Results Specificity of PCSK9 iodinated by the tyramine cellobiose TC method [125I]TC-PCSK9 binding to HepG2 cells Incubation of cells in lipoprotein-free medium increa-sed the level of L
Trang 1subtilisin ⁄ kexin type 9 binding to low-density lipoprotein receptors on HepG2 cells
Seyed A Mousavi1, Knut E Berge1, Trond Berg2and Trond P Leren1
1 Unit for Cardiac and Cardiovascular Genetics, Department of Medical Genetics, Oslo University Hospital Rikshospitalet, Norway
2 Department of Molecular Biosciences, University of Oslo, Norway
Keywords
association; dissociation; dissociation
constants; low-density lipoprotein receptor;
proprotein convertase subtilisin ⁄ kexin 9
Correspondence
T P Leren, Unit for Cardiac and
Cardiovascular Genetics, Department of
Medical Genetics, Oslo University Hospital
Rikshospitalet, P.O Box 4950 Nydalen,
NO-0424 Oslo, Norway
Fax: +47 23075561
Tel: +47 23075552
E-mail: trond.leren@rikshospitalet.no
(Received 29 March 2011, revised 7 June
2011, accepted 16 June 2011)
doi:10.1111/j.1742-4658.2011.08219.x
Proprotein convertase subtilisin⁄ kexin type 9 (PCSK9) is a secreted protein that regulates the number of cell surface low-density lipoprotein receptors (LDLRs) and the levels of low-density lipoprotein cholesterol in plasma Intact cells have not previously been used to determine the characteristics
of binding of PCSK9 to LDLR Using PCSK9 iodinated by the tyramine cellobiose (TC) method ([125I]TC-PCSK9), we measured the affinity and kinetics of binding of PCSK9 to LDLR on HepG2 cells at 4C The extent
of [125I]TC-PCSK9 binding increased as cell surface LDLR density increased Unlabeled wild-type and two gain-of-function mutants of PCSK9 reduced binding of [125I]TC-PCSK9 The Scatchard plot of the binding-inhibition curve was curvilinear, indicative of high-affinity and low-affinity sites for PCSK9 binding on HepG2 cells Nonlinear regression analysis of the binding data also indicated that a two-site model better fit-ted the data The time course of [125I]TC-PCSK9 binding showed two phases in the association kinetics Dissociation of [125I]TC-PCSK9 also occurred in two phases Unlabeled PCSK9 accelerated the dissociation of [125I]TC-PCSK9 At low pH, only one phase of dissociation was apparent Furthermore, the dissociation of [125I]TC-PCSK9 under pre-equilibrium conditions was faster than under equilibrium conditions Overall, the data suggest that PCSK9 binding to cell surface LDLR cannot be described by
a simple bimolecular reaction Possible interpretations that can account for these observations are discussed
Introduction
Proprotein convertase subtilisin⁄ kexin type 9 (PCSK9)
is a protein secreted by the liver that was recognized as
an important regulator of cholesterol homeostasis
through its link to autosomal dominant
hypercholester-olemia [1–3] Central to its role as a
cholesterol-regula-tory protein is the ability of PCSK9 to downregulate the
low-density lipoprotein (LDL) receptor (LDLR) [4–6]
Hepatic LDLR seems to be particularly susceptible to
this effect of PCSK9 PCSK9-mediated downregulation
of hepatic LDLR inhibits LDL uptake from plasma, thus increasing the concentrations of LDL cholesterol
in plasma [4,7] Besides the effect on hepatic LDLR levels, endocytosis of PCSK9 by the liver is also responsible for clearance of PCSK9 from the circulation [7]
The importance of PCSK9 in maintaining choles-terol homeostasis is clinically evident, in that PCSK9 gain-of-function mutations are associated with elevated
Abbreviations
ECD, extracellular domain; EGF-A, epidermal growth factor-like repeat A; LDL, low-density lipoprotein; LDLR, low-density lipoprotein receptor; LPDS, lipoprotein-depleted serum; PCSK9, proprotein convertase subtilisin ⁄ kexin type 9; PCSK9-WT, wild-type proprotein
convertase subtilisin ⁄ kexin type 9; TC, tyramine cellobiose.
Trang 2levels of plasma LDL cholesterol, whereas
loss-of-func-tion mutaloss-of-func-tions are associated with low levels of plasma
LDL cholesterol [1,8–12] Further data supporting the
importance of PCSK9 in cholesterol homeostasis come
from studies in mice, demonstrating that
overexpres-sion of PCSK9 in the liver results in increased plasma
LDL cholesterol levels, whereas knockout of the
PCSK9 gene decreases plasma levels of LDL
choles-terol [4–6,13] (for a recent review see [14])
PCSK9 exerts its action by interacting with a
bind-ing site within the epidermal growth factor-like
repeat A (EGF-A) of LDLR [15] The EGF-A-binding
site of PCSK9 has been shown to be composed of
resi-dues within the catalytic domain [16,17] Recent data
also indicate the potential involvement of other sites in
PCSK9 in LDLR binding [18]
Many of the data on parameters that describe
PCSK9 binding to LDLR have been obtained from
Biacore surface plasmon reasonance studies, using a
purified extracellular domain (ECD) of the LDLR
However, although sensitive and powerful, this
method may not completely reflect the situation of
membrane-embedded, full-length LDLR in intact cells
In this report, we present studies on the interaction of
PCSK9 with LDLR on HepG2 cells, a cell line that is
widely used to study PCSK9-mediated degradation of
LDLR
Results
Specificity of PCSK9 iodinated by the tyramine
cellobiose (TC) method ([125I]TC-PCSK9) binding
to HepG2 cells
Incubation of cells in lipoprotein-free medium
increa-sed the level of LDLR expresincrea-sed on the cell surface
1.8-fold (1.82 ± 0.10, n = 3) as compared with cells
that had been grown for 48 h in complete growth
med-ium (Fig 1A) Under both conditions, the amount of
[125I]TC-PCSK9 specifically bound was linearly related
to cell density Incubation of cells in lipoprotein-free
medium was also associated with a 1.6-fold increase
(1.64 ± 0.16, n = 3) in the binding of [125I]TC-PCSK9
(Fig 1B) The binding of [125I]TC-PCSK9-D374Y is
presented for comparison Approximately five times
less [125I]TC-PCSK9-D374Y than [125I]TC-PCSK9 was
needed to achieve equivalent binding (1.7 ± 0.07,
n= 3) (Fig 1C), which is consistent with the higher
affinity of PCSK9-D374Y for LDLR at neutral pH
(see below) The extent of binding to blank wells was
similar for both ligands, and the radioactivity
asso-ciated with cells was at least 10 times of the counts
associated with blank wells
Fig 1 Cell density dependence of the binding of [125I]TC-PCSK9 and [ 125 I]TC-PCSK9-D374Y to HepG2 cells Varying numbers of HepG2 cells were grown in complete growth medium (lower curves) or LPDS-containing Opti-MEM (upper curves), as described
in Experimental procedures The relative amount of cell-surface immunoreactive LDLR (A), as measured by the amount of specific binding of 125I-labeled anti-(rabbit IgG) to cells, and the specific binding of [ 125 I]TC-PCSK9 (5 lgÆmL)1, 70 n M ) (B) and [ 125 I]TC-PCSK9-D374Y (1 lgÆmL)1, 14 n M ) (C) to cells were determined as described in Experimental procedures Mean cell numbers in wells seeded with the highest cell density were 7.55 · 10 5 and 7.5 · 10 5 for cells grown in complete growth medium and LPDS-containing Opti-MEM, respectively The numbers of cells in wells with lower cell numbers could not be reliably determined The results are means ± standard deviations of triplicate determinations from a sin-gle experiment Similar results were obtained in several indepen-dent single-point binding experiments at high cell density, each performed in duplicate The values given in the text are the means ± standard deviations of three independent experiments performed at high cell density, including the results obtained with the highest cell density in this experiment.
Trang 3As the gain-of-function D374Y mutation is localized
in the LDLR-binding region of PCSK9, the enhanced
binding of [125I]TC-PCSK9-D374Y to HepG2 cells can
be attributed entirely to its higher affinity for LDLR
We therefore conclude that LDLR is the main receptor
responsible for binding of [125I]TC-PCSK9 and
[125I]TC-PCSK9-D374Y to HepG2 cells
To further demonstrate the specificity of [125
I]TC-PCSK9 binding, we incubated the cells for 30 min at
4C in the presence of different unlabeled ligands
prior to incubating the cells with [125I]TC-PCSK9 The
inclusion of a 200-fold excess of unlabeled
PCSK9-D374Y reduced binding by 82–90% (depending on the
batch used) Unlabeled wild-type PCSK9 and
PCSK9-S127R were also able to reduce the binding of
[125I]TC-PCSK9 (see below) TLDLR possesses distinct
binding domains for apoB-100 (the main
apolipopro-tein in LDL) and PCSK9 In agreement with previous
studies showing that LDL can inhibit the uptake
of PCSK9 in cells [19,20], binding of [125I]TC-PCSK9
was inhibited by > 60% when unlabeled LDL
(1 mgÆmL)1) was included in the incubation medium
(data not shown), presumably because of steric
block-ing of the adjacent EGF-A domain Incubation with
formaldehyde-treated BSA at a concentration sufficient
to saturate scavenger receptors (1 mgÆmL)1) [21] had
little effect on specific [125I]TC-PCSK9 binding to
HepG2 cells (data not shown), further supporting the
specificity of the binding
Estimation of binding affinity of wild-type and
two mutant variants of PCSK9
In preliminary saturation experiments, we found that it
was very difficult to achieve complete saturation curves
for [125I]TC-PCSK9 binding to HepG2 cells
More-over, large amounts of unlabeled PCSK9 (wild type
and D374Y) were required to determine nonspecific
binding These technical limitations precluded
determi-nation of equilibrium dissociation constants (Kd) for
[125I]TC-PCSK9 and [125I]TC-PCSK9-D374Y
In order to estimate the binding affinities of PCSK9
and PCSK9-D374Y for HepG2 cell LDLR, we
incu-bated the cells with a fixed concentration of [125I]
TC-labeled ligand in the presence of increasing
concen-trations of the unlabeled counterpart (Fig 2B,C) The
abilities of unlabeled D374Y and
PCSK9-S127R to reduce [125I]TC-PCSK9 binding were also
compared with that of unlabeled wild-type PCSK9
(Fig 2A) The binding data were analyzed either by
nonlinear regression analysis or by the method of
Scat-chard Nonlinear regression analysis of data from
indi-vidual binding curves indicated that the data were
described best by a two-binding site model The IC50 (the concentration of the competing ligand that inhib-its 50% of the specific binding of [125I]TC-PCSK9) val-ues for the higher-affinity and lower-affinity sites are shown in Table 1, where it can be seen that binding is inhibited most effectively by unlabeled PCSK9-D374Y Unlabeled PCSK9-D374Y was also able to reduce the binding of its labeled counterpart ([125 I]TC-PCSK9-D374Y) to HepG2 cells in a concentration-dependent manner (Fig 2C; Table 1) The relative proportions of the two binding sites were roughly equal
Analysis of the same binding data by the Scatchard method yielded concave upward curves (Fig 2B,C, insets), suggesting the presence of two binding sites⁄ states The Kd values obtained for the higher-affinity and lower-affinity binding sites are listed in Table 1 The high-affinity and the low-affinity sites accounted for 25% and 75% of the total binding, respectively These estimates of the relative proportions of sites are different from those estimated by nonlinear regression analysis However, it has been well established that the Scatchard method is sensitive to slight experimental errors, making accurate estimates of the number of binding sites from Scatchard plots difficult [22,23] The higher-affinity and lower-affinity sites for PCSK9 binding on HepG2 cells may be indicative of either the presence of two subpopulations of LDLR with different affinities for PCSK9, or negative cooper-ativity among interacting LDLRs, although other explanations are also possible (see below) The Hill coefficient (the slope of Hill plot) is often used as a measure of the extent of cooperativity, and a Hill coef-ficient < 1.0 might suggest negative cooperativity [24] However, the average Hill coefficient calculated for PCSK9-WT was equal to unity (0.98 ± 0.06, n = 3), and that obtained for PCSK9-D374Y was not signifi-cantly different from unity (0.87 ± 0.07, n = 3) (not shown)
Kinetic characteristics of [125I]TC-PCSK9 binding
to HepG2 cells
To determine whether the kinetics of binding of PCSK9 to HepG2 cells can be described as a simple bimolecular reaction, the kinetics of [125I]TC-PCSK9 and [125I]TC-PCSK9-D374Y dissociation from and association with HepG2 cells were determined
Kinetic association The time course of [125I]TC-PCSK9 association with HepG2 cells at 4C is shown in Fig 3A,B Specific binding of [125I]TC-PCSK9 (5 lgÆmL)1, 70 nm) to cells
Trang 4reached an apparent equilibrium within 4 h Binding
of [125I]TC-PCSK9-D374Y (1 lgÆmL)1, 14 nm) to
HepG2 cells indicates that [125I]TC-PCSK9-D374Y at
a concentration five times lower than that of [125
I]TC-PCSK9 bound to the cells in a similar time-dependent
manner, and approached binding equilibrium at nearly
the same rate, suggesting a higher affinity of [125
I]TC-PCSK9-D374Y for cell surface LDLR For both
[125I]TC-PCSK9 and [125I]TC-PCSK9-D374Y, the time
courses of binding were biphasic, and data from
indi-vidual association curves were well fitted by a
two-phase exponential association model, suggesting that
surface binding has two components, one rapid and
one slow The half-time for association of [125
I]TC-PCSK9 with the rapid component, representing 35%
(± 5.3%) of specific equilibrium binding, was 6.6 min
(± 1.03 min), whereas the half-time for binding to the
slow component was 94 min (± 23 min) (n = 3) The corresponding half-times for [125I]TC-PCSK9-D374Y binding were 6.1 min (± 0.8 min) and 89 min (± 9 min) (n = 3), respectively The observed associa-tion rate constants kobs for the rapid phase [kobs(rapid)] and for the slow phase [kobs(slow)] are shown in Table 2
Kinetic dissociation The data in Fig 4 show the time course of [125 I]TC-PCSK9 and [125I]TC-PCSK9-D374Y dissociation from HepG2 cells It is evident that the dissociation of both [125I]TC-labeled ligands is biphasic, with two kinetic components Data from individual dissociation curves were best described by a model of two exponential decay phases Approximately 25% of the bound [125I]TC-PCSK9 dissociated during the rapid phase, with a half-time of 19 min (± 2.7 min), and the remaining bound [125I]TC-PCSK9 dissociated slowly, with a half-time of 270 min (± 22 min) (n = 3) The
A
B
C
Fig 2 Nonlinear regression and Scatchard analyses of binding-inhi-bition data (A) Inhibinding-inhi-bition of [125I]TC-PCSK9 (2.5 lgÆmL)1) binding to HepG2 cells ( 9.8 · 10 5 ) by increasing concentrations of unlabeled wild-type (upper curve), D374Y (lower curve) and S127R (middle curve) variants of PCSK9 Bars the denote range of duplicate deter-minations For wild-type PCSK9 and PCSK9-S127R, only the upper and lower, respectively, halves of the ranges are shown, to avoid overlap of the error bars The results are expressed as percentage
of control (c.p.m in the absence of unlabeled ligand), plotted against the concentration of unlabeled ligands (B) Inhibition of [ 125 I]TC-PCSK9 binding to HepG2 cells (9.2–9.8 · 10 5 ) by increasing concentrations of unlabeled wild-type PCSK9 Error bars are stan-dard deviations for data from three separate experiments [including the one shown in (A)] The results are expressed as a percentage
of c.p.m in the absence of unlabeled wild-type PCSK9, plotted against the concentration of unlabeled wild-type PCSK9 Inset: a representative Scatchard plot of the competition for [ 125 I]TC-PCSK9 binding by unlabeled wild-type PCSK9 (C) Inhibition of [125 I]TC-PCSK9-D374Y (1 lgÆmL)1) binding to HepG2 cells (9.2–9.8 · 10 5 )
by increasing concentrations of unlabeled PCSK9-D374Y Error bars are standard deviations for data from three separate experiments The results are expressed as a percentage of c.p.m in the absence
of unlabeled PCSK9-D374Y, plotted against the concentration of unlabeled PCSK9-D374Y Inset: a representative Scatchard plot of the competition for [ 125 I]TC-PCSK9-D374Y binding by unlabeled PCSK9-D374Y All of the data shown in the three panels were described best by a two-binding site model Scatchard plots of the competition for [125I]TC-PCSK9 binding by PCSK9-S127R and PCSK9-D374Y [shown in (A)] exhibit curvatures similar to those of wild-type PCSK9 and PCSK9-D374Y (not shown) Each curve was analyzed separately, and the parameters (IC 50 , K d ) determined from these experiments (means ± standard deviations) are shown in Table 1 The asterisk indicates the range of duplicate determina-tions from a single experiment.
Trang 5amount of [125I]TC-PCSK9 that remained bound after
6 h was about 40% Approximately 20% of the bound
[125I]TC-PCSK9-D374Y dissociated rapidly, with a
half-time of 21 min (± 3.7 min), and the remaining
bound [125I]TC-PCSK9-D374Y dissociated more
slowly, with a half-time of 297 min (± 25 min) (n = 3) The fraction of [125I]TC-PCSK9-D374Y that remained bound after 6 h was about 45% The dissoci-ation rate constants [koff] for the rapid phase [koff(rapid)] and for the slow phase [koff(slow)] are shown in Table 2 Taken together with the association data, these results suggest that the increased affinity of PCSK9-D374Y for cell surface LDLR is mainly determined by the rate of association It should be noted that quanti-tative analysis of the more rapid phase of association requires measurement of the binding on time scales of minutes, a time resolution that is difficult to achieve in experiments with adherent cell cultures This phase appears to be complete by the second measurable time point (15 min) in our system, and this precludes esti-mation of the association rate constant [kon] and the kinetic Kd(i.e the Kdrepresenting the koff⁄ konratio)
Dissociation of [125I]TC-PCSK9 in the presence of unlabeled PCSK9
In order to establish whether the [125I]TC-PCSK9 bound
to the cells after 6 h was dissociable, dissociation experi-ments were performed in the absence and presence of unlabeled PCSK9 (Fig 4A) In the presence of a high concentration of unlabeled wild-type PCSK9, the half-time of the rapid phase was reduced from 19 to 16 min and the half-time of the slow phase was reduced from
270 to 154 min In the presence of unlabeled PCSK9-D374Y, the half-time of the rapid phase of [125 I]TC-PCSK9-D374Y dissociation was reduced from 21 to
19 min, and the half-time of the slow phase was reduced from 297 to 179 min (Fig 4B) These results suggest that the reason for ligand remaining bound to cells after
6 h is not irreversibility of the binding
Dissociation of [125I]TC-PCSK9 at low pH
As the affinity of PCSK9 for the ECD or the EGF-A domain of LDLR is known to increase at acidic pH
Table 1 Parameters obtained from binding-inhibition experiments Best-fit values for IC 50 were derived from nonlinear regression analysis.
K d values were derived separately from Scatchard plots High and low represent affinities of binding sites for unlabeled ligands Values in parentheses indicate the number of experiments performed The data from each experiment were analyzed separately, and mean val-ues ± standard deviations were calculated from these valval-ues.
Fig 3 Association time courses of the binding of [ 125 I]TC-PCSK9
and [ 125 I]TC-PCSK9-D374Y to HepG2 cells at 4 C Cells ( 9.2–
9.5 · 10 5
) were incubated for the indicated times in binding
medium containing [ 125 I]TC-PCSK9 (5 lgÆmL)1) or [ 125
I]TC-PCSK9-D374Y (1 lgÆmL)1) At each time point, the cells were washed, and
the specific binding was determined (A) Binding presented as
per-centage of total radioactivity added (B) Binding presented as the
amount of [ 125 I]TC-labeled ligand specifically bound Error bars are
standard deviations for data from three separate experiments The
binding data were normalized for cell number (per 106cells) The
curves were fitted with the two-phase exponential association
model.
Trang 6[15,25], it is believed that, subsequent to internalization
of the PCSK9–LDLR complex, LDLR is diverted to a
degradation pathway, owing to persistence of the
com-plex at endosomal pH [15] However, the rates of
dis-sociation of PCSK9 that has previously bound to
LDLR at neutral pH have not been determined under
different pH conditions to test this hypothesis The
effect of pH on the dissociation of [125I]TC-PCSK9
was measured at pH 6.2 (to mimic the early endosomal
pH) As shown in Fig 4A, low pH did indeed
mark-edly reduce the dissociation of [125I]TC-PCSK9 from
cells Dissociation occurred as a monophasic process
with a rate constant [koff] of 0.0009 ± 0.00012 min)1
(n = 3), which corresponds to a half-time of
dissocia-tion of 770 min Lowering the pH of dissociadissocia-tion
med-ium to 6.2 also led to monophasic and slow
dissociation of [125I]TC-PCSK9-D374Y with a
half-time of 810 min (Fig 4B)
This observation is in marked contrast to what is
seen in many receptor systems, including the insulin
receptor [26], where lowering of the pH leads to
disso-ciation of ligand from receptor The reduced
dissocia-tion at low pH may reflect dissociadissocia-tion from a single
high-affinity site or, alternatively, it may reflect the
sum of two dissociation processes It is likely that a
similar mechanism may be at work in the slightly
acidic early endosomal compartments, where reduced
pH will decrease dissociation of the internalized
PCSK9–LDLR complexes
Effect of association time on the dissociation of
[125I]TC-PCSK9
The dependence of dissociation of [125I]TC-PCSK9
and [125I]TC-PCSK9-D374Y on the length of
associa-tion time was investigated to determine whether the
proportions of the two kinetic components correspond
to the presence of two distinct receptor sites⁄ states
with different and fixed affinities The prediction of
this mechanism is that the proportions of the two
components will be constant ( 3 : 1 for [125 I]TC-PCSK9 and 4 : 1 for [125I]TC-PCSK9-D374Y) and independent of the length of association time This prediction was tested by comparing the dissociation rate for binding under the pre-equilibrium conditions (60 min) and equilibrium conditions (240 min) As can
be seen in Fig 5A, dissociation of [125I]TC-PCSK9 was biphasic at both association times Approximately 42% and 22% of dissociation occurred in the rapid phase after short and long incubation times, respec-tively The dependence of the kinetics of [125 I]TC-PCSK9-D374Y dissociation on the length of binding time was also examined Again, dissociation from the rapid component was found to be faster at 60 min than at 240 min (36% versus 19% under equilibrium conditions) (Fig 5B)
These data suggest that there is a fraction of rapidly dissociating receptors, 40% after 60 min of associa-tion, that converts with time to a receptor state that releases bound ligand very slowly The size of the frac-tion undergoing conversion may be larger at shorter association times (i.e shorter than 60 min) However, determination of the half-time for this conversion requires an analysis of shorter-term aspects of this pro-cess, which is not possible in our system, owing to the relatively long durations of such experiments
Discussion
This is the first study aimed at characterizing the bind-ing of PCSK9 to intact cells by usbind-ing radiolabeled PCSK9 Several observations indicate that LDLR is the main surface receptor mediating PCSK9 binding to HepG2 cells First, the number of LDLRs on HepG2 cells increased following growth in the absence of lipo-proteins, and there was a corresponding increase in the binding of [125I]TC-PCSK9 and [125 I]TC-PCSK9-D374Y Second, the extent of specific binding of both [125I]TC-labeled ligands was a linear function of the cell density Finally, the binding is specific, as the
Table 2 Parameters obtained from kinetic experiments k off values were obtained by fitting the time course data with a two-exponential decay phase model k off(rapid) and k off(slow) are the dissociation rate constants for the rapid and slow dissociation components, respectively.
kobsvalues were obtained by fitting the time course data with a two-exponential association model kobs(rapid)and kobs(slow)are the observed association rate constants for the rapid and slow association phases, respectively The values for the constants are means ± standard devia-tions of three experiments The data from each experiment were analyzed separately The concentradevia-tions used are those described in the text.
Labeled ligand
Trang 7binding of [125I]TC-PCSK9 was reduced by unlabeled
wild-type as well as by two mutant variants of PCSK9
and LDL, but not by an unrelated ligand, suggesting
that they all compete with [125I]TC-PCSK9 for the
same receptor
Linear [27–29] and curvilinear [30,31] Scatchard
plots for LDL binding to LDLR are observed
Analy-sis of inhibition curves of [125I]TC-PCSK9 and
[125I]TC-PCSK9-D374Y by the Scatchard method
consistently showed curvilinear plots that implicated the presence of high-affinity and low-affinity sites⁄ -states with affinities for PCSK9 that differ approxi-mately five-fold An apparently good fit of nonlinear regression analysis of binding data to a two-site model was also obtained, suggesting that the equilibrium binding of PCSK9 to cell surface LDLR is not a sim-ple bimolecular reaction (see Doc S1, model A) One possibility that could explain the observed cur-vilinear Scatchard plots is that unlabeled and labeled ligands have different affinities for the receptors [32] However, this seems to be less likely, as the presence
of two classes of PCSK9 binding site were also observed in kinetic experiments where only [125 I]TC-labeled ligands were employed Moreover, we believe that the [125I]TC-labeling method, in contrast to the direct 125I-labeling method, does not appreciably alter the binding properties of PCSK9, as [125
I]TC-PCSK9-Fig 5 Dissociation kinetics of [125I]TC-PCSK9 (A) and [125 I]TC-PCSK9-D374Y (B) as a function of time of association: HepG2 cells ( 9 · 10 5 ) were incubated at 4 C in medium containing [ 125 I]TC-labeled ligand for 60 or 240 min After removal of unbound [125 I]TC-labeled ligand, the cells were incubated at 4 C in fresh medium, and dissociation was measured as described in Experimental proce-dures Data are presented as percentage of total ligand bound at zero time (100%) The 240-min data are from a single experiment, and error bars represent range of duplicate determinations Error bars in 60-min curves represent mean ± one-half the range from two independent experiments, each performed in duplicate.
Fig 4 Dissociation time courses and effects of unlabeled ligands
and low pH on the dissociation rates of [ 125 I]TC-PCSK9 (A) and
[ 125 I]TC-PCSK9-D374Y (B) from HepG2 cells Binding to equilibrium
and removal of unbound [ 125 I]TC-labeled ligands were performed as
described in Experimental procedures The cells were then
incu-bated in dissociation medium (pH 7.4) without (control) or with
unlabeled ligand or medium (pH 6.2), and dissociation of specifically
bound ligands was followed as a function of time Data are
pre-sented as percentage of total ligand bound at zero time (100%).
Error bars are standard deviations for data from three separate
experiments Asterisks indicate error bars representing mean ±
one-half the range from two separate experiments Dissociation in
the absence (control) and presence of unlabeled ligand was best
described by a two-exponential decay phase model, whereas
disso-ciation at low pH was best described by a one-exponential decay
phase model Final concentrations of unlabeled ligands in the
disso-ciation medium were 120 lgÆmL)1 (wild-type PCSK9) and
30 lgÆmL)1 (PCSK9-D374Y) If the Kd values estimated here are
assumed, then about 70% of the high-affinity sites and about 12%
of the low-affinity sites are expected to be occupied at the
concen-trations of unlabeled ligands used.
Trang 8D374Y consistently displayed a much higher affinity
for HepG2 cell surface receptors than did [125
I]TC-PCSK9 Binding to sites other than LDLR, such as
LDLR related protein 1 (LRP1) [4], may also be
responsible for the observed curvature However, we
consider this possibility to be less likely, because LRP1
has been shown to be not regulated by cellular
choles-terol levels [33], and therefore cannot account for the
observed increased binding of [125I]TC-PCSK9 to
HepG2 cells grown in the absence of lipoproteins It is
more likely that the observed heterogeneity in binding
of [125I]TC-PCSK9 to HepG2 cells is attributable to
binding to different populations of LDLR that exist
prior to ligand binding (see below)
Previous Biacore studies have primarily used the
ECD of LDLR to determine the affinity and kinetics
of binding of PCSK9 The Kdvalues reported for
wild-type PCSK9 interaction with the ECD of LDLR at
neutral pH [19,25,34,35] differ by about an order of
magnitude (ranging from 90 to 840 nm) The Kd of
wild-type PCSK9 for the high-affinity LDLRs
(626 ± 113 nm) in intact cells estimated from
Scat-chard plots is within this range The calculated Kd
value for PCSK9-D374Y binding to the high-affinity
sites was 125 ± 20 nm Other investigators have
reported an apparent Kdfor binding of PCSK9-D374Y
to the ECD of LDLR that is similar to (101 nm) [19]
or 20-fold lower (6 nm) [25] than the Kd estimated
here The apparent Kd of PCSK9-S127R binding to
the high-affinity LDLR site (548 nm), as measured by
its ability to inhibit [125I]TC-PCSK9 binding to HepG2
cells, was slightly lower (i.e slightly higher affinity)
than that for wild-type PCSK9, and is comparable to
the 648 nm Kd obtained in a Biacore study [19] The
Kd values derived for the lower-affinity class of sites
are shown in Table 1 In this context, it is worth
men-tioning that the existence of high-affinity (32 nm) and
low-affinity (86 nm) states has previously been
demon-strated for binding of PCSK9-S127R to the ECD of
LDLR at pH 7.5 [25] This study also found
high-affinity (1 nm) and low-high-affinity (42 nm) binding states
for the interaction between wild-type PCSK9 and the
ECD of LDLR at pH 5.4, whereas PCSK9-D374Y
binds with only one affinity (Kd= 6 nm) at pH 7.5
and with a slightly higher affinity (Kd= 1.6 nm) at
pH 5.4 It therefore seems likely that the ECD of
LDLR also adopts different conformations when
immobilized on the biosensor chip
The observation of two kinetic components in the
association and dissociation kinetics also suggests the
presence of two populations of binding site However,
in experiments that examined the dissociation as a
function of association time (Fig 5), it was found that
increasing the association time increased the propor-tion of the slowly dissociating component, at the expense of the component with rapid dissociation This result cannot be simply explained by the presence of two populations of LDLR that have different and fixed affinities for PCSK9 (see Doc S1, model B) A possible explanation might be that binding of PCSK9
to the low-affinity form of LDLRs is followed by a slow conformational change of the ligand–receptor complex to the higher-affinity state, whereas this con-formational transition is faster when PCSK9 binds to the high-affinity form of receptor
A model (see Doc S1, model C) that appears to be consistent with the kinetic data is one in which LDLRs on HepG2 cells are in equilibrium between monomer and dimer states and PCSK9 interacts with both populations of the receptor via a two-step reac-tion in which the first binding step, representing bind-ing to the EGF-A domain, is followed by bindbind-ing of PCSK9 to a second site within the receptor In this model, dimeric and monomeric receptor states bind PCSK9 with equal affinity, but they differ in their con-version rates, i.e rate constants governing the confor-mational change that leads to the second binding step Thus, the rapid phase of PCSK9 association could rep-resent binding of PCSK9 to dimeric receptors that release bound ligand slowly because they convert rap-idly The slow phase of association could represent binding to monomeric receptors that release bound ligand rapidly because they convert slowly The pro-posed model is supported by the finding that a signifi-cant proportion of LDLRs in the plasma membrane pre-exist as noncovalent dimers (or higher oligomers)
in coated pits [36–38] or even outside coated pits [39], and by the recent demonstration that PCSK9 can also bind, via its C-terminal domain, to the LDL-binding domain of LDLR [18]
The molecular basis of the enhanced rate of dissoci-ation observed in the presence of unlabeled ligand is unclear This phenomenon has often, but not always, been interpreted as indicative of the presence of nega-tive cooperativity, i.e a decrease in affinity with increasing site occupancy [40] At the present time, a mechanistic explanation of negative cooperativity, if present, in this system would be difficult, although negative cooperativity among partially occupied dimeric receptors or between two binding sites on a monomeric divalent receptor [41] cannot be excluded
It should be noted that, given the small amount of [125I]TC-PCSK9 initially bound and its low affinity for LDLR, rebinding of dissociated [125I]TC-PCSK9 from the bulk solution cannot account for the slowly dissociating component, although rebinding from the
Trang 9putative ‘unstirred layer’ surrounding the cells [42]
can-not be excluded
The presence of two apparent classes of binding site
with different affinities for PCSK9 on HepG2 cells
raises the question of whether both classes of site are
involved in internalization and whether the rate
con-stants of association and dissociation for [125
I]TC-PCSK9 at 37C are similar to values obtained at
4C Our preliminary data indicate that HepG2 cells
are able to internalize and degrade [125I]TC-PCSK9 at
37C However, the interaction of ligands with cell
surface receptors at 37C is a function not only of the
rate constants of association and dissociation, but also
of the endocytic rate constant, and measurements of
these rate constants require a method to discriminate
between the surface-bound and internalized ligand [43]
However, an accurate measurement of 125I-TC-PCSK9
association and dissociation rates at 37C is difficult
to obtain, because, as discussed in Results, in contrast
to many ligand–receptor systems, acid wash does not
favor dissociation of [125I]TC-PCSK9 from the plasma
membrane We are currently trying to develop a wash
method that can effectively remove cell surface-bound
[125I]TC-PCSK9
Experimental procedures
Materials
Culture media and antibiotics, l-glutamine and nonessential
amino acids were from Gibco BRL (Invitrogen, Carlsbad,
CA, USA) Antibody against LDLR was from RDI
Research Diagnostic (Concord, MA, USA) BSA and fetal
bovine serum were from Sigma Aldrich (St Louis, MO,
USA) Na125I was purchased from PerkinElmer (Waltham,
MA, USA) IodoGen-precoated tubes were from Pierce
Biotechnology (Rockford, IL, USA) All other chemicals
and reagents were obtained from Sigma Aldrich unless
otherwise specified
Protein expression and purification
PCSK9-D374Y and PCSK9-S127R are two naturally
occur-ring gain-of-function mutants of PCSK9 that cause severe
hypercholesterolemia [1,11,12] Histidine-tagged PCSK9s
(wild type, D374Y, and S127R) were produced by
transfec-tion of HEK239 cells, and purified from conditransfec-tioned media
as previously described [44]
Radiolabeling of proteins
We initially attempted to investigate the binding of PCSK9
to HepG2 cells using PCSK9 directly labeled with 125I
([125I]PCSK9), which reacts with tyrosines of the protein However, analysis in single-point binding assays showed only 50–60% of the cell-associated [125I]PCSK9 could be inhibited in the presence of a large excess of unlabeled PCSK9, indicating a high level of nonspecific binding This, combined with the low affinity of PCSK9 for LDLR, made reliable measurements of specific binding difficult To over-come the problem of low specific binding, we used labeling
by [125I]TC, which reacts with lysines, and we found [125I]TC labeling of PCSK9 to be more suitable for equilib-rium and kinetic binding studies, because of the much lower level of nonspecific binding The reason why different labeling methods produce molecules with different binding properties is unclear, but the results suggest that the bind-ing of [125I]PCSK9 to a nonreceptor site is particularly enhanced by the radio-iodination of a tyrosine(s) It should
be noted that the EGF-A-binding region of PCSK9 con-tains no tyrosines (or lysines)
Purified wild-type PCSK9 was covalently coupled to [125I]TC by the method of Pittman et al [45], with modifi-cations as described previously [46] Briefly, [125I]TC was prepared by reacting TC (6 lL of 10 mm solution in NaCl⁄ Pi) with Na125I (1.0 mCi) in IodoGen-precoated tubes (Pierce) for 30–40 min at room temperature, followed
by transfer to a tube containing cyanuric chloride (6 lL of
10 mM solution in acetonitrile) and potassium iodide (6 lL
of 0.1 m solution) for 3 min The activated [125I]TC adduct was then incubated with wild-type PCSK9 (300–400 lg in
200 lL of carbonate buffer containing 0.5 mm CaCl2,
pH 8.9) for 30–45 min Unreacted [125I]TC (and free 125I) were removed by gel filtration with Sephadex G25 columns (PD-10; GE Healthcare) equilibrated with NaCl⁄ Pi contain-ing 0.2 mm CaCl2 [125I]TC labeling of PCSK9-D374Y was carried out as previously described, except that 80–100 lg
of purified protein was used for labeling Specific activities obtained were in the range of 6–7· 105c.p.m lg)1 The quality and integrity of the [125I]TC-labeled proteins were evaluated by SDS⁄ PAGE followed by EZBlue staining (Pierce) and autoradiography of the gel [125I]TC-PCSK9 is stable, and can be stored at 4C for at least 2 weeks For direct iodination, purified PCSK9 (100 lg in 100 lL of NaCl⁄ Pi containing 0.5 mm CaCl2, pH 7.4) was incubated with Na125I (0.3 mCi) in IodoGen-precoated tubes for 10–
15 min at room temperature The reaction was stopped by transfer to a tube containing 0.9 mL of NaCl⁄ Picontaining 0.5 mm CaCl2(pH 7.4) Free125I was removed by gel filtra-tion as described above Goat anti-(rabbit IgG) was radio-labeled with Na125I as described for direct iodination of PCSK9, except that CaCl2was omitted
Cell culture, buffers, and cell treatments HepG2 cells (European Collection of Cell Cultures, Porton Down, UK) were routinely cultured in collagen-coated 75-cm2 tissue culture flasks (BD Biosciences, San Diego,
Trang 10CA, USA) in MEM supplemented with 200 mm
l-gluta-mine, nonessential amino acids, and 10% fetal bovine
serum (referred to as complete growth medium) in a
5% CO2 atmosphere at 37C For binding and kinetic
experiments, cells were plated at 250 000–320 000 cells per
well in collagen-coated 12-well culture plates (BD
Bio-sciences, San Diego, CA, USA) After incubation for 22–
24 h at 37C in complete growth medium, cells were
washed twice with 1 mL of Opti-MEM and incubated in
Opti-MEM containing 3% lipoprotein-depleted serum
(LPDS) for 24–26 h at 37C to upregulate LDLR In some
experiments, the cells were incubated in Opti-MEM without
LPDS Cell-free wells (blanks) were treated in an identical
manner, and were included in all experiments
Collagen-coated culture plates were used to avoid cell detachment
during the relatively long (> 13 h) durations of some
experiments All experiments were performed at 4C to
prevent endocytosis of bound ligand The cells were also
kept at 4C during the washing steps The standard
bind-ing medium for all experiments was DMEM containbind-ing
20 mm Hepes and 1% BSA (pH 7.4) The dissociation
(chase) medium was the same as that for binding, except
that, in some experiments, the pH of the chase medium was
adjusted to 6.2 with Mes at 4C Exposure of cells to low
pH did not damage the cells during the chase period
( 6 h), as judged by the ability of cells to exclude Trypan
blue dye At the end of each experiment, two wells were
treated with trypsin and the average cell number was
deter-mined with a hemocytometer All experiments were
per-formed on several different HepG2 cell batches and, with
the exception of PCSK9-S127R (one preparation), at least
three different wild-type PCSK9 and PCSK9-D374Y
prepa-rations and five different prepaprepa-rations of [125I]TC-labeled
proteins were used
Determination of specific binding of
[125I]TC-PCSK9 to HepG2 cells
Varying numbers of HepG2 cells (70 000, 140 000 and
280 000) were seeded in 12-well plates and incubated either
for 48 h in complete growth medium or for 24 h in
com-plete growth medium, and then for 26 h in 3%
LPDS-con-taining Opti-MEM In experiments with cells grown in
complete growth medium for 48 h, cells were incubated in
lipoprotein-free medium for 2 h at 37C before the start of
the experiments, to allow internalization of cell
surface-bound LDL After 50 h of incubation at 37C, cells were
washed once with 1 mL of DMEM and incubated in the
same medium for 15 min at 4C The medium was then
removed, and triplicate wells were incubated with 0.5 mL
of incubation medium containing 10 lgÆmL)1 of antibody
against LDLR for 90 min at 4C The cells were then
washed, and the amount of bound antibody against LDLR
was measured by subsequent binding of 125I-labeled goat
anti-(rabbit IgG) (10 lgÆmL)1) A control in which the
anti-body against LDLR was omitted was used to measure the amount of nonspecific binding Triplicate wells were also incubated with 0.5 mL of incubation medium containing either [125I]TC-PCSK9 (5 lgÆmL) or [125 I]TC-PCSK9-D374Y (1 lgÆmL)1) for 4 h at 4C The cells were washed once with 1 mL of cold wash buffer, NaCl⁄ Pi containing 0.1 mm CaCl2 and 0.5% BSA, and three times with 1 mL
of wash buffer without BSA The cells were solubilized in 0.5 mL of 0.2 m NaOH, and transferred to counting tubes following a 15-min incubation at room temperature Wells were washed with an additional 0.5 mL of 0.2 m NaOH, added to the counting tube, and counted on a gamma counter Similar experiments were carried out in single-point binding assays at high initial cell density
Binding-inhibition experiments Cells grown in 12-well plates were washed once with 1 mL
of DMEM and incubated in the same medium for at least
15 min at 4C The medium was removed, and cells were incubated with increasing concentrations of unlabeled (seri-ally diluted two-fold) proteins in a total volume of 480 lL
at 4C for 15 min Labeled ligand was added in a small volume (20 lL) and the cells were incubated at 4C for
4 h At the end of incubation, the cells were washed once with 1 mL of cold wash buffer and three times with 1 mL
of wash buffer without BSA The wash procedure took
4 min per plate After washing, cells were solubilized and radioactivity was measured as described above
Kinetic association experiments Cells grown in 12-well plates were washed as described above The medium was removed, and the cells were incu-bated with 0.5 mL of binding medium containing [125 I]TC-PCSK9 ( 5 lgÆmL)1, 70 nm) or [125I]TC-PCSK9-D374Y ( 1 lgÆmL)1, 14 nm) At various times, cells were washed four times and solubilized, and radioactivity was measured
as previously described A 200-fold excess of unlabeled PCSK9-D374Y was added to selected wells, in order to allow estimation of nonspecific binding Specific binding was calculated by subtracting nonspecific from total binding
Kinetic dissociation experiments Cells grown in 12-well plates were washed as described above, and were incubated for 4 h at 4C in binding med-ium containing [125I]TC-PCSK9 ( 5 lgÆmL)1) or [125 I]TC-PCSK9-D374Y ( 1 lgÆmL)1) At the end of incubation, the cells were washed four times with wash buffer as described above, and then incubated in fresh binding medium At various times, cells were washed once with
1 mL of cold DMEM and solubilized, and radioactivity was measured as described above Nonspecific binding was