(Received for publication, November 26, 1996, and in revised form, January 9, 1997)
From the Shriners Hospital for Children, Portland,
Oregon 97201 and the ¶ Department of Biochemistry and
Molecular Biology, Oregon Health Sciences University,
Portland, Oregon 97201
Velocity sedimentation experiments using authentic fibrillin-1 demonstrated sedimentation coefficients of s20,w0 = 5.1 ± 0.1 in the Ca2+ form and s20,w0 = 6.2 ± 0.1 in the Ca2+-free form. Calculations based on these results and the corresponding molecular mass predicted a shortening of fibrillin by ~25% and an increase in width of ~13-17% upon removal of Ca2+. These observations were confirmed by analysis of Ca2+-loaded and Ca2+-free rotary shadowed fibrillin molecules. Analysis of recombinant fibrillin-1 subdomain rF17, consisting primarily of an array of 12 Ca2+-binding epidermal growth factor (cbEGF)-like repeats, by analytical ultracentrifugation and rotary shadowing further confirmed Ca2+-dependent structural changes in the tertiary structure of fibrillin-1. Based on these results, the contribution of a single cbEGF-like repeat to the length of tandem arrays is predicted to be ~3 nm in the Ca2+ form. Ca2+-free forms demonstrated a decrease of 20-30% in length, indicating significant structural changes of these motifs when they occur in tandem. Circular dichroism measurements of rF17 in the presence and absence of Ca2+ indicated secondary structural changes within and adjacent to the interdomain regions that connect cbEGF-like repeats. The results presented here suggest a flexible structure for the Ca2+-free form of fibrillin which becomes stabilized, more extended, and rigid in the Ca2+ form.
Fibrillins are integral components of extracellular supramolecular aggregates, called microfibrils (10-12 nm in diameter), which are found both in conjunction with elastic fibers and as isolated microfibrils (1, 2). Characteristic of the two highly homologous fibrillins is the mosaic composition of different types of extracellular modules (2-6). Most of the thread-like fibrillin molecule (7) is contributed by 43 epidermal growth factor (EGF)1-like motifs which contain a consensus sequence for calcium binding (cb).
Calcium binding to fibrillin is now well established. It has been demonstrated with authentic fibrillin purified from cell culture medium (5), with pepsin-resistant fragments of fibrillin (8), with recombinant subdomains of fibrillin-1 (9, 10), and with synthetic peptides (11, 12). Calcium binding to fibrillin has been suggested to be important for protein-protein interaction (13), stabilization of the lateral packing of the microfibrils (14), protection of fibrillin against proteolysis (15), and maturation of a proform of fibrillin (16). The importance of calcium binding to the fibrillins is especially emphasized by the fact that mutations, predicted to disturb calcium binding, result in the pathological manifestations of the Marfan syndrome (reviewed in Ref. 17) and congenital contractural arachnodactyly (18).
The EGF-like motif is a widely used module found in more than 70 extracellular proteins (reviewed in Ref. 19). The motif consists of
40-50 amino acid residues with six highly conserved cysteine residues
that form three disulfide bonds. In a subset of EGF-like modules, a
characteristic pattern of amino acid residues ((D/N)X(D/N)(Q/E)Xn(D*/N*)Xm(Y/F);
residues with an asterisk are potentially -hydroxylated) has been
identified, which is responsible for calcium binding (20-22). The
Ca2+-binding site of cbEGF-like motifs has been well
characterized by nuclear magnetic resonance studies of a single
cbEGF-like motif of factor X (23), of a pair (numbers 32-33) of
cbEGF-like motifs in fibrillin-1 (24), and by x-ray diffraction of a
isolated cbEGF-like motif from human factor IX (25) and of the complex of factor VIIa with soluble tissue factor (26). From these studies, it
is clear that Ca2+ binding occurs in a
NH2-terminal pocket of cbEGF-like repeats between the major
two-stranded
-sheet and the NH2-terminal loop.
In many proteins of the extracellular matrix, cbEGF-like repeats are arranged in tandemly repeated arrays from 2 to up to 36 repeats. Examples are proteins for specification of cell fate (27-29), blood coagulation factors (reviewed in Ref. 30), anticoagulation factors (31), basement membrane proteins (32), and the fibrillins (2-6). While Ca2+-binding sites in single cbEGF-like motifs are well understood (23-25), less information is available on how Ca2+ binding affects the secondary and tertiary structure of tandemly repeated cbEGF-like motifs. Based on asymmetric crystallization of a cbEGF-like module, Rao and co-workers (25) suggested a Ca2+-stabilized helical arrangement for tandemly repeated cbEGF-like motifs (25). Contrary to this hypothesis, nuclear magnetic resonance analyses of a covalently linked pair of cbEGF-like repeats suggested an extended and rigid conformation for tandem repeats (24).
To clarify these important issues, we investigated the effect of calcium on the shape and secondary structure of authentic fibrillin and a recombinant subdomain comprising the longest stretch of cbEGF-like repeats present in fibrillin-1. We found that in the presence of Ca2+ the shape of fibrillin and of the recombinant subdomain was more extended compared to the Ca2+-free forms. Additionally, circular dichroism measurements indicated subtle changes within the interdomain regions between cbEGF-like repeats. We suggest that these interdomain regions are more flexible without Ca2+ and more rigid in the presence of Ca2+.
Fibrillin was purified from fibroblast cell culture medium according to previously described methods (7). Design of recombinant fibrillin-1 subdomain rF17, comprising the longest stretch of cbEGF-like repeats and the preceding 8-cysteine motif (amino acid residues 952-1527), and purification of rF17 by chelating chromatography has been described (15). Concentrations of purified rF17 were determined in triplicate after hydrolysis (6 M HCl, 110 °C, 24 h) using an amino acid analyzer (Beckman 6300).
Velocity SedimentationNormal human skin fibroblasts (low passage numbers) were grown to confluency in 500-cm2 cell culture flasks (Nunc). The cells were incubated with serum-free medium (Dulbecco's modified Eagle's medium) for 48 h. 200 ml of the serum-free medium was treated with 2 µl/ml diisopropyl fluorophosphate and passed over gelatin-Sepharose 4B (8 ml bed volume; Pharmacia) to remove fibronectin. The flow-through was concentrated to ~3 ml by ultrafiltration, dialyzed against 50 mM Tris-HCl, pH 7.5, 150 mM NaCl (TBS), and then supplemented with either 5 mM CaCl2 or 5 mM EDTA. Aliquots (100 µl) were then pipetted on top of a 5-20% (w/v) sucrose gradient (3.6-ml total volume) buffered with TBS including either 5 mM CaCl2 or 5 mM EDTA in Polyallomer tubes (11 × 60 mm; Beckman). Ultracentrifugation experiments were performed for 17 h at 40,000 rpm (average relative centrifugal force = 164,000 × g) at 4 °C in a Beckman L8-M ultracentrifuge using a Beckman SW 60Ti rotor. After a small hole was pricked with a pin in the bottom of the tubes, 8 drop fractions were collected. Aliquots (10 µl) of the fractions were mixed with 10 µl of 2-fold concentrated SDS nonreducing sample buffer and then separated by SDS-gel electrophoresis on 4.5% (w/v) acrylamide gels (33). After transfer of the proteins to nitrocellulose membranes (Micron Separations Inc.) in 10 mM sodium borate, pH 9.2, at 0.4 Å for 45 min, fibrillin-1 was visualized by a typical Western blot analysis using monoclonal antibody 201 (~10 µg/ml) and goat anti-mouse IgG horseradish peroxidase conjugate (1:2000 diluted; Bio-Rad). Monoclonal antibody 201 is specific for fibrillin-1 and does not react with fibrillin-2.2 The membranes were developed with SuperSignalTM as instructed by the manufacturer (Pierce). X-ray films (X-Omat AR; Kodak) were exposed to the membranes for a few seconds and then developed. The intensity of fibrillin-1 bands were then quantified on a Macintosh (9500/132) using the public domain NIH Image program version 1.6 (developed at the National Institutes of Health), which is available on the Internet.3
Sedimentation coefficients from sucrose gradients were calculated as described previously (34). Calculations predicting the shape of fibrillin-1 were performed according to Bloomfield et al. (35).
After purification of rF17 by chelating chromatography, the subdomain was dialyzed against TBS and then supplemented to a final concentration of either 5 mM CaCl2 or 5 mM EDTA. The subdomain was then analyzed in concentrations of 0.55-2.9 mg/ml by analytical ultracentrifugation in a Spinco Model E centrifuge (Beckman) equipped with an optical scanner operating at 280 nm. Velocity sedimentation experiments were performed at 52,000 rpm in double sector cells at 20 °C. The absorbance scans were analyzed by the Ultrascan software (supplied by Dr. Borries Demeler).
Rotary Shadowing and Electron MicroscopyFibrillin purified from cell culture medium was dialyzed against H2O and then supplemented with either 2 mM CaCl2 or 10 mM EDTA. Recombinant subdomain rF17 was dialyzed against H2O including 2 mM CaCl2 or 0.5 mM EDTA. Samples were diluted to a final concentration of 70% (v/v) glycerol, sprayed onto freshly cleaved mica, and dried under vacuum (Balzers BAE 250 evaporator). Rotary shadowing was performed as described previously (36). Replicas were examined at 80 kV in a transmission electron microscope (Philips 410LS). Measurements of well resolved particles were performed using the public domain NIH Image program version 1.6 as described above.
Circular Dichroism MeasurementsTo deplete rF17 of Ca2+, pools of rF17 after chelating chromatography were supplemented to a final concentration of 10 mM EDTA and then dialyzed against 10 mM MOPS, pH 7.2, containing 5 g of Chelex 100 resin (Bio-Rad) per 100 ml of buffer. Alternatively, Chelex 100-treated rF17 was dialyzed against 10 mM MOPS, pH 7.2, 100 mM NaCl, 0.2 mM EDTA. Spectra from 260 to 180 nm were recorded on a Jasco J-500A instrument in the presence and absence of 2 mM CaCl2 in a 0.1-mm cell at 25 °C. Secondary structure analysis was performed with the variable selection method (37).
The experiments described were designed to study the effects of calcium on the shape and secondary structure of fibrillin.
Results from velocity sedimentation experiments using fibrillin-1 were
visualized by fluorescence-enhanced Western blot analysis using
monoclonal antibody 201, specific for fibrillin-12 (Fig.
1). The peak fractions of fibrillin-1, determined by
densitometric measurements of duplicate samples, clearly differed for
sedimentation in the presence of CaCl2 (fraction 16.2)
versus EDTA (fraction 13.9). These data resulted in
sedimentation coefficients of
s20,w0 = 5.1 ± 0.1 (± S.D.) for the Ca2+ form and
s20,w0 = 6.2 ± 0.1 (± S.D.) for the Ca2+-free form (Table
I). Predicted shapes were calculated using the
experimentally determined sedimentation coefficients, estimates for
partial specific volume and degree of hydration, and estimated molecular masses (35) (Table I). Based on previous data (10), we
assumed that 6.5% of the total mass of fibrillin-1 is contributed by
N-linked oligosaccharides. Since it is not clear whether the fibrillin-1 used was processed as has been suggested (10, 13, 38),
predicted shapes were calculated for the non-processed (334 kDa) and
the processed (312 kDa) form of fibrillin-1 (Table I). Upon removal of
Ca2+ by EDTA, the relative length of fibrillin molecules is
predicted to decrease by about 25%, whereas the width of the molecules
is predicted to increase by 13-17%. Assuming a degree of hydration of
0.1 cm3/g, a rod-like shape, and non-processed fibrillin,
the length of the molecules would be 156.1 nm in the presence of
Ca2+ versus 117.0 nm when Ca2+ is
removed. Assuming the fibrillin used was processed, the molecules would
be predicted to decrease from 142.0 nm in the Ca2+ form to
106.7 nm in the Ca2+-free form. These results demonstrate
significant structural changes of fibrillin in the presence and absence
of Ca2+ ions.
|
Purified fibrillin was visualized by rotary shadowing and transmission
electron microscopy in the presence of low amounts of either
CaCl2 or EDTA (Fig. 2). The molecules were
nicely resolved in the presence of CaCl2 and appeared as
extended thread-like particles as reported previously (7) (Fig.
2A). In the presence of EDTA single particles often appeared
shorter, somewhat wider and more diffuse (Fig. 2B). Length
measurements resulted in 140.3 ± 14.9 nm (± S.D.;
n = 41) for the Ca2+ form and 113.7 ± 14.9 (± S.D.; n = 36) for the Ca2+-free
form of fibrillin. Measurements for the width, after subtraction of 1.8 nm for the platinum coating, were 2.0 ± 0.6 nm (± S.D.; n = 62) for the Ca2+ form and 3.0 ± 0.8 nm (± S.D.; n = 56) for the Ca2+-free
form. These data corresponded with the calculations based on the
sedimentation coefficients (Table I) and suggested that previous
measurements of fibrillin (7) were performed using molecules in the
Ca2+ form.
Since Ca2+ binds to cbEGF-like motifs, which are tandemly repeated in fibrillin-1, the structural changes observed are most likely to occur within these tandem repeats. To focus on structural changes of tandemly repeated cbEGF-like modules, we analyzed a recombinant fibrillin-1 subdomain (rF17), comprising the longest stretch of cbEGF-like motifs present in fibrillin-1 and the preceding 8-cysteine motif.
After dialysis against water including 2 mM
CaCl2 or 0.5 mM EDTA, rF17 was rotary shadowed
and visualized by transmission electron microscopy (Fig.
3). The representative fields clearly show that the
Ca2+ form of rF17 adopts an extended thread-like and often
straight shape (Fig. 3A), whereas the Ca2+-free
form of rF17 appears shorter, somewhat wider and less straight (Fig.
3B). Length measurements revealed 38.0 ± 2.7 nm (± S.D.; n = 100) for the Ca2+ form and
31.5 ± 3.3 nm (± S.D.; n = 110) for the
Ca2+-free form of rF17 (Table II). The
length distribution is somewhat wider for the Ca2+-free
form of rF17, indicating a more variable structure compared to the
Ca2+ form (Fig. 4). The width of rF17 after
subtraction of 1.8 nm for the platinum coating was 1.8 ± 0.5 nm
(± S.D.; n = 100) for the Ca2+ form and
2.7 ± 0.7 nm (± S.D.; n = 100) for the
Ca2+-free form.
|
The sedimentation coefficients for rF17 in the presence of
CaCl2 or EDTA were determined by analytical
ultracentrifugation (Fig. 5). Extrapolated to zero
protein concentration, we determined coefficients of
s20,w0 = 3.2 ± 0.1 (± S.D.) for the Ca2+ form and
s20,w0 = 3.9 ± 0.2 (± S.D.) for the Ca2+-free form of rF17
(Table II). No di- or multimerization of rF17 at high protein
concentrations up to 2.9 mg/ml was observed. As demonstrated
previously, about 10% of the total mass of rF17 originates from
N-linked oligosaccharides (15), and therefore the total mass
of rF17 was calculated to be 71 kDa. Using this molecular mass, the
shape of rF17 was calculated for different degrees of hydration (35)
(Table II). Upon removal of Ca2+ by EDTA the predicted
length of rF17 decreased by about 29-32% and the predicted width
increased by about 18-22%, similar to observations with authentic
fibrillin. Assuming 0.1 cm3/g degree of hydration and a
rod-shaped form, the length of rF17 is predicted to be 38.4 nm in the
Ca2+ form and 26.6 nm in the Ca2+-free form,
which is in agreement with measurements of rotary shadowed particles of
rF17 (Table II). Assuming the 8-cysteine motif is globular with a
diameter of 2-3 nm, the length contribution of a single cbEGF-like
motif to a tandem array is about 3.0 nm when Ca2+ is bound
and may decrease to as little as 2.0 nm upon removal of
Ca2+.
Far-UV circular dichroism spectra for rF17 in the presence and absence
of Ca2+ are shown in Fig. 6. The overall
appearance of the two spectra are different, suggesting electronic or
conformational changes as a result of coordination of Ca2+.
The Ca2+-free form showed a maximum at 185 nm ( =
0.45) and a minimum at 202 nm (
=
3.23). When Ca2+
ions bound to rF17, there was a decrease of
between 180 and 195 nm with a minimum at 191 nm (
=
3.0) and an increase of
between 196 and 216 nm with a maximum at 202 nm (
=
1.27). When
calculating the relative amounts of secondary structure by the variable
selection method (Table III), as expected, relatively low amounts of
-helix (8%), and high amounts of
-sheets
(25-26%) and
-turns (24-26%) were observed. Only small
differences in secondary structure between the Ca2+ form
and the Ca2+-free form were observed. These small
differences were between
-sheets (1-2%),
-turns (2%), and
"other" structural elements (4%). These data indicate that
secondary structural changes cannot account for the observed
differences in length and that secondary structural changes likely also
occur in interdomain regions between cbEGF-like repeats.
|
In this study, we demonstrated significant differences in the tertiary structure of fibrillin in the presence and absence of Ca2+. As determined by velocity sedimentation experiments and confirmed by measurements of rotary shadowed molecules, fibrillin shrinks in length by about 25% and becomes wider by about 13-17% when Ca2+ is removed from the molecule. Since Ca2+ binding to fibrillin is mediated by cbEGF-like modules, which are arranged in arrays of variable numbers of repeats, the structural changes observed must be associated with this type of motif or with its tandem arrangement. In order to substantiate these results, we analyzed a small (71 kDa) recombinant subdomain of fibrillin-1 (rF17), consisting primarily of cbEGF-like repeats (12) plus the preceding 8-cysteine motif, by analytical ultracentrifugation and rotary shadowing in the presence and absence of Ca2+. Based on the results presented here, we calculated that a single cbEGF-like repeat contributes about 3 nm in the Ca2+ form to the length of fibrillin. In the Ca2+-free form, this length may be decreased by 20-30%.
The best characterized interaction of Ca2+ with cbEGF-like
motifs has been demonstrated with single cbEGF-repeats from blood coagulation factors IX and X (23, 25). The Ca2+-binding
site has been described in an amino-terminal pocket with seven
liganding oxygen atoms in a pentagonal bipyramide geometry (23, 25).
The NH2-terminal amino acid residues are linked by a
Ca2+ ion to the -turn in the main
-sheet. Upon
removal of Ca2+, only locally restricted structural changes
occur at the NH2 terminus and at the
-turn on top of the
main
-sheet (23). We compared dimensions of Ca2+-loaded
and Ca2+-free forms of published structures of human factor
IX (25, 39) and bovine factor X (23, 40) using Insight II (version 2.3.0; Biosym Technologies). Two solution structures of factor X
determined by nuclear magnetic resonance were compared and the solution
structure of the Ca2+-free form of factor IX (39) was
compared with the crystal structure of the Ca2+ form (25).
Depending on the residues chosen for measurements, in factor X the
length of the cbEGF-like repeat is maximally ~6% shorter in the
Ca2+-free form. In factor IX the repeat is maximally
~11% shorter in the Ca2+-free form. Thus, structural
changes within single cbEGF-like repeats cannot account solely for the
shortening effect of tandemly repeated cbEGF-like motifs observed in
this study. Other changes, most likely in the connecting region between
adjacent repeats, must occur. This suggestion was substantiated by
circular dichroism measurements of rF17 in the presence and absence of
Ca2+ which demonstrated only minor changes in secondary
structure.
Based on an asymmetric crystallization of two units of a cbEGF-like repeat from human factor IX, which are not covalently linked to each other, it was suggested that tandemly arranged cbEGF-like repeats adopt a tightly wound helical arrangement stabilized by Ca2+ (25). In such a model, a single cbEGF-like repeat would only contribute about 1.5 nm to the length of a tandem array. The results presented in our study contradict this model.
Recently, the solution structure of a pair of cbEGF-like repeats of fibrillin-1 (numbers 32 and 33) in the Ca2+ form was reported, suggesting that Ca2+ rigidifies the interdomain region between the two cbEGF-like repeats (24). The two repeats were in a nearly straight orientation along the long axes with only a slight tilt angle of about 18°. A model for repeat numbers 32-36 predicted a length of 14.5 nm, suggesting that a single cbEGF-like repeat contributes about 2.9 nm to the length of the tandem array (24). These suggestions are in good agreement with the results presented in this study.
No structural data are available for the Ca2+-free form of
tandemly repeated cbEGF-like motifs. Condensation upon removal of Ca2+, as observed for fibrillin-1 and recombinant rF17,
could be explained by either more flexibility in the interdomain region
or, alternatively, by adopting a certain fixed tilt angle between
adjacent cbEGF-like repeats. Some evidence for more flexibility in the
interdomain region comes from single repeats of clotting factors. In
the Ca2+-free form, the amino termini were reported to be
poorly defined (39-41), which was attributed to physical mobility
within this region (40, 41). Since Ca2+ connects the
NH2 terminus (linking region between to adjacent repeats)
to the major -sheet of the repeats, it is conceivable that upon
removal of Ca2+, the linking domain becomes more flexible.
If the interdomain region is indeed more flexible in the absence of
Ca2+ one would expect wider length distributions of tandem
arrays in the Ca2+-free form compared to the
Ca2+-form with the "locked" extended conformation.
Length distributions of rotary shadowed rF17 in the presence and
absence of Ca2+ in fact showed a somewhat wider
distribution for the Ca2+-free form. However, in
sedimentation experiments of authentic fibrillin-1, approximately the
same band width was observed for the Ca2+-loaded and the
Ca2+-free form.
To reduce the apparent length of a single motif within a tandem array from about 3 nm in the extended form to as little as 2 nm, a tilt angle of about 96° would be required between adjacent repeats. Interestingly, x-ray diffraction data from porcine factor IXa, which contains two tandemly repeated EGF-like repeats (one NH2-terminal Ca2+-binding and one COOH-terminal non-calcium binding repeat)4 demonstrated a tilt angle of 110° between the repeats (42). In clotting factors, the interdomain region between the two EGF-like repeats cannot be stabilized by Ca2+ since the COOH-terminal repeat is not of the Ca2+-binding type and the Ca2+-binding site of the NH2-terminal repeat is distal from the interdomain region.
EGF-like motifs occur in at least 70 different proteins with a wide variety of biological functions. They often occur in tandemly repeated arrays (19). These repeated arrays can consist of non-calcium binding EGF-like repeats as found for example in the tenascin family (43, 44) or of cbEGF-like repeats as in Drosophila Notch (27), the fibulins (32, 45, 46), or the fibrillins (2-6). Why was Ca2+ binding to EGF-like motifs invented? Tandem arrays of non-calcium binding EGF-like repeats may serve as flexible hinge-like domains in modular proteins, whereas in arrays of cbEGF-like motifs, Ca2+ binding will stabilize an extended and rigid conformation that might be necessary to maintain the correct tertiary structure for certain protein-protein interactions or for assembly of supramolecular structures. We reported previously that cbEGF-like repeats in fibrillin-1 are protected against proteolysis by Ca2+ (15). Tandem stretches of cbEGF-like repeats may provide this protection to proteins which are supposed to form stable structures.
More than 50 mutations in fibrillin-1 give rise to the Marfan syndrome, a connective tissue disorder with skeletal, cardiovascular, and ocular symptoms (17). The majority of these mutations are predicted to disturb Ca2+ binding to cbEGF-like motifs. Structural changes upon removal of Ca2+, as demonstrated in this study, may resemble local structural changes in mutated fibrillin-1. The interdomain region NH2-terminal to the cbEGF-like repeat harboring the mutation may be more flexible or adopt a certain angle different from the Ca2+-loaded form. These structural changes might then disturb protein interactions or supramolecular assembly processes, or expose certain regions to proteolytic degradation. These types of interference with biological functions of fibrillin might in turn lead to the progressive nature of the disorder.
We are thankful for the excellent technical assistance of Robert N. Ono, Catherine C. Ridgway, Noé L. Charbonneau, and Jay E. Gambee.