(Received for publication, May 21, 1996, and in revised form, October 3, 1996)
From the Department of Medical Biochemistry,
University of Aarhus, DK-8000 Aarhus C, Denmark, ¶ Department of
Cellular Biology, University of Barcelona, 08028 Barcelona, Spain, and
Department of Medical Genetics, University of British Columbia,
British Columbia, Canada V6T 1Z4
Lipoprotein lipase (LpL) can mediate cellular
uptake of chylomicron and VLDL remnants via binding to heparan
sulfate proteoglycans (HSPG) and the endocytic
2-macroglobulin receptor/low density lipoprotein
receptor-related protein (
2MR/LRP). Whereas it is established that the C-terminal folding domain binds to
2MR/LRP, it remains uncertain whether it binds to
heparin and to HSPG. To identify segments important for binding to
2MR/LRP and to clarify possible binding to heparin, we
produced constructs of the human C-terminal folding domain,
LpL-(313-448), and of the fragment LpL-(347-448) in Escherichia
coli. In addition to binding to
2MR/LRP,
LpL-(313-448) displayed binding to heparin with an affinity similar to
that of the LpL monomer, whereas it bound poorly to lipoprotein
particles. Moreover, LpL-(313-448) displayed heparin sensitive binding
to normal, but not to HSPG deficient Chinese hamster ovary cells.
LpL-(313-448) and LpL-(347-448) showed similar affinities for binding
to both purified
2MR/LRP and to heparin. Deletion of LpL
residues 380-384 abolished the binding to LRP, whereas binding to
heparin was unperturbed. The binding to both heparin and
2MR/LRP was essentially abolished following deletion of
residues 404-430, and pretreatment of CHO cells with the peptide
comprising aa 402-423 inhibited the binding of LpL-(313-448). We
conclude that the C-terminal folding domain of human LpL has a site for
binding to heparin and to HSPG, presumably involving amino acids within
residues 404-430. Two segments of the domain are necessary for
efficient binding to
2MR/LRP, one comprising residues
380-384 and another overlapping the segment important for binding to
heparin.
Chylomicron and very low density lipoprotein
(VLDL)1 remnants can be taken up via the
endocytic 2-macroglobulin receptor/LDL receptor-related
protein (
2MR/LRP) either mediated by apolipoprotein E,
or by lipoprotein lipase (LpL) or hepatic lipase associated with the
lipoproteins (1-4). Recent results provide evidence that the uptake
via
2MR/LRP occurs in vivo, since remnant
lipoproteins accumulate in mice that express reduced amounts of
2MR/LRP and also lack functional LDL receptors (5). It
has been suggested that the LpL-mediated pathway may be particularly
important in the vascular wall, since
2MR/LRP is
abundant in macrophages and smooth muscle cells of atherosclerotic
lesions and since LpL is secreted by macrophages in the lesions (6). In
addition, it was recently shown that, whereas normal macrophages
incubated with VLDL accumulate massive amounts of lipids, this does not occur in LpL-deficient macrophages (7).
LpL binds to cell surface heparan sulfate proteoglycans (HSPG) as
reflected in its strong affinity for binding to heparin. This property
is physiologically important for the docking on endothelial cells
exposing the lipase to circulating lipoproteins. LpL also binds to HSPG
of other cell types as recently shown directly by immunofluorescence
studies (8), and studies in Chinese hamster ovary cells have shown that
uptake and degradation of LpL can be mediated by HSPG (9). In addition,
the docking of LpL on HSPG of cells rich in 2MR/LRP is
important for uptake of LpL and for LpL-mediated uptake of lipoproteins
via this receptor (4, 10, 11).
LpL is a member of the mammalian lipase family also comprising the
homologous hepatic and pancreatic lipases. The crystallographic structure of pancreatic lipase (12) shows that it consists of two
folding domains, a larger N-terminal and a smaller C-terminal domain.
Based upon the similarities in sequence it is thought that the domain
organization is similar for LpL, and a three-dimensional model was
recently proposed (13). LpL circulates both as a 96-kDa homodimer,
which is the normally secreted and catalytically active form, and as a
catalytically inactive monomer (14-16). Although not known, it is
likely that the LpL dimer is arranged head-to-tail in a way that allows
enough space for conformational changes following substrate binding
(13). The dimeric structure of LpL causes an increase in the affinity
for heparin, presumably because sites in both monomers can participate
in binding of one heparin molecule (13), and helps efficient
2MR/LRP-mediated lipoprotein uptake, possibly because
only the dimer can bind to a lipoprotein particle and to the receptor
at the same time (10, 17).
The N-terminal folding domain of human LpL, comprising residues 1-312,
harbors the catalytic triad with its covering loop, which is important
for interaction with lipid substrates (18), sites for binding to
heparin (16), and the site for binding of the LpL cofactor
apolipoprotein C-II (2). The C-terminal folding domain of the human
lipase, comprising residues 313-448, is important for binding to
2MR/LRP and also participates in binding to lipoprotein
particles and to heparin. Thus, it is known that sequences within aa
378-423 largely account for the affinity to
2MR/LRP
(11). The presence of a secondary lipid binding site in the C-terminal
folding domain is strongly suggested by the finding that truncation of
LpL at residue 390, which does not disrupt the dimeric structure,
abolishes binding to particles such as chylomicrons (19) and rabbit
-VLDL (10). In addition, LpL-(313-448) has been reported to mediate
binding of lipoproteins to
2MR/LRP (20). Mutational
studies have suggested that the secondary lipid binding site includes
the tryptophan residues 390 and 393 (17, 20). Previous results
concerning heparin binding affinity of the C-terminal domain have been
ambiguous. Studies including LpL-hepatic lipase chimeras strongly
suggest that the C-terminal folding domain has a site for binding of
heparin (21, 22). This is in accordance with the result that heparin abolishes binding of a C-terminal LpL fragment (residues 378-423) to
2MR/LRP, which itself does not bind heparin. Earlier
studies of mutated LpL species have shown that candidate segments
comprising residues 390-393 and 439-448 appear not to contribute to
the heparin affinity (16). Finally, the result that dimeric
LpL-(1-390) and LpL-(1-448) are eluted at the same salt
concentrations following heparin affinity chromatography (19) may be
taken as an argument against a separate functional heparin binding site
in the C-terminal folding domain.
The purpose of the present work was to identify segments important for the receptor recognition of the C-terminal folding domain, to clarify whether the domain binds to lipoproteins, and to characterize possible binding to heparin and to HSPG. We report that the C-terminal folding domain binds poorly to lipoproteins, whereas the domain does bind to heparin and to cellular HSPG. Moreover, we provide evidence that two segments are important for binding to the receptor, one of them overlapping with the heparin binding domain.
Human 2MR/LRP was prepared from
solubilized placental membranes by affinity chromatography using
immobilized activated
2M. Elution was in 150 mM NaCl, 5 mM EDTA, 10 mM sodium
phosphate, 0.1% Triton X-100, pH 6.0. Copurified RAP was removed from
the receptor by incubation with heparin-Sepharose, pH 8.0 (23). Purified
2MR/LRP was iodinated to a specific activity of
about 6 × 106 Bq/mol as described (10, 24). Bovine
LpL (bLpL), prepared as described previously (25), was a gift from Dr.
G. Olivecrona, Department of Medical Biochemistry and Biophysics,
University of Umeå, Sweden. This preparation contains enzymatically
active (~600 units/mg) dimeric enzyme (25). Rabbit
-VLDL, isolated from the blood of cholesterol-fed rabbits, was a gift from Dr. U. Beisiegel, Medizinische Klinik, Universitätskrankenhaus
Eppendorf, Hamburg, Germany. The composition of this preparation, and
the 125I-iodination, have been described previously (11,
26).
The
construct H6FX-LpL-(347-448), containing the
hexahistidine-factor X substrate sequence
MGSH6SIEGR and aa 347-448 of human LpL, was prepared as
described by Nielsen et al. (24). This and other constructs
containing human LpL sequences are referred to as LpL constructs.
H6FX-LpL-(313-448) was generated in the same
way, except that the primer
5-CAGGATCCATCGAGGGTAGGGTCTTCCATTACCAAGTA-3
was used for the
N-terminal. The mutated H6FX-LpL-(347-448)
constructs were generated using full-length human LpL cDNAs
containing different point mutations and deletions generated as
described previously (16, 17). All LpL fragment constructs were
cDNA-sequenced to eliminate possible PCR errors. The recombinant
C-terminal folding domain construct,
H6FX-LpL-(313-448), and the recombinant
wild-type and mutated LpL fragment constructs were expressed in
Escherichia coli DH1 cells and purified on a
Ni2+ nitriloacetic acid column (11, 24). The recombinant
proteins were more than 95% pure, as judged from Coomassie Brilliant
Blue-stained SDS-polyacrylamide gels, and were generally soluble in
concentrations up to 20-50 µM in aqueous buffer. The
H6FX-LpL-(313-448) construct, referred to as
LpL-(313-448), was 125I-iodinated to a specific activity
of about 4 × 106 Bq/mol using chloramine T as the
oxidizing agent (11, 24).
Unless otherwise stated, heparin was from Leo Pharmaceuticals, Denmark. Tinzaparin, a low molecular mass heparin, and [3H]tinzaparin with a specific activity of about 8 mCi/mg (27), were kind gifts from Dr. P. B. Østergaard, Novo Nordisk A/S, Denmark. Heparin-Sepharose (Cl6B) was from Pharmacia, Sweden.
Solid Phase Assays2MR/LRP was immobilized
in Maxisorp, and bLpL and recombinant LpL fragments were immobilized in
Polysorp microtiter wells (both from NUNC, Denmark) by incubation for
2 h at 20 °C in 50 mM NaHCO3, pH 9.6, as described previously (10, 11, 24). Separate experiments showed that
the bLpL was catalytically inactive after immobilization to the
wells,2 and the immobilized lipase is
therefore thought to be in monomeric form. After washing, the wells
were blocked in 5% bovine serum albumin when using
125I-
2MR/LRP or
125I-LpL-(313-448), and with 2% Tween 20 when using
[3H]tinzaparin. All incubations in microtiter wells were
performed for 16 h at 4 °C in 100 µl of 140 mM
NaCl, 10 mM Hepes, 2 mM CaCl2, 1%
bovine serum albumin, pH 7.8. Following washing with 2 × 200 µl
of incubation buffer, bound radioactivity was removed with 10% SDS and
counted. Competition-inhibition curves were drawn as the best least
square fit to triplicate determinations using the equation
b/f = R/(Ki + F), where b/f is the ratio of bound and free
labeled ligand, R is the concentration of binding sites
immobilized in the well, and F is the concentration of free competitor.
COS-1 cells (American Type
Culture Collection CRL 1650) were incubated in monolayers (about 3 × 105 cells/ml) as described previously (24). After
incubation for 4 h at 37 °C, degradation of
125I--VLDL was assessed by measuring radioactivity in
the medium soluble in 12% trichloroacetic acid. After washing, the
cells were lysed in 1 M NaOH and assayed for
cell-associated radioactivity. Normal human skin fibroblasts were
obtained as described previously (8) and cultured in Dulbecco's
modified Eagle's medium with 10% fetal calf serum containing
glutamine (2 mM), penicillin, and streptomycin. Wild-type
and mutant (CHO-745) Chinese hamster ovary cells lacking heparan
sulfate and chrondroitin sulfate glycosaminoglycans (28) were kindly
provided by Dr. J. D. Esko, Department of Biochemistry, University of
Alabama at Birmingham. The CHO cells were grown in Ham's F-12 culture
medium supplemented with 10% fetal calf serum, glutamine, and
antibiotics. Cells were plated on glass coverslips for
immunofluorescence experiments and used between the 3rd and 4th day of
seeding.
The binding of bLpL (2.5 µg/ml) or LpL(313-448) (25 µg/ml) to fibroblasts and CHO cells was
performed by incubating prechilled cells for 30 min at 4 °C in
Dulbecco' modified Eagle's medium containing 20 mM Hepes
(pH 7.4) and 1% defatted bovine serum albumin. The cells were then
fixed for 15 min at room temperature in 3% paraformaldehyde, 2%
sucrose, 0.1 mM phosphate buffer, pH 7.4. After rinsing
twice in phosphate-buffered saline (PBS, 10 mM phosphate, 150 mM NaCl, pH 7.4) containing 20 mM glycine,
the fixed cells were blocked with PBS-glycine containing 1% bovine
serum albumin. As a primary antibody, we used the monoclonal antibody
5D2 (Oncogene) at dilution 1:100. This antibody is reported to react
within residues 396-405 of the C-terminal folding domain (11, 29), and
separate analysis showed that it did not react with the peptide
comprising aa 402-423 used in inhibition experiments. A fluorescein
isothiocyanate-conjugated rabbit anti-mouse IgG (F(ab)2
fragment, DAKO, Denmark) at dilution 1:50 was used to visualize the
primary antibody. Both antibodies were diluted in PBS-glycine, 1%
bovine serum albumin. The primary antibody was applied for 45 min at
37 °C and, following washing in PBS-glycine for 15 min, the
secondary antibody was applied for 45 min. The coverslips were then
labeled with the nuclear stain Hoechst 33342, mounted upside-down on a
glass slide with a drop of Immunofluore (ICN Biomedicals) and viewed on
a Zeiss Axioskop equipped for epifluorescence. For confocal microscopy studies we used a Leica TCS 4D confocal scanning laser microscope adapted to an inverted Leitz DMIRBE microscope and a 63× (numerical aperture 1.4, oil) Leitz Plan-Apo objective. Fluorescein isothiocyanate was excited at 488-nm lines from a 75-mW krypton-argon laser. Each
image was the average of four serial scans at the confocal microscope
normal scan rate (30). Image sizes were 512 × 512 and were
photographed on a color high resolution video printer. The results are
representative for at least five separate experiments.
We
first examined whether the the C-terminal folding domain of the human
lipase, LpL-(313-448), has an affinity toward 2MR/LRP similar to that of the fragment, LpL-(347-448). Fig. 1
shows binding of 125I-labeled folding domain to immobilized
purified
2MR/LRP. The binding was completely inhibited
by unlabeled LpL-(313-448) or LpL-(347-448) at high concentrations.
The potency of the fragment toward inhibition of the binding of
125I-LpL-(313-448) was similar to or slightly lower than
that of the folding domain. Other experiments (not shown) demonstrated complete inhibition of the binding of tracer in the presence of heparin
(10 units/ml).
In the reverse experiment, 125I-labeled
2MR/LRP was bound to immobilized bovine LpL, and the
inhibitory potencies of LpL-(313-448) and LpL-(347-448) were again
found to be similar and inhibitable by heparin (not shown). We decided
to use the LpL-(347-448) fragment as the basis for mutational analysis
since the yields of the recombinant products were satisfactory. The
inhibitory potencies toward binding of
125I-
2MR/LRP to immobilized bLpL were
measured as illustrated in Fig. 2. The results
summarized in Table I show that deletion of aa 380-384
caused a 10-20-fold reduction in the inhibitory potency. Deletion of
aa 390-393, or exchange with the equivalent aa in human hepatic
lipase, had no effect. Deletion of a large segment comprising aa
404-430 also caused a marked reduction in the inhibitory potency,
whereas the deletion of aa 404-414 caused only an approximately 4-fold
reduction. The displayed point mutations had little or no effects.
Previous results have shown that LpL-(378-448) and LpL-(378-423) have
the same affinity to
2MR/LRP (11). When taken together,
the results therefore indicate that two segments rich in basic residues
are important for the interaction with receptor: aa 380-384 (LKWKS)
and aa 404-423 (IRVKAGETQKKVIFCSREKV).
|
It has been reported that LpL-(313-448)
(20), in contrast to the fragment LpL-(378-448) (11) can mediate the
the binding of lipoproteins to 2MR/LRP. We therefore
compared the abilities of the folding domain, LpL-(313-448), and the
fragment, LpL-(347-448), to bind 125I-labeled rabbit
-VLDL. Fig. 3A shows that LpL-(347-448)
does not bind
-VLDL. The binding of
-VLDL to LpL-(313-448) was
confirmed, although this binding is poor compared with the binding to
bLpL. Fig. 3B shows that LpL-(313-448) can mediate binding
of
-VLDL to
2MR/LRP, although poorly, and that the
concentrations of LpL-(313-448) required for the mediation of
lipoprotein binding are more than 100-fold higher than those needed for
bLpL. Finally, Fig. 4 shows that, whereas 100 nM bLpL clearly mediates degradation of
125I-
-VLDL in COS-1 cells, which express abundant
2MR/LRP (24), 1 µM of the C-terminal
folding domain is incapable of mediating degradation of the labeled
lipoprotein. We therefore conclude that the C-terminal folding domain
cannot on its own mediate efficient uptake of lipoproteins.
Binding to Heparin
We next turned to the question whether the
C-terminal folding domain has a site for heparin binding. Fig.
5 shows that LpL-(313-448) was retained on
heparin-Sepharose and eluted at 0.8 M NaCl. The position of
elution was similar to that of the bLpL monomer, whereas dimeric bLpL
eluted at 1.3 M NaCl. This result suggests that a segment
in the C-terminal folding domain may account for a large part of the
heparin affinity of the LpL monomer. We used binding of tinzaparin, a
low molecular mass heparin (average Mr ~4,500) for further analysis of the putative heparin binding segment. Fig.
6 shows an experiment in which bLpL, LpL-(313-448), or
LpL-(347-448) immobilized in microtiter wells were incubated with
[3H]tinzaparin and increasing concentrations of unlabeled
tinzaparin. The results show that tinzaparin bound with approximately
equal affinities to LpL-(313-448) and LpL-(347-448), and that the
affinity was only moderately lower than that for binding to bLpL that
was catalytically inactive after immobilization2 and
therefore thought to be in monomeric form. Similar experiments with
immobilized 2MR/LRP showed no binding of heparin (data
not shown) confirming that heparin inhibits binding of the C-terminal folding domain due to interaction with the ligand and not with the
receptor. Table II summarizes the results concerning
binding of tinzaparin to LpL and wild-type and mutated LpL fragments. It appears that the deletion of aa 380-384, which essentially abolished the binding to
2MR/LRP (Table I), had no
effect on the binding of tinzaparin. Deletion of the large segment
comprising aa 404-430 completely abolished the binding of labeled
tinzaparin, whereas deletion of the aa 404-414 segment had only a
minor effect.
|
The results show that heparin can bind directly to the C-terminal
folding domain and that residues within the segment comprising aa
404-430 are essential. Since heparin inhibits the binding of LpL-(378-423) to 2MR/LRP (11), it seemed possible that
aa 424-430 are not essential. This prompted us to prepare the peptide
comprising aa 402-423 and analyze for possible binding to
heparin-Sepharose. The peptide was retained on the heparin column used
for chromatography of LpL-(313-448) and eluted at about 0.3 M NaCl, whereas the irrelevant basic peptide
urokinase-(148-159) (AGQKTLRPRFK) was not retained (not shown).
The
binding to normal human fibroblasts and to wild-type as well as heparan
and condroitin sulfate glycosaminoglycan-deficient CHO cells was
investigated in order to evaluate the possible physiological significance of the heparin binding. Fig. 7, row
1, shows that LpL-(313-448) was able to bind to fibroblasts. As
for dimeric bLpL, the binding of the C-terminal folding domain was
mainly surface-associated. In addition, scattered bright fluorescence spots were observed when using LpL-(313-448), indicating some nonspecific binding. Heparin (50 units/ml) inhibited the binding to the
fibroblasts (not shown). Wild-type CHO cells (row 2) also bound LpL-(313-448) as well as bLpL, and the binding was inhibited by
heparin (row 3). Moreover, binding was inhibited by the
peptide comprising aa 402-423 (row 4), but not by the
irrelevant peptide (not shown), Finally, the binding of both bLpL and
the folding domain was greatly reduced when using the CHO-745 cells
deficient in HSPG (row 5). It should be noted that the
apparent nonspecific binding persisted in the presence of heparin and
in incubations of the CHO-745 cells. When taken together, the results
indicate that the C-terminal folding domain binds to proteoglycans and might participate in the docking of LpL to cell surfaces.
The present results help defining the segments involved in the
functions of the C-terminal folding domain of LpL. Fig.
8 shows a model of the folding domain (13) based on the
known three-dimensional structure of pancreatic lipase (12) and
highlights the segments important for binding to receptor and to
heparin. We suggest that residues within the segment comprising aa
380-384 are important for recognition of 2MR/LRP and
that residues within aa 404-430 are important for binding to receptor
as well as to heparin. This interpretation presumes that the functional
changes of the C-terminal folding domain are the direct result of the
deletion mutations. However, it cannot be excluded that some functional
changes may be due to conformational changes of the domain induced by
the mutations. This applies particularly to the large deletion of aa
404-430.
Binding to
The most notable result is
perhaps the dramatic decrease in affinity following deletion of aa
380-384 (LKWKS). The findings that the affinity for binding of heparin
remains unperturbed, and that other deletions, e.g. of aa
390-393, have little or no effect on the receptor binding affinity,
argue that residues in this region interact with the receptor. Since
basic residues of other ligands have been shown to be important for
binding to the receptor (31), we propose that Lys-381, Lys-383, and
possibly Lys-379 participate in the binding. Surprisingly, deletion of aa 404-414, including Lys-407 which is thought to be critical for
binding to 2MR/LRP (20), caused only a moderate
reduction in the affinity. On the other hand, deletion of the large
segment comprising aa 404-430 caused a marked decrease in the affinity for the receptor. Since previous results have shown the same receptor binding affinity of the fragments comprising aa 378-448 and 378-423 (11), it seems likely that residues within the segment aa 404-423 participate in the binding. These may include Arg-405, Lys-407, Lys-413, Lys-414, and particularly, as judged from the minor effect of
deletion of aa 404-414, Arg-420 and Lys-422. Since neither deletion of
aa 380-384 nor of aa 404-430 completely abolished the affinity to the
receptor, we propose that two segments located on either side of the
C-terminal folding domain interact with
2MR/LRP to
provide high affinity binding.
The exact location of sites on the receptor for binding of LpL is not
known. However, we have previously shown that wild-type LpL-(347-448)
binds to the same sites on the receptor as the C-terminal fragment of
RAP (24) which primarily binds in the region of 2MR/LRP
containing a cluster of eight LDL receptor class A repeats (32, 33).
Recently, the three-dimensional structure of class A repeats has shown
a clustering of many of the side chains of acidic residues, including
the conserved Ser-Asp-Glu triad (34), thus emphasizing the importance
of exposed basic residues in ligands. It is also of note that LpL, like
RAP, binds to two other members of the LDL receptor family, megalin
(35) and the VLDL receptor (36), with about the same affinity as to
2MR/LRP, and it may be inferred that the same segments
of LpL participate in binding to class A repeats of these
receptors.
Previous studies showed that the
receptor binding fragment LpL-(378-448) does not bind to -VLDL and
therefore inhibits the LpL-mediated uptake of lipoproteins via
2MR/LRP. On the other hand, the C-terminal folding
domain was reported to bind to lipoproteins and to mediate their uptake
into cells (37). We find that although LpL-(313-448) does bind to
-VLDL, the concentrations needed are several magnitudes higher than
for LpL. This also applies to the mediation of
-VLDL binding to
2MR/LRP. We therefore conclude that the secondary lipid
binding site in the C-terminal folding domain is insufficient on its
own, even though it is important for binding of full-length LpL to
lipoprotein particles (10, 19). Although not proven, it is likely that
Trp-393 and Trp-394 participate in binding to lipoproteins, since
substitution of both residues with alanine abolishes binding of
LpL-(313-448) to lipoproteins (20), and since deletion of aa 390-393
reduces the ability of dimeric LpL to promote lipoprotein uptake by
about 40% (17). The position of the putative secondary lipid binding site adjacent to residues 380-384, which are necessary for binding to
receptor, may explain why the dimeric structure of LpL appears important for binding to both lipoproteins and to
2MR/LRP and thereby for efficient mediation of
lipoprotein binding and uptake (10, 11).
The present results show for
the first time that the C-terminal folding domain binds to heparin.
Previous studies have examined two candidate regions in this domain by
using mutational analysis of dimeric LpL. One region is the cluster of
positive charges in the extreme C-terminal region (13, 38). Ma et
al. (16) found that replacement of the 10 C-terminal aa with the
equivalent residues in pancreatic lipase did not disrupt the affinity
for heparin, in agreement with our previous result that heparin at low
concentration inhibited binding of both LpL-(378-448) and LpL-(378-423) to 2MR/LRP (11). The other region is the
noncharged sequence aa 390-393 (WSDW) homologous to the heparin
binding type-1 repeat of thrombospondin, and, as in our experiments
with the C-terminal folding domain, deletion or exchange with the
equivalent sequence in hepatic lipase had little influence on the
heparin affinity of LpL (16). A third candidate region is the cluster of basic residues originating from a noncontiguous peptide stretch and
comprising Lys-319-Lys-403-Arg-405-Lys-407-Lys-413-Lys-414 (13).
In addition, contacts may be made to glutamine residues (e.g. aa 402 and 412) as shown for basic fibroblast growth
factor (39). It appears that Lys-319 is not important since
LpL-(313-448) and LpL-(347-448) showed similar affinities.
Surprisingly, the heparin affinity was only moderately affected
following deletion of aa 404-414, whereas deletion of aa 404-430
abolished the binding to heparin. This may suggest that Arg-420- and
Lys-422 are important as supported by the heparin affinity, although
weak, of the peptide LpL-(402-423). In addition, the ability of this
peptide to inhibit binding of bLpL and LpL-(313-448) to wild-type CHO
cells is presumably due to a blocking of the sites on HSPG.
The finding that the C-terminal folding domain binds to heparin with
about the same affinity as the LpL monomer suggests that it is an
important determinant for binding to HSPG and therefore presumably for
catabolism of the monomer. The considerably higher affinity of the
dimer is thought to be due to interaction of one heparin molecule
(e.g. dodecamer or longer) with the two LpL monomers at the
same time. The heparin molecule may span the three positively charged
clusters in the N-terminal domains and even occupy positively charged
residues of the C-terminal domains (13). It remains unknown whether the
site for binding of heparin in the C-terminal domain may be shielded in
the dimer and therefore contribute only to a limited extent to the
heparin affinity of dimeric LpL. In conclusion, we have shown that the
C-terminal folding domain has affinity for heparin and for HSPG, and we
have identified segments important for the interactions with
2MR/LRP and with heparin.
We thank G. Olivecrona for bLpL, P. B. Østergaard for tinzaparin, and J. D. Esko for cell lines. Helle Nielsen, David García, and Susanne Castel are thanked for excellent technical assistance.