Article No. jmbi.1999.2714 available online at http://www.idealibrary.com on
J. Mol. Biol. (1999) 288, 1027±1036
High-resolution AFM-imaging and Mechanistic
Analysis of the 20 S Proteasome
Ingmar T. Dorn1,2, Rainer Eschrich1, Erika SeemuÈller1
Reinhard Guckenberger1 and Robert TampeÂ1,2*
1
Max-Planck-Institut fuÈr
Biochemie, Am Klopferspitz
18a, D-82152 Martinsried
Germany
2
Lehrstuhl fuÈr Biophysik
Technische UniversitaÈt
MuÈnchen, James-Franck-Straûe
D-85747 Garching, Germany
As macromolecular protease complex, the 20 S proteasome is responsible
for the degradation of cellular proteins and the generation of peptide epitopes for antigen presentation. Here, structural and functional aspects of
the 20 S proteasome from Thermoplasma acidophilum have been investigated by atomic force microscopy (AFM) and surface plasmon resonance
(SPR). Due to engineered histidine tags introduced at de®ned positions,
the proteasome complex was pre-oriented at ultra-¯at chelator lipid
membranes allowing for high-resolution imaging by AFM. Within these
two-dimensional protein arrays, the overall structure of the proteasome
and the organization of individual subunits was resolved under native
conditions without ®xation or crosslinking. In addition, the substrate-proteasome interaction was monitored in real-time by SPR using a novel
approach. Instead of following enzyme activity by product formation, the
association and dissociation kinetics of the substrate-proteasome complex
were analyzed during proteolysis of the polypeptide chain. By blocking
the active sites with a speci®c inhibitor, the substrate binding step could
be dissected from the degradation step thus resolving mechanistic details
of substrate recognition and cleavage by the 20 S proteasome.
# 1999 Academic Press
*Corresponding author
Keywords: antigen processing; enzyme; interface; membrane;
multicatalytic proteinase
Introduction
The 20 S proteasome is a multicatalytic protease
found in all three domains of life. It is responsible
for the degradation of misfolded or regulatory proteins and is therefore essential for cell function
(Baumeister et al., 1998). In vertebrates, the resulting pool of protein fragments is used in a more
speci®c function for the loading of MHC class I
molecules in the antigen processing pathway (York
& Rock, 1996). The 20 S proteasome consists of 28
subunits which can be divided in type a and b.
Although the architecture of proteasomes is highly
conserved, eukaryotic proteasomes contain seven
different a and b-subunits, whereas the 20 S proPresent address: R. TampeÂ, Institute fuÈr
Physiologische Chemie, ZellulaÈre Biochemie &
Biophysik, Philipps-UniversitaÈt Marburg, Karl-vonFrisch-Str. 1, D-35033 Marburg, Germany.
Abbreviations used: AFM, atomic force microscopy;
SPR, surface plasmon resonance; RU, resonance units.
E-mail address of the corresponding author:
[email protected]
0022-2836/99/201027±10 $30.00/0
teasome of Thermoplasma acidophilum is composed
of identical a and b-subunits. Therefore, the
Archaea proteasome represents an ideal model to
study the basic structure and mechanism of this
macromolecular enzyme complex. The 28 subunits
of the 20 S proteasome are arranged in four heptameric rings which are stacked to form a barrelshaped complex of 15 nm in length and 11 nm in
diameter, whereby the a-subunits form the two
outer rings and the b-subunits the two inner rings.
The proteasome can be described as a nanocompartment, because perpendicular to the four rings
a narrow channel penetrates throughout the whole
macromolecular protease and connects three inner
cavities. The proteolytically active sites are located
within the central cavity, whereas the function of
the two outer cavities is as yet unknown. The catalytic sites are single N-terminal threonine residues
of the b-subunit (LoÈwe et al., 1995; SeemuÈller et al.,
1995). Consequently, the proteasome belongs to
the new class of Ntn (N-terminal nucleophile)
hydrolases (Brannigan et al., 1995). Protein degradation by the 20 S proteasome is controlled by selfcompartmentalization because only unfolded pro# 1999 Academic Press
1028
teins can pass through the narrow channel and
therefore have access to the inner compartments.
In contrast to other proteases, protein degradation
is rather slow and occurs processively, i.e. one
bound substrate is degraded completely before
another one is attacked (Akopian et al., 1997;
Kisselev et al., 1998). To gain more information
about the fundamental mechanistic steps of the
proteasome it is important to monitor the enzymesubstrate complexes during substrate degradation,
in addition to the overall analysis of product formation or substrate degradation.
Scanning probe techniques are of growing interest for high-resolution imaging of functional biological samples under native conditions in an
aqueous environment (Engel et al., 1997). In this
way, reversible structural changes in proteins
(MuÈller et al., 1996) as well as the interaction of
proteins with ligands (Florin et al., 1994), DNA
(Nettikadan et al., 1996), or substrate molecules
(Radmacher et al., 1994; Vinckier et al., 1998), can
be monitored online. An essential prerequisite for
high-resolution
imaging
by
atomic
force
microscopy (AFM) is an ultra-¯at and biocompatible support. This can be achieved by coating molecularly ¯at mica and gold surfaces with functional
and biocompatible ®lms of polymers (Ohnishi et al.,
1996), self-assembled monolayers (Wagner et al.,
1996), or lipid membranes (Weisenhorn et al.,
1990). In contrast to direct adsorption onto the
solid surface, immobilization via covalent or
ligand-receptor-mediated binding onto these coatings, suppresses non-speci®c interactions of the
protein with the solid surface. Additionally, buffer
conditions can be varied over a broad range to
minimize the interactions of the AFM tip with the
sample. In addition to structural studies, real-time
measurements on immobilized proteins can be performed, and thermodynamic as well as kinetic parameters of, e.g. ligand-receptor (Khilko et al., 1995;
Schmid et al., 1998) or enzyme-substrate interactions (Frazier et al., 1997; Haruki et al., 1997), can
be determined.
Since its introduction in 1994 (Schmitt et al.,
1994; Shnek et al., 1994), the chelator lipid concept
has evolved to become a powerful method for
structural and functional studies of immobilized
proteins. Following complex formation between
the chelator head group of the lipid and transition
metal ions, biomolecules can speci®cally be bound
via a short, engineered sequence of six histidine
residues (Dietrich et al., 1995; Dorn et al., 1998b;
Gritsch et al., 1995). Protein binding occurs in an
oriented fashion (Bischler et al., 1998) and the function of the proteins is preserved (Dietrich et al.,
1996; Dorn et al., 1998c). In addition to two-dimensional crystallization of proteins at chelator lipid
monolayers (Barklis et al., 1997; Frey et al., 1996;
Kubalek et al., 1994), polymerized chelator lipid
interfaces have been used as supports in order to
image single protein-DNA complexes by AFM
(Dorn et al., 1998a).
Structure and Function of the Proteasome
Here, we describe the formation of chelator lipid
membranes on solid substrates for structural and
functional studies on histidine-tagged 20 S proteasome by atomic force microscopy and surface plasmon resonance (SPR). Due to the introduction of
histidine tags at de®ned positions, the proteasome
complex was immobilized in a speci®c orientation
allowing for high-resolution AFM-imaging. In
addition, the substrate-proteasome complex could
be monitored in real-time by SPR, thus allowing
for the mechanistic analysis of substrate recognition and degradation by the proteasome.
Results and Discussion
Chelator lipid membranes as ultra-flat
docking interfaces
The stable immobilization and orientation of
proteins upon ¯at and biocompatible supports are
key requisites in structural and mechanistic studies
on proteins by scanning probe and surface sensitive techniques. Here, we used solid-supported
chelator lipid membranes to immobilize histidinetagged proteins in an oriented and highly speci®c
manner. The chelator membranes are spread by
vesicle fusion either as monolayers on hydrophobic
gold surfaces or as bilayers on freshly cleaved
hydrophilic mica. The formation of a chelator lipid
monolayer on a self-assembled monolayer of alkyl
thiol on gold was followed by SPR spectroscopy.
Repeated injection of a solution of 0.2 mM SOPC
vesicles containing 10 % chelator lipid (NTADODA) onto the hydrophobic surface resulted in a
change of refractive index at the surface (Figure 1)
which is proportional to an increase in mass. Due
to the deposition of a lipid layer we measured a
reproducible change in refractive index of approximately 3000 resonance units (RU). SOPC was chosen as the matrix lipid due to its ideal mixing and
phase behavior. These solid-supported membranes
Figure 1. Vesicle spreading at hydrophobic gold surfaces. A 100 ml sample of a 0.2 mM vesicle solution of
SOPC containing 10 % chelator lipid was injected three
times onto an HPA chip (Biacore AB, Uppsala, Sweden)
under a continuous ¯ow of 2 ml/min buffer containing
50 mM EDTA at 25 C.
Structure and Function of the Proteasome
1029
Figure 2. Supported chelator
lipid membranes on mica imaged
by AFM. (a) Unilamellar vesicles
(2 mM of total lipid) composed of
90 % SOPC and 10 % Ni-NTADODA was incubated on freshly
cleaved mica in buffer at 25 C.
After two hours, the surface was
rinsed extensively with buffer.
AFM images were recorded in tapping mode as described in detail in
Materials and Methods. Heights are
displayed as grey values with a
range of 15 nm height difference
between black and white. (b) In the
upper part an area with several
defects is shown at higher magni®cation. Height pro®les are taken along the black line and presented in the lower panel. These pro®les allow to
measure directly the thickness of the lipid membranes by the depth of the holes.
are stable over at least one week and highly resistant to osmotic stress or to treatment with bivalent
cations or non-speci®c protein adsorption. The
spreading of lipid monolayers is strongly in¯uenced on pretreatment of the surface, and it is
essential to clean the hydrophobic surfaces by
repeated injection of detergent solution prior to
vesicle spreading. The hydrophobic surfaces can be
reused many times, since lipid monolayers can be
entirely removed from the surface by detergent.
For bilayer formation we incubated a solution of
small unilamellar vesicles (2 mM lipid) composed
of 90 % SOPC and 10 % chelator lipid (Ni-NTADODA) on freshly cleaved mica. After incubation
for two hours at 25 C, the surfaces were washed
extensively with buffer and imaged by AFM. This
procedure leads to almost perfect bilayers over
large areas (>250 mm2) with only a few defects and
holes as shown in Figure 2(a). Because of the
dimensions of the holes, the relatively large AFMtip cannot touch the ground, thus the height of
these solid-supported membranes cannot exactly
be determined. In order to bypass this problem,
preparations with larger defects as shown in
Figure 2(b) (upper panel) were used for height
measurements. Here, a thickness of the supported
bilayers of approximately 5.5 nm was determined
(Figure 2(b), lower panel), which is in good agreement with published data (Shao et al., 1996).
ation level of 3000 RU. This value corresponds to a
surface concentration of 3 ng/mm2 (4.3 fmol/mm2)
or a surface coverage of 40 % assuming a molecular
area of 165 nm2 for the proteasome (LoÈwe et al.,
1995). During rinsing with buffer, almost no
change in signal was observed, demonstrating a
stable immobilization of the histidine-tagged proteasomes. Assuming a molecular area of 0.6 nm2
for one lipid in the ¯uid phase (Schmitt et al., 1994)
and 165 nm2 for one proteasome, there are
approximately 28 chelator lipids beneath each proteasome. Therefore, rebinding of the histidine tags
to the densely packed chelator groups in the lipid
membrane can occur, and this stabilizes the binding kinetically (Dorn et al., 1998c).
To provide further evidence for the speci®city of
the immobilization technique, the proteasome
could almost completely be removed (93 %) from
the surface by injection of 100 mM EDTA
Specific immobilization of the 20 S proteasome
To immobilize the 20 S proteasome in a speci®c
and oriented way on chelator lipid membranes,
histidine tags were introduced at the C terminus of
each a-subunit (see Figure 6). The binding of histidine-tagged proteasomes to chelator lipid membranes was followed in real-time by SPR. After
loading of the chelator lipid membrane with nickel
ions and subsequent injection of 0.3 mM proteasome onto the surface of the chelator lipid layer, a
rapid association reaction was observed (Figure 3).
Before reaching saturation, further adsorption was
stopped by washing with buffer at an immobiliz-
Figure 3. Reversible immobilization of 20 S proteasome at chelator lipid monolayers. (a) A solution of
0.3 mM proteasome was injected onto a nickel-loaded
chelator lipid monolayer (b) After injection, the surface
was rinsed with buffer. (c) To desorb bound proteasomes a solution of 100 mM EDTA (pH 7.0) was
injected. All the experiments were performed at a ¯ow
rate of 10 ml/min at 30 C.
1030
Structure and Function of the Proteasome
Figure 4. Speci®c adsorption of
proteasomes onto supported chelator lipid membranes. Proteasome
was injected onto nickel-loaded
chelator lipid membranes yielding
to a ®nal concentration of 7 nM in
buffer. AFM-images were recorded
after the (a) one minute, (b) 25 minutes, and (c) six hours at 25 C.
Image (d) was recorded one hour
after the addition of proteasome
(7 nM) to unloaded chelator lipid
membranes. The supported chelator lipid membranes were prepared
as described in the legend to
Figure 2. Details of the AFM experiments are given in Materials and
Methods. The thickness of the proteasome layer is approximately
10 nm in (b) and (c).
(Figure 3). This result is in line with the observation that less than 3 % non-speci®c adsorption of
histidine-tagged proteasomes onto unloaded chelator lipid membranes was observed (data not
shown).
In parallel with the SPR measurements, we followed the adsorption of proteasome molecules at
supported chelator lipid membranes on mica by
AFM (Figure 4). Within one minute following protein injection (7 nM ®nal concentration), proteasome complexes were detected at the lipid surface
(Figure 4(a)). The adsorbed proteasomes initially
form protein islands which are too mobile to be
imaged by the scanning AFM tip. With time, these
patches become larger and grow in a dendrite-like
fashion (Figure 4(b)). After six hours, a more or
less complete monomolecular protein layer was
formed on top of the lipid membrane which can be
imaged at high resolution (Figure 4(c)). The speci®city of the immobilization process was demonstrated by performing the same experiment as
described above, but without loading of the chelator lipid with nickel ions. In this case, speci®c binding of the protein via the histidine tag to the
chelator lipid is excluded. As a result, even after
one hour of incubation with 7 nM proteasome very
little protein adsorption to the surface was imaged,
indicating that non-speci®c interaction of proteasome with the chelator lipid membrane is very low
(Figure 4(d)).
To demonstrate the stability of the speci®cally
immobilized proteasome layer, the surface was
rinsed extensively with buffer (Figure 5). After six
hours of storage in proteasome-free buffer, the
immobilization level did not change signi®cantly.
Because the chelator lipid bilayers are in a ¯uid
phase, we observed a lateral mobility of the
immobilized proteasome. Depending on incubation
time and buffer conditions the proteasomes were
packed more densely and had oriented themselves
to each other (Figure 5(b)). These densely packed
proteasomes are suf®ciently stable to be imaged in
the contact mode of the AFM where higher lateral
forces are exerted on the sample. In this respect, it
is interesting to note that the two-dimensional crystallization of annexin V has recently been followed
by AFM at supported lipid bilayers (Reviakine
et al., 1998). It therefore seems possible to crystallize histidine-tagged proteins at solid-supported
chelator lipid membranes, and to follow the twodimensional crystallization process by AFM.
High-resolution AFM-imaging of the
20 S proteasome
In order to image single biomolecules at high
resolution, it is important to minimize the interaction of the tip with the substrate. In aqueous solution, this interaction strongly depends on vander-Waals and electrostatic forces which can be
Structure and Function of the Proteasome
1031
Figure 5. Stable immobilization
of proteasome at ¯uid chelator
lipid membranes. (a) Proteasome
complexes were imaged by AFM
microscopy after speci®c binding to
a mica-supported chelator lipid
bilayer. The 20 S proteasomes were
added to a ®nal concentration of
7 nM, and imaged by AFM after 60
minutes of incubation. Details of
the AFM experiments are given in
Materials and Methods. (b) The
surface was rinsed extensively with
proteasome-free buffer and an
additional image was recorded
after six hours. The arrow indicates
the preferred axial orientation of
the 20 S proteasomes. The supported membrane was prepared as
described in the legend to Figure 2.
altered by buffer conditions (MuÈller et al., 1997). By
using the chelator lipids for oriented protein immobilization, experiments were performed between
pH 6.5 and 9.5 and at concentrations up to
300 mM NaCl and no effect on the level of
immobilized proteasome was observed. The highest resolution was achieved at pH 7.5 and 150 mM
NaCl. In contrast, bivalent ions such as magnesium
ions at concentrations above 6 mM inhibited
adsorption of the proteasome almost completely
(data not shown).
To analyze the structure of single proteasome
molecules at high resolution by AFM, the lateral
and vertical forces during scanning had to be minimized. Therefore, we worked in the tapping mode
at very small free amplitudes (around 2-6 nm)
with a feedback set-point of 93-97 %. Although the
noise was increased, single proteasome molecules
were imaged at submolecular resolution (Figure 6).
Due to the position of the histidine tag at the C terminus of the a-subunit of the proteasome, only
side-views of the proteasome were observed. The
dimensions of single proteasomes (11 nm 15 nm)
measured by AFM in densely packed areas are in
good agreement with the three-dimensional structure (Baumeister et al., 1988; LoÈwe et al., 1995).
Interestingly, even structural details within single
particles can be resolved without any further ®xation or crosslinking of the proteins. The stripe-like
structure re¯ecting the a7-b7-b7-a7 rings of the 20 S
proteasome (Figure 6(b)) is in agreement with the
side-view of proteasomal particles imaged by electron microscopy. In some cases the four rings, in
other cases only three rings of the proteasome
could be resolved (Figure 6(c)). This can be
explained by the smaller gap between the two
inner b-rings in comparison with the larger gap
between the outer a and the inner b-ring
(Figure 6(b)). Within this substructure, individual
subunits can be distinguished. Although no image
processing was performed, a resolution of approximately 3 nm could be achieved for single mol-
Figure 6. High-resolution AFM
imaging of the 20 S proteasome.
(a) A detailed view of the adsorbed
proteasomes shows substructures
in the proteasome (the square has
dimensions of 15 nm 11 nm).
Details of the AFM experiments are
given in Materials and Methods. To
minimize the lateral and vertical
force exerted by the tip onto the
sample, we worked with a free
amplitude of the tip of 2-6 nm and
the feedback was set to 93-97 % of
the free amplitude. (b) Schematic
view of the proteasome. The histidine tags are illustrated as yellow
circles at the a-subunits. (c) Two
single proteasome molecules at a
higher magni®cation with their axis
oriented parallel with the diagonal from left top to right bottom. The substructured stripes perpendicular to the axis
represent the rings of the proteasome. Whereas the two inner rings appear in the left image as one, they are resolved
in the image presented at right.
1032
Structure and Function of the Proteasome
ecules. Thus, it seems possible to monitor the proteasome during substrate degradation or binding
of cofactors by AFM. Since the orientation of the
proteasome at the chelator lipid membrane can be
controlled by the position of the histidine tag in
the protein, top views of the 20 S proteasome
should predominate if mutants with histidine tags
at the end faces of the barrel-shaped complex are
used. In this orientation, interaction forces between
substrate bound to the AFM tip and single
immobilized proteasome molecules could be
measured during the processive degradation of the
polypeptide chain.
Real-time monitoring the substrate-proteasome
complex during substrate degradation
In order to demonstrate the function of proteasome complexes immobilized at the chelator lipid
interfaces, we monitored the substrate-proteasome
interaction in real-time by surface plasmon resonance. As a substrate we chose the oxidized
b-chain of insulin because it is a simple model substrate with a molecular mass suf®ciently high for
mass-sensitive SPR analysis. Although the 20 S
proteasome of the thermophilic achaeabacteria
T. acidophilum has its maximum activity at 95 C,
experiments were performed at 30 C due to instrumental limitations and in order to slow down substrate degradation. Injection of insulin under
continuous ¯ow onto the immobilized proteasome
resulted in an increase in resonance units at the
proteasome-lipid interface, indicating speci®c substrate binding to the 20 S proteasome (Figure 7,
black curve). Directly after this association phase,
the surface was rinsed with buffer and a rapid
decrease of the signal was monitored due to the
dissociation of the substrate-proteasome complex.
To extract information about substrate binding and
degradation from kinetic measurements, we
assumed the most simple reaction model:
k1
k2
E ‡ S „‰ES ! ! EPn Š ÿ! E ‡ nP
kÿ1
…1†
Here, the substrate S binds to the immobilized
enzyme E, building up the proteasome-substrate
complex ES. If we assume strictly processive degradation of substrate, the bound substrate is cleaved
at different positions resulting in complex EPn.
Because ES and EPn have the same overall mass
and geometry we assume that no change in refractive index at the surface will occur. Therefore, we
cannot dissect single mechanistic steps between
substrate binding and degradation by SPR. After
degradation of the substrate, all products were
released from the enzyme in one step. Because the
products are immediately removed by continuous
¯ow, rebinding of product as well as product inhibition is unlikely, and therefore does not in¯uence
the measurement. If formation of ES and EPn
results in the same change in refractive index, our
mechanism is identical with the Michaelis-Menten
Figure 7. Insulin degradation by the 20 S proteasome.
A 15 mM sample of insulin was injected onto immobilized proteasome (immobilization level: 3000 RU, for
details see Figure 3) from t ˆ 0 until t ˆ 120 seconds.
After subtraction of the non-speci®c interaction (for
details see Materials and Methods), the speci®c association and dissociation curve (black curve) from the
interaction of 15 mM insulin with immobilized proteasome was ®tted by equation (3) or (5) (black line). The
curve ®t started four seconds after changing buffer conditions, to account for mixing effects in the ¯ow
chamber. For inhibition, 100 mM LLnL was injected onto
immobilized proteasome for ten minutes. Directly thereafter, 15 mM insulin was injected and the speci®c signal
(red curve) was ®tted by equation (3) or (5) (red line).
Kinetic constants obtained from the ®ts are given in
Table 1. Residuals of the ®ts are displayed on bottom.
The experiments were performed under a continuous
¯ow of 10 ml/min of buffer at 30 C. Inset: Evaluation of
rate constants for insulin degradation by 20 S proteasome. The kinetic constants shown in Table 1 are plotted
against the insulin concentration. From a linear ®t of the
equation (4), k1 ˆ 2110 Mÿ1 sÿ1 and k-1 ‡ k2 ˆ 0.041 sÿ1
were calculated.
reaction and can be described as follows:
dC=dt ˆ k1 S0 E0 ÿ …k1 S0 ‡ kÿ1 ‡ k2 †C
…2†
taking into account E ˆ E0 ÿ C.
In Equation (2), C denotes the concentration of
the sum of ES and EPn assuming that ES and EPn
have an identical refractive index. So is the substrate concentration which is kept constant by
injection of an excess of substrate in comparison to
immobilized enzyme. E0 denotes the initial concentration of immobilized enzyme, E the concentration
of immobilized enzyme, and k1, kÿ1, and k2 the rate
constants of the reactions as shown in equation (1).
Because during the association phase binding of
substrate (reaction described by k1 in equation (1))
as well as release of uncleaved substrate (kÿ1 in
equation (1)) and formed products (k2 in equation
(1)) occurs, solution of equation 2 takes the form:
Cˆ
B
…1 ÿ exp…ÿqass t††
qass
…3†
whereby B ˆ k1 S0 E0 and the observed association
1033
Structure and Function of the Proteasome
constant qass is de®ned by:
qass ˆ k1 S0 ‡ kÿ1 ‡ k2
…4†
During the dissociation phase, the release of
cleaved products (k2 in equation (1)) and nondegraded substrate (kÿ1 in equation (1)) is monitored. Therefore, equation (2) can be solveXd by:
Cˆ
B
exp…ÿqdiss t†
qass
…5†
whereby the observed dissociation constant
qdiss ˆ kÿ1 ‡ k2.
The dissociation and association phase of the
substrate-proteasome interaction can be ®tted by a
monoexponential function (equations (3) or (5),
respectively) as shown in Figure 7 (black line).
Similar association and dissociation kinetics were
measured at ¯ow rates up to 30 ml/min, demonstrating that the reaction is not diffusion-limited
(data not shown). By varying the concentration of
insulin up to 30 mM, a linear dependence of the
observed association constant qass on the substrate
concentration was revealed, whereas the observed
dissociation constant qdiss was concentration independent (Table 1). At high substrate concentrations, saturation of the association phase is
reached at 35 RU, indicating in a rough estimation
that two insulin molecules per proteasome can be
bound. By plotting qass versus the insulin concentration, the rate constant k1 ˆ 2110 Mÿ1sÿ1 as well
as kÿ1 ‡ k2 ˆ 0.041 sÿ1 can be calculated from a ®t
of equation (4) (Inset in Figure 7). This value of
is
about
37 %
smaller
than
kÿ1 ‡ k2
kÿ1 ‡ k2 ˆ 0.056 sÿ1 determined directly from the
®t of the dissociation phase. Although it is argued
that the value of kÿ1 ‡ k2 is poorly de®ned in qass
(O'Shannessy et al., 1993) this result could also
indicate a more complex reaction mechanism than
assumed in equation (1).
In order to separate the reaction of the substrate
binding from substrate degradation, the proteolytically active centers were block by the proteasomal
inhibitor LLnL. By X-ray analysis it has already
been shown, that the peptide aldehyde LLnL binds
to all 14 active sites of the b-subunits (LoÈwe et al.,
1995), thereby inhibiting substrate degradation by
the 20 S proteasome. Consequently, we incubated
the immobilized proteasome with 0.1 mM LLnL
for ten minutes (data not shown). Directly thereafter, 15 mM insulin was added to the inhibited
proteasome and a sensogram was recorded. The
dissociation and association phase can be ®tted by
equations (3) or (5), respectively (Figure 7, red
curve). Interestingly, the observed association constant qass was only reduced of about 10 % in comparison to experiments with active proteasome,
whereas the observed dissociation constant qdiss
was signi®cantly decreased by about 60 % (Table 1).
Similar results were obtained for longer pre-incubation with LLnL (up to 60 minutes) of the proteasome with LLnL, indicating that all proteolytic
centers are blocked. In addition, no difference was
seen to proteasomes which were ®rst inhibited by
LLnL in solution and then immobilized. Since the
association phase of the experiment is dominated
by the binding of substrate (see equations (3) and
(4)), there is evidence that the substrate degradation occurs in more than one step whereby the
rate-limiting step is not affected by blocking of the
active sites. In contrast, the dissociation phase of
the experiment is dominated by the release of substrate and product (see equation (5)). By complete
inhibition of substrate degradation reaction, k2 in
equation (1) does not take place and therefore does
not contribute to the observed dissociation constant qdiss so that the dissociation phase is slower
in comparison to the experiments with active proteasome. Thus, using inhibited proteasomes substrate binding can be dissected from substrate
degradation and the observed rate constants qass
and qdiss describe substrate association and dissociation.
Although the assumed reaction mechanism ®ts
the experimental data and therefore seems most
likely, other reaction mechanisms cannot be
excluded. It is also possible that after each cut products (P1, P2, . . . Pn ‡ 1) are sequentially released
(equation (6)):
…6†
Table 1. Kinetic parameters for insulin degradation by
20 S proteasome
Proteasome
Active
Active
Active
Inhibited by LLnL
Insulin conc.
(mM)
qass (sÿ1)
qdiss (sÿ1)
10
15
30
15
0.062
0.072
0.104
0.064
0.053
0.060
0.056
0.025
Kinetic constants were calculated from best ®ts of equations
(3) or (5), to the measured SPR data as shown for one concentration in Figure 7 (for details see the Figure legend). Here, qass
denotes the observed association constant and qdiss the observed
dissociation constant.
In this model ES, B1, B2, and Bn have a different
mass and therefore a different refractive index.
Since these complexes also react with different velocities, the recorded SPR signal would be a superposition of several exponential functions with
different decay times. Although the experimental
data could be ®tted by a monoexponential function, it is also possible to ®t them by higher-order
exponential functions.
Degradation of proteins by the proteasome is a
multistep process. Loosely folded proteins are supposed to gain access to the central cavity by two
narrow openings and must wind their
way through a system of internal cavities and
1034
restrictions until they reach the active sites in the
central cavity. The underlying mechanism of translocation is unknown. By using a novel approach,
we followed the life-time of the substrate-proteasome complex in real-time by SPR. Typically, proteasome activity is assayed in the presence of a
large excess of substrate by product formation
after many enzyme cycles (Wenzel et al., 1994; Ehring et al., 1995; Kisselev et al., 1998). Here, the formation of the enzyme-substrate complex and the
subsequent release of products or substrates was
followed as a more or less single turnover event.
The method cannot discriminate between reactions
occurring within the substrate-proteasome complex
unless any release of products or substrates happens. However, by blocking the active sites in the
central cavity, the substrate binding event can be
dissected from the ®nal product release. As the
association kinetic is not signi®cantly affected by
blocking the active sites, substrate binding appears
to be rate-limiting. Consequently, the substrate
binding site of the proteasome must be distinct
from the catalytic site. However, experiments
using proteasome mutants with altered properties
in substrate binding, translocation and degradation
have to been analyzed, since a more complex
mechanism cannot so far be excluded. It has been
recently shown that the glycine-alanine repeat
domain of the Epstein Barr virus nuclear antigen-1
prevents proteasomal protein degradation and
thereby epitope generation for cytotoxic T-cell
response (Levitskaya et al., 1997). By applying this
novel approach, it will be interesting to see
whether the Gly-Ala repeat prevents substrate
binding, translocation into the active sites or the
®nal cleavage.
Materials and Methods
Materials
The chelator lipid NTA-DODA was synthesized as
described (Schmitt et al., 1994). 1-Stearoyl-2-oleyl-sn-glycero-3-phosphatidylcholine (SOPC) was ordered from
Avanti Polar Lipids (Birmingham, AL), oxidized insulin
b-chain from Sigma (Sigma, St. Louis, MO), and acetylleucyl-leucyl-norleucinal (LLnL) from Bachem AG
(Bubendorf, Switzerland). All other solvents and chemicals were purchased from Fluka (Neu-Ulm, Germany).
For all experiments, a sterile-®ltered and degassed buffer
(10 mM Hepes, 150 mM NaCl (pH 7.5)) was used if not
stated otherwise. Vesicles were prepared by mixing
appropriate molar ratios of lipid in organic solution,
evaporation of the solvent in vacuum, swelling in buffer
to a ®nal lipid concentration of 2 mM, and repeated
extrusion (21 times) through 100 nm ®lters (LiposoFastBasic, Avestin, Ottawa, Canada). In some experiments
NTA-DODA was preloaded with nickel ions (Ni-NTADODA) in chloroform/methanol (6:1) by adding equimolar amounts of NiCl2-6H2O dissolved in methanol.
Expression and purification of the 20 S proteasome
The a and b-subunit of the 20 S proteasomes were
coexpressed in Escherichia coli BL21(DE3), and the
Structure and Function of the Proteasome
assembled 20 S complex was puri®ed on Ni-NTA resin
(Qiagen GmbH, Hilden, Germany) via histidine tags at
the C terminus of each a-subunit as described (SeemuÈller
et al., 1995).
Structural studies by AFM
Supported membranes on mica were prepared by
incubation of a 2 mM vesicle solution at 25 C on freshly
cleaved mica. After two hours the surfaces were
thoroughly rinsed with buffer and mounted into the
¯uid cell of the AFM. Care was taken to always keep
the solid-supported membranes submerged in buffer.
A Nanoscope IIIa AFM (Digital Instruments, Santa Barbara, California) with oxide-sharpened silicon nitride
pyramidal tips (NPS, Digital Instruments) was used in
tapping mode for imaging. Cantilevers with nominal
spring constants of 0.12 and 0.32 N/m were used. The
tapping frequency was between 9 and 15 kHz. The tips
were cleaned by UV radiation of about one hour prior to
use. To minimize the lateral and vertical force exerted by
the tip onto the sample, we worked with a free amplitude of the tip of 2-6 nm and the feedback was set to 9397 % of the free amplitude in Figures 5 and 6. Figures 2
and 4 were taken with 25 nm amplitude and a set-point
of 90 %.
Functional studies by SPR
Supported membranes on gold were formed by vesicle fusion onto a alkyl thiol gold chip (HPA chip, Biacore AB, Uppsala, Sweden) by repeated injection of a
0.2 mM vesicle solution into the Biacore-X instrument
(Biacore AB, Uppsala, Sweden) at a ¯ow rate of
2 ml/min and at a temperature of 25 C. Prior to vesicle
spreading the HPA chip was cleaned by rinsing the surface at least three times with 40 mM n-octyl glucoside
in deionized water for ten minutes at a ¯ow rate of
10 ml/min. Shortly before protein immobilization, the
chelator lipids were loaded with nickel ions by injection
of a 1 mM NiCl2 solution in buffer. After rinsing the surface by buffer, a 0.3 mM solution of proteasome was
injected into the ¯ow cell at a ¯ow rate of 10 ml/min
until an immobilization level of approximately 3000 RU
was reached. A total 1000 RU is equivalent to 1 ng
protein/mm2. The substrate-proteasome interaction was
analyzed by 20 ml injections of insulin onto the proteasome-derivatized surface at various concentration
(0-30 mM) and at a ¯ow rate of 10 ml/min. Non-speci®c
insulin binding to the chelator lipid membrane without
proteasome was recorded in parallel and the non-speci®c
signal was subtracted from the overall signal to yield the
speci®c signal. To inhibit immobilized proteasome,
100 mM LLnL was injected into the ¯ow cell for 10-60
minutes at a ¯ow rate of 10 ml/min. All experiments
except vesicle spreading were performed at 30 C.
Acknowledgments
We acknowledge Alexei Boulbitch, Klaus Neumaier
and Ulf RaÈdler for helpful discussions, and Erich
Sackmann for continuous encouragement. This work
was supported by the Deutsche Forschungsgemeinschaft,
the European Commission (BIOTECH program) and the
Volkswagen-Stiftung.
Structure and Function of the Proteasome
References
Akopian, T., Kisselev, A. & Goldberg, A. (1997). Processive degradation of proteins and other catalytic
properties of the proteasome from Thermoplasma
acidophilum. J. Biol. Chem. 272, 1791-1798.
Barklis, E., McDermott, J., Wilkens, S., Schabtach, E.,
Schmid, M. F., Fuller, S., Karanjia, S., Love, Z.,
Jones, R., Rui, Y., Zhao, X. & Thompson, D. (1997).
Structural analysis of membrane-bound retrovirus
capsid proteins. EMBO J. 16, 1199-1213.
Baumeister, W., Dahlmann, B., Hegerl, R., Kopp, F.,
Kuehn, L. & Pfeifer, G. (1988). Electron microscopy
and image analysis of the multicatalytic proteinase.
FEBS Letters, 241, 239-245.
Baumeister, W., Walz, J., ZuÈhl, F. & SeemuÈller, E. (1998).
The proteasome: paradigm of a self-compartmentalizing protease. Cell, 92, 367-380.
Bischler, N., Balavoine, F., Milkereit, P., Tschochner, H.,
Mioskowski, C. & Schultz, P. (1998). Speci®c interaction and two-dimensional crystallization of histidine-tagged yeast RNA polymerase I on nickelchelating lipids. Biophys. J. 74, 1522-1532.
Brannigan, J., Dodson, G., Duggleby, H., Moody, P.,
Smith, J., Tomchick, D. & Murzin, A. (1995). A protein catalytic framework with an N-terminal nucleophile is capable of self-activation. Nature, 378, 416419.
Dietrich, C., Schmitt, L. & TampeÂ, R. (1995). Molecular
organization of histidine-tagged biomolecules at
self-assembled lipid interfaces using a novel class of
chelator lipids. Proc. Natl Acad. Sci. USA, 92, 90149018.
Dietrich, C., Boscheinen, O., Scharf, K. D., Schmitt, L. &
TampeÂ, R. (1996). Functional immobilization of a
DNA-binding protein at a membrane interface via
histidine tag and synthetic chelator lipids. Biochemistry, 35, 1100-1105.
Dorn, I. T., Hofmann, U. G., Peltonen, J. & TampeÂ, R.
(1998a). Diacetylene chelator lipids as supports for
immobilization and imaging of proteins by atomic
force microscopy. Langmuir, 14, 4836-4842.
Dorn, I. T., Neumaier, K. R. & TampeÂ, R. (1998b). Molecular recognition of histidine-tagged molecules by
metal-chelating lipids monitored by ¯uorescence
energy transfer and correlation spectroscopy. J. Am.
Chem. Soc. 121, 2753-2763.
Dorn, I. T., Pawlitschko, K., Pettinger, S. C. & TampeÂ, R.
(1998c). Orientation and two-dimensional organization of proteins at chelator lipid interfaces. Biol.
Chem. 379, 1151-1159.
Ehring, B., Meyer, T. H., Eckerskorn, C., Lottspeich, F. &
TampeÂ, R. (1996). Effects of major-histocompatibility-complex-encoded subunits on the peptidase and
proteolytic activities of human 20 S proteasomes.
Eur. J. Biochem. 235, 404-415.
Engel, A., Schoenenberger, C.-A. & MuÈller, D. J. (1997).
High-resolution imaging of native biological sample
surfaces with scanning probe microscopy. Curr.
Opin. Struct. Biol. 7, 279-284.
Florin, E.-L., Moy, V. & Gaub, H. (1994). Adhesion
forces between individual ligand-receptor pairs.
Science, 264, 415-417.
Frazier, R., Davies, M., Matthijs, G., Roberts, C.,
Schacht, E., Tendler, S. & Williams, P. (1997). In situ
surface plasmon resonance analysis of dextran
monolayer degradation by dextranase. Langmuir,
13, 7115-7120.
1035
Frey, W., Schief, W. R., Pack, D. W., Chen, C. T.,
Chilkoti, A., Stayton, P., Vogel, V. & Arnold, F. H.
(1996). 2-Dimensional protein crystallization via
metal-ion coordination by naturally-occurring surface histidines. Proc. Natl Acad. Sci. USA, 93, 49374941.
Gritsch, S., Neumaier, K., Schmitt, L. & TampeÂ, R.
(1995). Engineered fusion molecules at chelator
lipid interfaces imaged by re¯ection interference
contrast microscopy (RICM). Biosens. Bioelect. 10,
805-812.
Haruki, M., Noguchi, E., Kanaya, S. & Crouch, R.
(1997). Kinetic and stochiometric analysis for the
binding of Escherichia coli ribonuclease HI to RNADNA hybrids using surface plasmon resonance.
J. Biol. Chem. 272, 22015-22022.
Khilko, S., Jelonek, M., Corr, M., Boyd, L., Bothwell, A.
& Margulies, D. (1995). Measuring interactions of
MHC class I molecules using surface plasmon resonance. J. Immunol. Methods. 183, 77-94.
Kisselev, A. F., Akopian, T. N. & Goldberg, A. L. (1998).
Range of sizes of peptide products generated
during degradation of different proteins by archaeal
proteasomes. J. Biol. Chem. 273, 1982-1989.
Kubalek, E. W., Le, Grice F. J. & Brown, P. O. (1994).
Two-dimensional crystallization of histidine-tagged
HIV-1 reverse transcriptase promoted by a novel
nickel-chelating lipid. J. Struct. Biol. 113, 117-123.
Levitskaya, J., Sharipo, A., Leonchiks, A., Ciechanover,
A. & Masucci, M. G. (1997). Inhibition of ubiquitin/
proteasome-dependent protein degradation by the
Gly-Ala repeat domain of the Epstein-Barr virus
nuclear antigen 1. Proc. Natl Acad. Sci. USA, 94,
12616-12621.
LoÈwe, J., Stock, D., Jap, B., Zwickl, P., Baumeister, W. &
Huber, R. (1995). Crystal structure of the 20 S proteasome from the archaeon Thermoplasma acidophiÊ resolution. Science, 268, 533-539.
lum at 3.4 A
MuÈller, D., Baumeister, W. & Engel, A. (1996). Conformational change of the hexagonally packed intermediate layer of Deinococcus radiodurans monitored
by atomic force microscopy. J. Bacteriol. 178, 30253030.
MuÈller, D., Amrein, M. & Engel, A. (1997). Adsorption
of biological molecules to a solid support for scanning probe microscopy. J. Struct. Biol. 119, 172-188.
Nettikadan, S., Tokumasu, F. & Takeyasu, K. (1996).
Quantitative analysis of the transcription factor AP2
binding to DNA by atomic force microscopy. Biochem. Biophys. Res. Commun. 226, 645-649.
O'Shannessy, D. J., Brigham-Burke, M., Soneson, K. K.,
Hensley, P. & Brooks, I. (1993). Determination of
rate and equilibrium binding constants for macromolecular interaction using surface plasmon resonance: use of nonlinear least squares analysis
methods. Anal. Biochem. 212, 457-468.
Ohnishi, S., Hara, M., Furuno, T. & Sasabe, H. (1996).
Imaging two-dimensional crystals of catalase by
atomic force microscopy. Jpn. J. Appl. Phys. 35,
6233-6238.
Radmacher, M., Fritz, M., Hansma, H. & Hansma, P.
(1994). Direct observation of enzyme activity with
the atomic force microscope. Science, 265, 1577-1579.
Reviakine, I., Bergsma-Schutter, W. & Brisson, A. (1998).
Growth of protein 2-D crystals on supported planar
lipid bilayers imaged in situ by AFM. J. Struct. Biol.
121, 356-361.
Schmid, E. L., Tairi, A. P., Hovius, R. & Vogel, H.
(1998). Screening ligands for membrane-protein
1036
Structure and Function of the Proteasome
receptors by total internal-re¯ection ¯uorescence the 5HT3 serotonin receptor. Anal. Chem. 70, 13311338.
Schmitt, L., Dietrich, C. & TampeÂ, R. (1994). Synthesis
and characterization of chelator lipids for reversible
immobilization of engineered proteins at selfassembled lipid interfaces. J. Am. Chem. Soc. 116,
8485-8491.
SeemuÈller, E., Lupas, A., Stock, D., LoÈwe, J., Huber, R.
& Baumeister, W. (1995). Proteasome from Thermoplasma acidophilum: a threonine protease. Science,
268, 579-582.
Shao, Z., Mou, J., Czajkowsky, D., Yang, J. & Yuan, J.
(1996). Biological atomic force microscopy - What is
achieved and what is needed. Advan. Phys. 45, 1-86.
Shnek, D. R., Pack, D. W., Sasaki, D. Y. & Arnold, F. H.
(1994). Speci®c protein attachment to arti®cial membranes via coordination to lipid-bound copper(II).
Langmuir, 10, 2382-2388.
Vinckier, A., Gervasoni, P., Zaugg, F., Ziegler, U.,
Lindner, P., Groscurth, P., PluÈckthun, A. &
Semenza, G. (1998). Atomic force microscopy
detects changes in the interaction forces between
GroEL and substrate proteins. Biophys. J. 74, 32563263.
Wagner, P., Hegner, M., Kernen, P., Zaugg, F. &
Semenza, G. (1996). Covalent immobilization of
native biomolecules onto Au(111) via N-hydroxysuccinimide ester functionalized self-assembled
monolayers for scanning probe microscopy. Biophys.
J. 70, 2052-2066.
Weisenhorn, A., Drake, B., Prater, C., Gould, A.,
Hansma, P., Ohnesorge, F., Egger, M., Heyn, S. &
Gaub, H. (1990). Immobilized proteins in buffer
imaged at molecular resolution by atomic force
microscopy. Biophys. J. 58, 1251-1258.
Wenzel, T., Eckerskorn, C., Lottspeich, F. & Baumeister,
B. (1994). Existence of a molecular ruler in proteasomes suggested by analysis of degradation products. FEBS Letters, 349, 205-209.
York, I. & Rock, K. (1996). Antigen processing and presentation by the class I major histocompatibility
complex. Annu. Rev. Immunol. 14, 369-396.
Edited by I. B. Holland
(Received 16 December 1998; received in revised form 11 March 1999; accepted 16 March 1999)