transactivator, were stably
transfected with a plasmid supporting mouse
MAGP-2 expression under the control of the tetresponsive element. In the absence of doxycycline
(dox), a tet analog, MG-TE cells showed high
levels of MAGP-2 in the culture medium (Fig. 1C,
dox -). In the presence of dox, MG-TE cells
showed only low levels of endogenous MAGP-2
in the medium as expression of exogenous
MAGP-2 is repressed (Fig. 1C, dox +).
Overexpression of MAGP-2 did not change
mRNA expression or medium levels of
tropoelastin (Fig. 1D and 1E, dox + vs. dox -). Of
note, TGF- greatly enhanced levels of exogenous
MAGP-2 present in the medium (Fig. 1C, dox-)
This is associated with similar increase in levels of
exogenous MAGP-2 mRNA (data not shown).
Since expression of exogenous MAGP-2 mRNA is
driven by a CMV minimum promoter (see
material section), this suggests that TGF-
stabilizes the MAGP-2 mRNA.
To look at the effect of MAGP-2
overexpression on elastic fiber assembly, hyperconfluent MG-TE cells were cultured for one
week with or without dox in the presence of TGF, and then elastic fibers analyzed by EGFP
fluorescence. Consistent with parental MEF cells
(Fig. 1B), MG-TE cells cultured with dox, and
therefore expressing only low levels of
endogenous MAGP-2, showed an extremely poor
and limited network of elastic fibers associated

ImmunoResearch
laboratories).
Coimmunolocalization of MAGP-2-V5 and elastinEGFP were performed using two antibody
combinations. The first combination used mouse
antibody to V5 (1:100) and rabbit antibody to GFP
(1:300 - Ab290, Abcam, Cambridge, MA)
associated with 6 nm-gold labeled anti-mouse and
18 nm-gold labeled anti-rabbit IgG antibodies. The
second combination used rabbit antibody to
MAGP-2 (1:200) and mouse antibody to bovine
elastin (1:200 - clone BA4) associated with 18 nmgold labeled anti-rabbit and 6 nm-gold labeled
anti-mouse IgG antibodies. Following antibody
incubation, cells were washed three times 5 min
with PBS, prior to final fixation in 1%
glutaraldehyde in 0.1M sodium cacodylate buffer
for 15 min on ice. Cells were then stained with
1.25% osmium tetraoxide-0.1M Na cacodylate for
30 min, washed three times 5 min with 0.1m Na
cacodylate buffer, incubated with 2% tannic acid
in 0.1M cacodylate for 30 min, washed three times
with cacodylate buffer, rinsed twice 5 min with
15% ethanol, followed by 30-min incubation in
4% aqueous uranyl acetate before sequential
dehydration with ethanol. The samples were then
embedded in PolyBed812-filled gelatin capsules
and baked at 60C for 18 h. 60-nm thin sections
were counterstained with uranyl acetate and lead
citrate and examined on a Zeiss 902 electron
microscope.
RESULTS
MAGP-2 overexpression stimulates elastic
fiber assembly in MEF cells As a tool to
investigate the molecular mechanisms underlying
elastic fiber assembly, we used cultured MEF
cells, a 3T3 derived-mouse embryonic fibroblast
cell line that assembles fibrillin microfibrils but
not elastic fibers since tropoelastin is not
expressed (Fig. 1A). Because these cells produce
all components required to form the scaffolding of
elastic fibers, but not tropoelastin itself, they
provide an excellent in vitro system for studying
elastic fiber assembly in association with
constitutive expression of an exogenous
tropoelastin molecule. To generate MEF cells
constitutively expressing tropoelastin (MEF-TE
cells), MEF cells were transfected with a vector
supporting CMV-driven expression of bovine
tropoelastin tagged with EGFP. Their ability to
deposit exogenous tropoelastin onto microfibrils
was then tested by EGFP fluorescence and

with numerous and prominent elastin globules

(Fig. 2A, dox +). Thin, sparse, developing elastic
fibers stretched between these large elastin
globules that were arranged in a discontinuous,
loose network (Fig. 2B, d and j). In contrast,
MEG-TE cells cultured with dox, and therefore
overexpressing exogenous MAGP-2, showed a
strong and extended network of elastic fibers (Fig.
2A, dox -). Elastic fibers were thick and dense and
seemed to be generated from the aggregation of
the elastin globules (Fig. 2B, c and i). Of note,
both control transfection with the empty pTRE2
and control dox-treatment of parental MEF-TE
cells failed to alter elastic fibers assembly,
indicating that the effect of MAGP-2 on elastic
fibers was not due to either the plasmid pTRE-2 or
dox (data not shown). These results suggest that
MAGP-2 overexpression stimulates assembly of
tropoelastin onto microfibrils, independent of
extracellular levels of soluble tropoelastin
available to the matrix (Fig. 1E).
MAGP-2-induced elastic fibers is associated
with increased matrix-associated levels of MAGP2 - To investigate the molecular basis of the
stimulation of elastic fiber assembly by MAGP-2,
MG-TE cells were also stained in parallel for
MAGP-2 by immunofluorescence. MG-TE
overexpressing exogenous MAGP-2 showed a
dramatic increase in matrix-associated MAGP-2
compared to the same cells expressing only
endogenous MAGP-2 (Fig. 2B, a vs. b and g vs.
h). Remarkably, increases in elastic fibers and
matrix-associated MAGP-2 upon expression of
exogenous MAGP-2 were closely associated as
they largely colocalized (Fig. 2B, e and. k).
Similarly, in the absence of exogenous MAGP-2,
the sparse network of elastic fibers also
colocalized with basal endogenous matrixassociated MAGP-2 (Fig. 2B, f and l).
MAGP-2-induced matrix-associated elastin
primarily colocalizes with MAGP-2-associated
microfibrils and forms elastic fibers
Ultrastructural analysis of elastic fibers assembled
by MGV5-TE cells overexpressing V5-tagged
MAGP-2 and EFGP-tagged tropoelastin was
performed by electron microscopy (EM)
associated with immunolocalization of MAGP-2
alone (Fig 3A and Fig. 3B) or in combination with
elastin (Fig 3C and Fig. 3D). Gold-labeled anti-V5
and MAGP-2 antibodies targeting MAGP-2-V5
specifically localized to 10-12 nm fibrillin

microfibrils within the matrix (Fig. 3A, 3B and

3C, 6 nm-gold labeling Fig 3D, 18 nm-gold
labeling). This is consistent with previous
immunoelectron microscopy studies performed on
various animal tissues showing specific
localization of MAGP-2 to fibrillin microfibrils
(26). Dark, electron dense, globular structures
typical of elastin were found exclusively
associated with bundles of fibrillin microfibrils
(Fig. 3A, 3B, 3C, and 3D), providing molecular
evidence for the presence of elastic fibers in our in
vitro system. Although a substantial amount of
MAGP-2-associated microfibrils were elastin-free,
elastin globules primarily localized to MAGP-2associated microfibrils (Fig. 3A, 3B, 3C, and 3D),
not to MAGP-2-free microfibrils (Fig. 3A, 3B and
3C, arrowhead). These globules were gold-labeled
by antibody to GFP targeting elastin-EGFP (Fig.
3C, 18 nm-gold particles) and antibody to bovine
tropoelastin (Fig. 3D, 6 nm-gold particles),
confirming their elastin nature. The lack of
uniform labeling of these globules by the antibody
to GFP is likely to represent a technical artifact
inherent to the complexity of immuno-EM
technique and / or lack of accessibility of EGFP
epitopes within highly-cross-linked elastin
globules. Of note, control experiments using
secondary 6 nm- or 18 nm gold-labeled secondary
antibodies alone showed no immunogold labeling
(data not shown).
The close association of elastin globules with
MAGP-2 molecules within the matrix suggests
that MAGP-2 facilitates elastic fiber assembly
from its matrix-associated form, as opposed to its
soluble non-bound form. In order to selectively
assess the effect of matrix-associated and soluble
MAGP-2 on elastic fiber assembly, we took
advantage of the inability of soluble MAGP-2
provided as an external source from conditioned
medium to assemble within the matrix of ME-TE
cells (data not shown). ME-TE cells were cultured
in high content MAGP-2-containing media
conditioned
from
dox-regulated
MG-TE
overexpressing MAGP-2, and then elastic fibers
analyzed by EGFP fluorescence. Although
conditioned media were renewed every other day
for 2 weeks, ME-TE cells cultured in MAGP-2containing medium (dox -) showed no increase in
elastic fibers compared to cells cultured in MAGP2-free media (dox +) (data not shown). Since
soluble MAGP-2 failed to stimulate elastic fiber

assembly in matrix-associated MAGP-2-free

conditions, this result supports the electron
microscopy data, indicating that matrix-associated
MAGP-2 mediates MAGP-2-induced-elastic fiber
assembly.
Matrix-associated fibrillin-1 regulates matrix
assembly of MAGP-2 - Since fibrillin-1 binds
MAGP-2 (18,32), we studied whether matrixassociated fibrillin-1 regulates matrix assembly of
MAGP-2. We used tet-regulated MEF cell lines
conditionally overexpressing a Marfan-like
fibrillin- 1 deletion mutant (cFbn) where the Cterminal region was replaced at amino acid 2780
by EGFP. EGFP fluorescence analysis of MEFcFbn cells cultured without dox (dox -) showed
high intracellular levels of EGFP- cFbn but no
matrix-associated form of this mutant (Fig. 4A, b).
dox treatment (dox +) completely repressed
intracellular expression of EGFP- cFbn (Fig. 4A,
a). Immunostaining of fibrillin-1 with antibody
detecting both mutant and wild-type fibrillin-1
molecules showed that overexpression of EGFPcFbn dramatically inhibited matrix assembly of
fibrillin-1 (Fig. 4A, red-stained fibrils, c vs. d and i
vs. j). Western blot analysis of EGFP-cFbn using
an antibody targeting both wild-type and mutant
forms of fibrillin-1 showed higher levels of total
fibrillin-1 in both cell layers and culture media of
cell overexpressing EGFP-cFbn (Fig. 4B, top
panel, dox vs. dox +), indicating that EGFPcFbn, even though not detected within the matrix
by immunofluorescence, is secreted. This suggests
a dominant-negative effect of the c-truncated
EGFP-cFbn
mutant
on
extracellular
polymerization of fibrillin-1 monomers, similar to
cases of severe Marfan syndrome caused by ctruncated fibrillin-1 (33).
Importantly, inhibition of matrix assembly of
fibrillin-1 by cFbn dramatically decreased levels
of matrix-associated MAGP-2 (Fig. 4C, g vs. h).
Since overexpression of cFbn did not alter
medium levels of soluble MAGP-2 available to the
matrix (Fig. 4D, dox + vs. dox -) and since
fibrillins are the only known matrix proteins
binding MAGP-2, this suggested that fibrillin-1associated matrix regulates matrix assembly of
MAGP-2, by binding MAGP-2. This in turn
suggested that MAGP-2 overexpression might
stimulate elastic fiber assembly via stimulating
matrix assembly of fibrillin-1.

MAGP-2 does not stimulate elastic fiber

assembly by altering matrix assembly of fibrillin-1
To investigate whether MAGP-2 stimulates
elastic fiber assembly by altering matrix assembly
of fibrillins, we looked at the effect of MAGP-2
overexpression on matrix-associated fibrillin-1 in
our dox-regulated MG-TE cells (Of note, fibrillin2 is not expressed in MEF cells, data not shown).
MAGP-2 overexpression was associated with an
extended matrix-associated fibrillin-1 network that
largely colocalized with both matrix associated
MAGP-2 (Fig. 5A, a, c and e) and elastic fibers
(Fig. 5B, a, c and e). However, while doxrepression
of
MAGP-2
overexpression
dramatically reduced both matrix-associated
MAGP-2 and elastic fibers, as expected (Fig. 5A
and 5B, b), it did not alter matrix-associated
fibrillin-1 (Fig. 5A, and 5B, d). These data indicate
that MAGP-2 does not stimulate elastic fiber
assembly by altering matrix assembly of fibrillin1. They also suggest that in our in vitro system
matrix-associated fibrillin-1 is necessary but not
sufficient for elastic fiber assembly, as it also
required MAGP-2.
MAGP-2 does not stimulate elastic fiber
assembly by altering matrix-associated levels of
MAGP-1, fibulin-2, fibulin-5 or emilin-1 To
further understand the molecular basis for MAGP2-induced elastic fiber assembly, we next
investigated whether elastic fiber components
other than fibrillin-1 might mediate the stimulating
effect of MAGP-2 on elastic fibers, including
major structural component of microfibrils
MAGP-1, as well as microfibril-elastin interface
molecules fibulin-2, fibulin-5 and emilin-1
implicated in elastogenesis (3-5). Using doxregulated MG-TE cells, we looked at the effect of
MAGP-2 overexpression on matrix-associated
MAGP-1,
fibulin-2,
and
emilin-1
by
immunostaining and on fibulin-5 by western blot.
MG-TE cells overexpressing MAGP-2 showed an
extended elastic fiber network that largely
colocalized with matrix-associated MAGP-1 (Fig.
6A, a, c and e), fibulin-2 (Fig. 6B, a, c and e) and
emilin-1 (Fig. 6C, a, c and e). However, while
dox-repression of MAGP-2 overexpression
dramatically reduced elastic fiber levels, it did not
alter matrix-associated levels of MAGP-1 (Fig.
6A, b and d), fibulin-2 (Fig. 6B, b and d) and
emilin-1 (Fig. 6C, b and d). Similar, MAGP-2
overexpression did not alter levels of fibulin-5

(Fig. 7). These data indicate that MAGP-2 does

not stimulate elastic fiber assembly by altering
matrix-associated levels of MAGP-1, fibulin-2,
fibulin-5 or emilin-1, and that their presence is
unable to compensate for a lack of MAGP-2.
Matrix assembly of elastic fibers and MAGP-2
induced by MAGP-2 overexpression is not
mediated by the MAGP-2 RGD MAGP-2
contains an RGD, integrin binding motif. Such a
motif is known to be required in matrix assembly
of proteins such as fibronectin as reported for
fibronectin (34). To investigate whether binding of
MAGP-2 to integrin receptors may mediate
MAGP-2-induced elastic fiber assembly, we
mutated the RGD sequence of MAGP-2 to RVD
and looked at the effect of this mutation on elastic
fiber assembly using cells conditionally
overexpressing this MAGP-2 mutant (MGRGDTE cells). Similar to MG-TE cells overexpressing
wild-type
MAGP-2,
MGRGD-TE
cells
overexpressing MAGP-2 with a mutated RGD
showed concomitant, dramatic increase in matrix
levels of MAGP-2 (Fig. 8, a vs. b) and elastic
fibers (Fig. 8, c vs. d) that colocalized (Fig. 8, e).
These results indicate that the MAGP-2 RGD does
not regulate elastic fiber assembly, suggesting that
MAGP-2 does not stimulate elastic fiber assembly
by interacting with an integrin receptor such as
v3 (22). Moreover, since matrix assembly of
MAGP-2 is increased upon overexpression of the
MAGP-2 RGD mutant, this also indicates that
matrix assembly of MAGP-2 is not mediated by its
RGD.
DISCUSSION
Using an in vitro model featuring steady-state
levels of extracellular soluble tropoelastin
monomers, we show in this study that microfibrilassociated-MAGP-2 stimulates elastic fiber
assembly, independently of its RGD sequence.
MAGP-2 did not stimulate elastic fiber assembly
by altering microfibril assembly, as primary
structural components of microfibrils, fibrillin-1
and MAGP-1, were not affected by MAGP-2.
Since fibrillin-2 is not expressed in MEF cells,
evaluation of fibrillin-1 along with MAGP-1
represents a fair assessment of the complete
elastin-associated microfibril structure, especially
since fibrillin-1 strictly colocalized with elastic
fibers (Fig. 5B). Our study also showed that
MAGP-2 does not stimulate elastic fiber assembly

by altering levels of elastogenesis-mediating

elastin-associated molecules, fibulin-2, fibulin-5
and emilin-1. Although MAGP-2 is not known to
bind to these molecules, we cannot exclude the
possibility that MAGP-2 might indirectly exert its
effect on elastin deposition by modifying the
conformation or ultrastructural organization of one
of these, possibly altering their interaction with
tropoelastin or cell surface molecules.
Additional molecular insights into the
regulation of elastic fiber assembly by MAGP-2
were provided by a detailed morphological and
ultrastructural analysis of elastic fibers assembled
by cells in the presence or the absence of
exogenous MAGP-2. In the absence of exogenous
MAGP-2 stimulation, the elastic fiber network is
predominantly represented by large tropoelastin
globules that are specifically aligned in a
discontinuous, loose structure defined by
connecting, developing, frail elastic fibers (Fig.
2A and 2B). These large globules are also present
upon overexpression of MAGP-2, although they
appear to have coalesced into larger, stretched
aggregates to form elastic fibers. Formation of
elastin globules corresponds to the elastic fiber
microassembly
step,
and
results
from
coacervation, an entropically driven, inverse
temperature transition produced by the interaction
of the tropoelastin hydrophobic domains, causing
tropoelastin monomers to associate in large
globular polymers (35). Since these elastin
globules are present in both cells overexpressing
or not MAGP-2, this suggests that MAGP-2 does
not stimulate elastic fibers assembly by promoting
or facilitating coacervation.
Cross-linking of tropoelastin monomers by
lysyl oxidase, another important step in elastic
fiber assembly, essentially in late stages of fiber
maturation, provides an alternative mechanism for
MAGP-2-induced
elastic
fiber
assembly.
However, recent work has clearly shown that
tropoelastin cross-linking is not required for the
deposition of tropoelastin onto developing elastic
fiber (6). If stimulation of elastic fiber assembly
by MAGP-2 were to be based on increased crosslinking activity, elastic fibers in exogenous
MAGP-2-free conditions would likely have a
thicker structure than actually seen (more similar
in appearance with fibers synthesized in the
presence of MAGP-2), as deposition of soluble
tropoelastin onto microfibrils would occur to the

same extent as with exogenous MAGP-2 (6).

Therefore, it appears unlikely that MAGP-2
augments elastic fiber assembly by stimulating
tropoelastin cross-linking. Consistent with this,
Affymetrix microarray analysis of gene expression
in conditional MG-TE cells upon MAGP-2
overexpression showed no change in expression of
lysyl oxidase and two of its homologs, lysyl
oxidase like protein 1 and 3 (data not shown).
One intriguing molecular aspect of the
stimulation of elastic fiber assembly by MAGP-2
is the co-localization of elastin aggregates and
MAGP-2-associated microfibrils. This suggests
that MAGP-2 controls elastic fiber assembly from
its matrix-associated form. The most obvious
mechanism in that context would be a direct
interaction of matrix-associated MAGP-2 with
elastin.
However,
co-immunoprecipitation
experiments using medium conditioned from MGTE cell cultures overexpressing MAGP-2 and
tropoelastin did not reveal any significant binding
of MAGP-2 to elastin (data not shown), consistent
with previous binding studies using the yeast
double hybrid system that showed direct
interaction of MAGP-2 with fibrillins but not
elastin (18). Alternatively, MAGP-2 could
indirectly affect tropoelastin deposition through
interfacing elastic fiber-associated molecules.
Fibrillin-1 is a good candidate since it binds both
MAGP-2
and
tropoelastin
(14,18,32).
Hypothetically, binding of MAGP-2 to fibrillin-1
may stimulate matrix assembly of fibrillin-1,
which in turn would stimulate tropoelastin
deposition onto fibrillin-1 rich microfibrils.
However, we show that MAGP-2 overexpression
does not alter assembly of matrix-associated
fibrillin-1.
To address the question of how matrixassociated MAGP-2 regulates elastic fiber
assembly, general mechanisms underlying
alignment of elastin globules onto microfibrils first
have to be considered. Latest advances in
microscopy technology have recently shown the
extraordinary dynamism of cells within the matrix,
providing evidence of cell motion-mediated elastic
fiber assembly (36,37). Dynamic imaging
microscopy has shown that elastic fiber assembly
is a two-step process that includes microassembly
and macroassembly. Microassembly corresponds
to the appearance of small cell surface-associated
tropoelastin globules that increase in size with

time. Macroassembly corresponds to the transfer

of these globules onto pre-existing elastic fiber
(early stage) where they coalesce into larger
aggregate structures that eventually will be
stretched into fibrillar form (late stage). As
discussed above, our data do not suggest that
MAGP-2 alters microassembly since elastin
globules are present in cell cultures whether or not
MAGP-2 is overexpressed. If MAGP-2 affects the
late stage of elastic fiber macroassembly where
microfibril-associated elastin globules coalesce
into larger aggregates, MAGP-2 overexpression
would be associated with alterations in
microfibrils. Our data show no alteration of major
microfibril structural components, fibrillin-1 and
MAGP-1, suggesting MAGP-2 does not alter this
late stage of macroassembly. Alternatively,
MAGP-2 may regulate early stages of elastic fiber
macroassembly, including transfer of tropoelastin
globules from the cell membrane onto the
developing fiber. In that stage, as the elastinproducing cell interacts with a preexisting elastic
fiber, the tropoelastin globules on the cell surface
are redistributed toward the site of elastic fiber
contact and eventually become incorporated into
the fibrous structure (37). The molecular basis of
elastic fiber microassembly is not known but
likely involves cell-matrix interaction factors,
maybe MAGP-2. Consistent with this view,
MAGP-2 interacts with Jagged-1 (23), a transmembrane protein, ligand for its receptor Notch
(38), that plays a critical role in matrix remodeling
via alteration of cell-matrix interaction and cell
motility (39). Therefore, matrix-associated
MAGP-2 might regulate elastic fiber assembly by
binding tropoelastin-producing cells via Jagged-1,
allowing targeted release of elastin globules from
the cell onto the MAGP-2-containing fiber.
We have previously reported that MAGP-2
stimulates type I collagen expression (25). The
present study identifies a function of MAGP-2 in
elastogenesis. Since collagen and elastic fibers are
widely expressed and play key roles in structural
integrity and remodeling of connective tissues,
MAGP-2 stands as a critical organizer of the
extracellular matrix and may greatly impact on
connective tissue homeostasis.
ACKNOWLEDGMENTS
The authors thank Marilyn Levy and Russel
Knutsen (Department of Cell Biology &

Physiology, Washington University School of

Medicine, St Louis, MO), as well as Paul Toselli
(Department of Biochemistry, Boston University
School of Medicine, Boston MA) for their
technical assistance in electron microscopy. This
work was supported by the Scleroderma

Foundation (Grant 006-004), the Boston

University Department of Medicine (Pilot grant
01-499-82-179-1) and National Institute of
Arthritis and Musculoskeletal and Skin Diseases
(Grant R01AR051089-01A2).

REFERENCES
1. Milewicz, D. M., Urban, Z., and Boyd, C. (2000) Matrix Biol 19(6), 471-480
2. Kielty, C. M., Sherratt, M. J., and Shuttleworth, C. A. (2002) J Cell Sci 115(Pt 14), 2817-2828
3. Chu, M. (2004) Communication - Second National Meeting of the American Society for Matrix
Biology
4. Yanagisawa, H., Davis, E. C., Starcher, B. C., Ouchi, T., Yanagisawa, M., Richardson, J. A., and
Olson, E. N. (2002) Nature 415(6868), 168-171
5. Zanetti, M., Braghetta, P., Sabatelli, P., Mura, I., Doliana, R., Colombatti, A., Volpin, D., Bonaldo, P.,
and Bressan, G. M. (2004) Mol Cell Biol 24(2), 638-650
6. Kozel, B. A., Ciliberto, C. H., and Mecham, R. P. (2004) Matrix Biol 23(1), 23-34
7. Mariencheck, M. C., Davis, E. C., Zhang, H., Ramirez, F., Rosenbloom, J., Gibson, M. A., Parks, W.
C., and Mecham, R. P. (1995) Connect Tissue Res 31(2), 87-97
8. Cleary, E. G., and Gibson, M. A. (1983) Int Rev Connect Tissue Res 10, 97-209
9. Greenlee, T. K., Jr., Ross, R., and Hartman, J. L. (1966) J Cell Biol 30(1), 59-71
10. Sakai, L. Y., Keene, D. R., and Engvall, E. (1986) J Cell Biol 103(6 Pt 1), 2499-2509.
11. Zhang, H., Apfelroth, S. D., Hu, W., Davis, E. C., Sanguineti, C., Bonadio, J., Mecham, R. P., and
Ramirez, F. (1994) J Cell Biol 124(5), 855-863
12. Gayraud, B., Keene, D. R., Sakai, L. Y., and Ramirez, F. (2000) J Cell Biol 150(3), 667-680.
13. Lemaire, R., Farina, G., Kissin, E., Shipley, J. M., Bona, C., Korn, J. H., and Lafyatis, R. (2004)
Arthritis Rheum 50, 915-926

14. Clarke, A. W., Wise, S. G., Cain, S. A., Kielty, C. M., and Weiss, A. S. (2005) Biochemistry 44(30),
10271-10281
15. Zhang, H., Hu, W., and Ramirez, F. (1995) J Cell Biol 129(4), 1165-1176.
16. Pereira, L., Andrikopoulos, K., Tian, J., Lee, S. Y., Keene, D. R., Ono, R., Reinhardt, D. P., Sakai, L.
Y., Biery, N. J., Bunton, T., Dietz, H. C., and Ramirez, F. (1997) Nat Genet 17(2), 218-222.
17. Jensen, S. A., Reinhardt, D. P., Gibson, M. A., and Weiss, A. S. (2001) J Biol Chem 276(43), 3966139666.
18. Penner, A. S., Rock, M. J., Kielty, C. M., and Shipley, J. M. (2002) J Biol Chem 16, 16
19. Trask, B. C., Trask, T. M., Broekelmann, T., and Mecham, R. P. (2000) Mol Biol Cell 11(5), 14991507
20. Reinboth, B., Hanssen, E., Cleary, E. G., and Gibson, M. A. (2002) J Biol Chem 277(6), 3950-3957.
21. Finnis, M. L., and Gibson, M. A. (1997) J Biol Chem 272(36), 22817-22823.
22. Gibson, M. A., Leavesley, D. I., and Ashman, L. K. (1999) J Biol Chem 274(19), 13060-13065.
23. Nehring, L. C., Miyamoto, A., Hein, P. W., Weinmaster, G., and Shipley, J. M. (2005) J Biol Chem
280(21), 20349-20355
24. Hansen, H. E., Hew, F. H., Moore, G., and Gibson, M. A. (2004) J Biol Chem 279, in press
25. Lemaire, R., Korn, J. H., Shipley, J. M., and Lafyatis, R. (2005) Arthritis Rheum 52(6), 1812-1823
26. Gibson, M. A., Hatzinikolas, G., Kumaratilake, J. S., Sandberg, L. B., Nicholl, J. K., Sutherland, G.
R., and Cleary, E. G. (1996) J Biol Chem 271(2), 1096-1103.
27. Gibson, M. A., Finnis, M. L., Kumaratilake, J. S., and Cleary, E. G. (1998) J Histochem Cytochem
46(8), 871-886.
28. Handford, P. A., Downing, A. K., Reinhardt, D. P., and Sakai, L. Y. (2000) Matrix Biol 19(6), 457470.
29. Kissin, E. Y., Lemaire, R., Korn, J. H., and Lafyatis, R. (2002) Arthritis Rheum 46(11), 3000-3009
30. Lemaire, R., Korn, J. H., Shipley, J. M., and Lafyatis, R. (2004) J Invest Dermatol 123, 1063-1069
31. Gossen, M., and Bujard, H. (1992) Proc Natl Acad Sci U S A 89(12), 5547-5551

10

32. Hanssen, H. E., Hew, F. H., Moore, G., and Gibson, M. A. (2004) J Biol Chem 279, 29185-29194
33. Raghunath, M., Kielty, C. M., and Steinmann, B. (1995) J Mol Biol 248(5), 901-909.
34. Wu, C., Keivens, V. M., O'Toole, T. E., McDonald, J. A., and Ginsberg, M. H. (1995) Cell 83(5),
715-724
35. Cox, B. A., Starcher, B. C., and Urry, D. W. (1974) J Biol Chem 249(3), 997-998
36. Czirok, A., Zach, J., Kozel, B. A., Mecham, R. P., Davis, E. C., and Rongish, B. J. (2006) J Cell
Physiol 207(1), 97-106
37. Kozel, B. A., Rongish, B. J., Czirok, A., Zach, J., Little, C. D., Davis, E. C., Knutsen, R. H.,
Wagenseil, J. E., Levy, M. A., and Mecham, R. P. (2006) J Cell Physiol 207(1), 87-96
38. Lindsell, C. E., Shawber, C. J., Boulter, J., and Weinmaster, G. (1995) Cell 80(6), 909-917
39. Lindner, V., Booth, C., Prudovsky, I., Small, D., Maciag, T., and Liaw, L. (2001) Am J Pathol
159(3), 875-883

FIGURE LEGENDS
Figure 1. MEF cells do not efficiently assemble tropoelastin onto developing elastic fibers. (A) MEF
cells were cultured for 1 week, in the presence of TGF-, and analyzed for elastic fibers by
immunostaining. Nuclei were stained with Hoechst (blue). Bar = 10 m. (B) MEF cells constitutively
overexpressing EGFP-tagged tropoelastin (MEF-TE cells) were cultured for 1 week in the presence of
TGF-. Elastic fibers were then analyzed by both EGFP fluorescence (green) and immunoflurorescence
(red). Overlapping EGFP fluorescence and immunofluorescence appears in yellow. (C, D and E) MEF
cells conditionally overexpressing MAGP-2 and constitutively EGFP-tagged elastin (MG-TE cells) were
cultured for 1 week with and without dox, in the presence or the absence of TGF-. Endogenous and
exogenous MAGP-2 secreted in the medium was analyzed by western blot (C). Elastin mRNA expression
was analyzed by northern blot (D) and level of elastin in the medium by western blot (E).
Figure 2. MAGP-2 overexpression stimulates elastic fiber assembly. (A) MG-TE cells conditionally
overexpressing MAGP-2, and constitutively elastin-EGFP, were cultured for 1 week with and without
dox, in the presence of TGF-. Elastic fibers were then analyzed by EGFP fluorescence (green). (B) In
parallel cultures, elastic fibers (green) and matrix-associated MAGP-2 (red) were also co-analyzed,
respectively by EGFP fluorescence and immunofluorescence. Nuclei were stained with Hoechst (blue).
Bar = 10 m.
Figure 3. MAGP-2-induced matrix-associated elastin primarily colocalizes with MAGP-2-associated
microfibrils and form elastic fibers. MG-TE cells overexpressing MAGP-2-V5 and elastin-EGFP were
cultured for 2 weeks without dox in the presence of TGF-. Elastic fibers were then analyzed by immuno-

11

electron microscopy. MAGP-2 was immunolocalized using 6 nm-gold-labeled anti-V5 antibody (A and
B). MAGP-2 and elastin were co-immunolocalized using 6 nm-gold-labeled anti-V5 and 18 nm-gold
labeled anti-EGFP antibodies (C) or 18 nm-gold-labeled anti-MAGP-2 and 6 nm-gold labeled anti-elastin
antibodies (D). Arrowheads points to MAGP-2-free microfibril stretches. Letter E points to some of the
elastin globules. Bar = 100 nm.
Figure 4. Inhibition of matrix assembly of fibrillin-1 in MEF cells overexpressing a Marfan-like
fibrillin-1 mutant is associated with dramatic decrease in matrix assembly of MAGP-2. (A) MEF
cells conditionally overexpressing an EGFP-tagged truncated fibrillin-1 mutant were cultured for 1 week
with or without dox, in the presence or the absence of TGF-. Total matrix-associated fibrillin-1 (red) and
C-terminal truncated fibrillin-1 mutant cFbn (green) were co-analyzed, respectively by EGFP
fluorescence and immunoflurorescence using polyclonal antibody to fibrillin-1 N-terminal half. Nuclei
were stained with Hoechst (blue). Bar = 10 m. (B) Levels of total and mutant fibrillin-1 were analyzed
in cell layer and culture medium by western blot using antibody to the fibrillin-1 N-terminal half. Total
protein levels were controlled by ponceau stain. (C) In parallel cultures, matrix-associated MAGP-2 (red)
and cFbn (green) were also co-analyzed, respectively by immunofluorescence and EGFP fluorescence.
(D) Level of MAGP-2 in the culture medium was also analyzed by western blot.
Figure 5. MAGP-2 overexpression does not stimulate elastic fiber assembly by stimulating matrixassociated fibrillin-1. (A) MEF cells conditionally overexpressing V5-tagged MAGP-2 were cultured for
1 week with or without dox, in the presence of TGF-. Matrix-associated MAGP-2 (green) and fibrillin-1
(red) were co-analyzed by immunofluorescence using antibody, respectively to V5 and fibrillin-1. Colocalizing matrix-associated MAGP-2 and fibrillin-1 appear in yellow. (B) MG-TE cells overexpressing
conditionally MAGP-2, and constitutively elastin-EGFP, were cultured for 1 week with and without dox,
in the presence of TGF-. Elastic fibers (green) and matrix-associated fibrillin-1 (red) were then coanalyzed respectively by EGFP fluorescence and immunofluorescence. Co-localizing elastic fibers and
matrix-associated fibrillin-1 appear in yellow. Nuclei were stained with Hoechst (blue). Bar = 10 m.
Figure 6. MAGP-2 overexpression does not stimulate elastic fiber assembly by stimulating matrix
assembly of MAGP-1, fibulin-2 and emilin-1. MG-TE cells conditionally overexpressing MAGP-2, and
constitutively elastin-EGFP, were cultured for 1 week with and without dox, in the presence of TGF-.
Elastic fibers (EFGP-fluorescence, green) were then co-analyzed with either matrix-associated MAGP-1
(A), fibulin-2 (B) or emilin-1 (C) by immunofluorescence (red). Co-localizing molecules appear in
yellow. Nuclei were stained with Hoechst (blue). Bar = 10 m.
Figure 7. MAGP-2 overexpression does not increase elastic fiber assembly by stimulating fibulin-5.
MG-TE cells conditionally overexpressing MAGP-2, and constitutively tropoelastin, were cultured for 1
week with or without dox, in the presence or the presence of TGF-. Matrix and cell layers were then
lysed in SDS-PAGE buffer / me and proteins analyzed for MAGP-2 and fibulin-5 by western blot.
Figure 8. Matrix assembly of elastic fibers and MAGP-2 induced by MAGP-2 overexpression is not
mediated by the MAGP-2 RGD. MGRGD-TE cells conditionally overexpressing MAGP-2 with a
mutated RGD, and constitutively elastin-EGFP, were cultured for 1 week with and without dox, in the
presence of TGF-. Elastic fibers (EFGP-fluorescence, green) were then co-analyzed with matrixassociated MAGP-2 by immunofluorescence (red). Co-localizing molecules appear in yellow. Nuclei
were stained with Hoechst (blue). Bar = 10 m.

12