Structure of Bovine Pancreatic Cholesterol Esterase at 1.6 Å: Novel Structural
Features Involved in Lipase Activation†,‡
Julian C.-H. Chen,§Larry J. W. Miercke,|Jolanta Krucinski,|Jacqueline R. Starr,⊥,#Gina Saenz,⊥,@Xingbo Wang,⊥
Curtis A. Spilburg,⊥,+Louis G. Lange,⊥Jeff L. Ellsworth,⊥,Oand Robert M. Stroud*,|
Graduate Group in Biophysics and Department of Biochemistry and Biophysics, UniVersity of California,
San Francisco, California 94143, and CV Therapeutics, Inc., 3172 Porter DriVe, Palo Alto, California 94306
ReceiVed December 4, 1997; ReVised Manuscript ReceiVed February 4, 1998
ABSTRACT: The structure of pancreatic cholesterol esterase, an enzyme that hydrolyzes a wide variety of
dietary lipids, mediates the absorption of cholesterol esters, and is dependent on bile salts for optimal
activity, is determined to 1.6 Å resolution. A full-length construct, mutated to eliminate two N-linked
glycosylation sites (N187Q/N361Q), was expressed in HEK 293 cells. Enzymatic activity assays show
that the purified, recombinant, mutant enzyme has activity identical to that of the native, glycosylated
enzyme purified from bovine pancreas. The mutant enzyme is monomeric and exhibits improved
homogeneity which aided in the growth of well-diffracting crystals. Crystals of the mutant enzyme grew
in space group C2, with the following cell dimensions: a ) 100.42 Å, b ) 54.25 Å, c ) 106.34 Å, and
? ) 104.12°, with a monomer in the asymmetric unit. The high-resolution crystal structure of bovine
pancreatic cholesterol esterase (Rcryst) 21.1%; Rfree) 25.0% to 1.6 Å resolution) shows an R-? hydrolase
fold with an unusual active site environment around the catalytic triad. The hydrophobic C terminus of
the protein is lodged in the active site, diverting the oxyanion hole away from the productive binding site
and the catalytic Ser194. The amphipathic, helical lid found in other triglyceride lipases is truncated in
the structure of cholesterol esterase and therefore is not a salient feature of activation of this lipase. These
two structural features, along with the bile salt-dependent activity of the enzyme, implicate a new mode
of lipase activation.
Pancreatic cholesterol esterase (CEase1), also known as
bile salt-activated lipase, is responsible for the hydrolysis
of dietary cholesterol esters, fat soluble vitamin esters,
phospholipids, and triglycerides (1, 2). As such, it is one of
the central enzymes that mediates absorption of dietary lipids
through the intestinal wall into the bloodstream. A number
of studies have suggested a possible role for CEase in the
absorption of free cholesterol at the brush border membrane
of the small intestine (3-6), though a CEase gene knockout
showed little change in the absorption of cholesterol (7).
CEase is present in high concentrations in breast milk and
plays a crucial role in development by supplying the enzyme
to infants whose pancreases are not yet fully developed (1,
2). A role has also been demonstrated for CEase in the
hepatic uptake of HDL-associated cholesterol esters (8). The
bile salt dependence of enzymatic activity identifies CEase
as a distinct lipase with necessarily different structural
changes upon activation.
Bovine pancreatic CEase is a glycoprotein of 579 amino
acids, with a notable proline-rich region between amino acids
540 and 573, and a highly conserved six-amino acid hydro-
phobic sequence forming the extreme C terminus of the
protein (9). The proline-rich region is composed of a con-
sensus repeat sequence of (PVPPTGDSGAP)n, and the num-
ber of repeats marks the major difference between CEases
across the species (2); bovine pancreatic CEase contains three
such repeats, while human CEase contains 16 repeats.
Studies of the proline-rich repeats show they have no role
in enzymatic activity, and a physiological role for this region
remains elusive (10-13). The proline-rich repeats are also
serine- and threonine-rich and contain putative sites for
O-glycosylation. There are also two sites for N-linked
glycosylation at residues 187 and 361 (9, 14).
The purpose of this study was to determine the high-
resolution crystal structure of bovine pancreatic CEase to
gain insight into the molecular recognition of sterols, the
mechanisms of cholesterol absorption, sites of interaction
†This work was supported by NIH Grant GM24485 (R.M.S.).
‡Coordinates have been deposited in the Brookhaven Protein Data
Bank under accession number 2bce and will be available within a year
of publication of this paper.
§Graduate Group in Biophysics, University of California.
|Department of Biochemistry and Biophysics, University of Cali-
⊥CV Therapeutics, Inc.
#Present address: Department of Epidemiology, Box 357236,
University of Washington, Seattle, WA 98195-7236.
@Present address: Dade Behring, MicroScan, 2040 Enterprise
Boulevard, West Sacramento, CA 95691.
+Present address: Department of Medicine, Cardiology Division,
Jewish Hospital of St. Louis, Washington University Medical Center,
St. Louis, MO 63310.
OPresent address: ZymoGenetics, Inc., 1201 Eastlake Avenue East,
Seattle, WA 98102.
1Abbreviations: CEase, cholesterol esterase; CEase-C, cholesterol
esterase with amino acids 574-579 deleted; PC, phosphatidylcholine;
TC, taurocholate; p-NPB, p-nitrophenyl butyrate; HEK, human em-
bryonic kidney; HDL, high-density lipoprotein; SDS-PAGE, sodium
dodecyl sulfate-polyacrylamide gel electrophoresis; IEF, isoelectric
focusing; DLS, dynamic light scattering; MES, 2-(N-morpholino)-
ethanesulfonic acid; HEPES, N-(2-hydroxyethyl)piperazine-N′-2-ethane-
sulfonic acid; CHES, 2-(N-cyclohexylamino)ethanesulfonic acid.
5107Biochemistry 1998, 37, 5107-5117
S0006-2960(97)02989-9 CCC: $15.00© 1998 American Chemical Society
Published on Web 03/24/1998
with bile salt, and the mechanism of activation of this distinct
lipase. Since high-resolution diffraction quality crystals
could not be initially obtained from native wild-type protein
(C. A. Spilburg and L. G. Lange, unpublished), the two
N-linked sugars were removed by mutating asparagines 187
and 361 in the N-linked glycosylation consensus sequences
to glutamines, followed by overexpression in HEK 293 cells.
The purified double-mutant CEase was evaluated for elec-
trophoretic purity (SDS-PAGE, native PAGE, and IEF),
functional activity by cholesteryl oleate hydrolysis and
p-nitrophenyl butyrate (p-NPB) hydrolysis, and oligomeric
state and homogeneity by dynamic light scattering (DLS).
As this paper was being submitted, the 2.8 Å structure of
the native enzyme alone and in complex with bile salt was
solved independently, in a different space group (15). In-
terestingly, their apoenzyme structure is in a different con-
formation than our apoenzyme structure at 1.6 Å resolution,
which suggests a role for the aliphatic chain of the fatty acid,
mimicked by detergent in their structure, and a critical role
for the C terminus in the activation of the lipase revealed in
ours (16). Our structure shows a clear biological role for
the C-terminal region of the enzyme and implicates novel
structural features in the activation of this lipase.
MATERIALS AND METHODS
Cloning, Mutagenesis, and Expression. Bovine pancreatic
CEase was cloned as previously described (9). The full-
length, double-mutant CEase was constructed by first creating
two single mutants, N187Q and N361Q, using PCR-based
mutagenesis (17). N187Q was created using the mutagenic
TTTG-3′, while N361Q used the oligonucleotide 5′-CTCA-
GAGGTGCCGACGCCACGTACGAG-3′. The two cloned
inserts were sequenced, and then cleaved with restriction
enzyme ClaI, located between the codons for amino acids
187 and 361. The 5′ portion of N187Q and the 3′ portion
of N361Q were then ligated together into a pCEP4 expression
vector (Invitrogen) and expressed in HEK 293 cells as
described previously (18).
Purification of NatiVe and Mutant CEases. Native CEase
from bovine pancreas and double-mutant N187Q/N361Q
CEase expressed in HEK 293 cells were purified as previ-
ously described (18), with the double-mutant CEase requiring
modifications. For the double mutant, 2.1 L of cell media
was adjusted to 20 mM MES (pH 6.0) with 1.0 M MES,
0.45 µm filtered, and CEase precipitated by addition of
ammonium sulfate (856 g to 2.1 L) with stirring for 3 h.
The pellet was resuspended in 100 mL of 20 mM MES (pH
6.0) and dialyzed against 20 mM MES (pH 6.0). Following
S-Sepharose chromatography [3 cm × 40 cm; equilibrated
and loaded with 20 mM MES (pH 6.0), washed with 0.15
M NaCl, and CEase eluted at 0.2-0.35 M NaCl using a 0.15
to 0.6 M NaCl gradient], p-nitrophenyl butyrate (p-NPB,
Sigma) active fractions were pooled, concentrated to ap-
proximately 10 mL, and dialyzed for 24 h against 100 mM
MES (pH 6.0). The sample was applied to a 1.2 cm × 10
cm Poros CM weak cation exchanger (Perseptive Biosys-
tems) run at 3 mL/min in 100 mM MES (pH 6.0) with 2
mL injection loads. CEase was retarded on the column with
elution post flow through at 1.5-4.5 min.
postinjection, a 3.5 min gradient from 0 to 0.7 M NaCl was
At 5 min
run to remove impurities followed by a 5 min re-equilibra-
tion. Pooled CEase was concentrated to approximately 5
mL and dialyzed against the appropriate buffer. The protein
was concentrated by high-pressure stirred cell ultrafiltration
using YM-10 membranes (Amicon). The protein concentra-
tion was determined by the Bradford method using bovine
serum albumin as a standard.
Electrophoretic and Molecular Size Characterization.
SDS-PAGE and native PAGE were performed using a
Novex Xcell II Mini Cell and 4-20% acrylamide/Tris-
glycine precast gels in the presence and absence of 0.1%
SDS in the running buffer. Samples were prepared by
mixing with an equal volume of 2× reducing application
buffer (SDS-PAGE) or running buffer (native PAGE).
Isoelectric focusing (IEF) was performed using precast pH
3-10 ampholite gels. IEF samples were diluted with equal
volumes of IEF sample buffer. Gels were run according to
Novex Precast Gel Instructions. Molecular weights and
oligomeric states were assayed using the DynoPro-801
Dynamic Light Scattering Instrument equipped with a micro-
cuvette-based MicroSampler (Protein Solutions). Prior to
data collection, 1.2 mg/mL protein in 10 mM MES (pH
6.0) was filtered through a 0.02 µm Anodisc membrane
using the MicroFilter system (Protein Solutions). Software
version 3.00 was used for instrument control and data
Enzymatic ActiVity Assays. Functional activities of both
native and mutant CEases were assayed by monitoring the
hydrolysis of cholesteryl oleate and p-NPB as a function of
pH, enzyme concentration, and taurocholate (TC, Sigma)
concentration.[1-14C]Oleic acid release after a 5 min
incubation period at 37 °C from vesicles containing choles-
teryl [1-14C]oleate (Amersham) and egg yolk phosphatidyl-
choline (PC, Sigma) was measured as previously described
(19). Samples (25 ng of CEase/275 µL) were prepared by
mixing 175 µL of 100 mM buffer [sodium acetate (pH 4.0
and 5.0), MES (pH 6.0), HEPES (pH 7.0), Tris (pH 8.0),
and CHES (pH 9.0)], 75 µL of cholesteryl [1-14C]oleate
vesicles in 2.0 mM Tris (pH 7.2), e5.5 µL of 500 mM TC,
20 µL of CEase diluted 1000-fold in the appropriate pH
buffer, and water to give a total volume of 275 µL. The
rate of p-NPB hydrolysis at 25 °C was determined from the
linear portion of the absorbance curve (10-70 s) at 405 nm
using a Shimadzu UV-160 spectrophotometer equipped with
a liquid temperature control system. CEase in 10 mM MES
(pH 6.0) was diluted 50-fold in buffer (same as above)
containing the appropriate TC concentration to give 0.02 mg/
mL CEase. p-NPB was prepared by mixing 250 µL of fresh
stock (9 µL in 1 mL of dioxane) into 25 mL of 25 mM MES
(pH 6.0). The reaction was initiated by addition of 2 µL of
p-NPB to 1 mL of protein solution directly in a quartz
Crystallization and Data Collection. For crystallization,
purified N187Q/N361Q CEase was dialyzed against 2 mM
sodium phosphate (pH 7.0) and further concentrated to 5-6
mg/mL.Crystals were grown at room temperature in
hanging drops by mixing equal volumes of protein with well
buffer containing 0.1 M sodium acetate (pH 4.5), 1.5 M
ammonium sulfate, and 6% 2-propanol. Crystals appeared
in 5-8 days as twinned clusters of thin rods, averaging 200
µm × 60 µm × 30 µm in size, with a small number of single,
untwinned crystals suitable for data collection. Data were
5108 Biochemistry, Vol. 37, No. 15, 1998Chen et al.
collected on a 30 cm image plate (MAR Research). Cryo-
protectant of mother liquor containing 30% glycerol was
applied to the crystal shortly before flash freezing in a 90 K
nitrogen gas stream. A complete data set was collected on
a single frozen crystal at Stanford Synchotron Radiation
Laboratory beamline 7-1 (λ ) 1.08 Å), with diffraction
observed to the edge of the image plate at 1.6 Å resolution.
Data were indexed, integrated, and scaled using DENZO and
SCALEPACK with no σ cutoff, indexing in space group C2
with the following unit cell dimensions: a ) 100.42 Å, b )
54.25 Å, c ) 106.34 Å, and ? ) 104.12°, with a monomer
in the asymmetric unit. A summary of data collection and
refinement statistics is presented in Table 1.
Crystallography and Structure Refinement. Molecular
replacement in AMoRe (20) was used to generate an initial
map, and the model was improved by cycles of manual
building in CHAIN (21) and refinement in XPLOR (22).
Briefly, the primary sequences of bovine pancreatic CEase
and Torpedo californica acetylcholinesterase were aligned
and the acetylcholinesterase coordinates (PDB code 2ace)
truncated to a discontinuous, 368-amino acid core corre-
sponding to conserved regions of sequence and secondary
structure (23). Rotation and translation searches, followed
by rigid-body refinement in AMoRe, yielded an unambigu-
ous solution and a starting Rcrystof 48.3%. A 7.0% set of
reflections was set aside for calculation of Rfree prior to
refinement in XPLOR (24). An immediate bulk solvent
correction followed by simulated annealing refinement
reduced the Rcrystto 44.3%, with clear density corresponding
to the primary sequence of CEase.
followed by cycles of conjugate gradient minimization,
simulated annealing, and temperature factor refinement (40-
1.6 Å) dropped Rcrystand Rfreeto 21.1 and 25.0%, respec-
tively. A bulk solvent correction was applied regularly
throughout refinement. MIDAS was used for docking the
cholesterol ester (25). Ramachandran parameters were
monitored using PROCHECK (26).
Protein Expression, Purification, and Characterization.
The three-step purification procedure produced 5-7 mg of
the purified, recombinant double mutant from 2.1 L of
expression media. The purified N187Q/N361Q mutant
CEase consists of one SDS-PAGE band, one native PAGE
band, and three IEF bands with pI’s between 4.8 and 4.9
(Figure 1). In comparison, native CEase purified from
bovine pancreas migrates as a single, slightly faster migrating
SDS-PAGE band, a smeared, impeded multibanded pattern
on native PAGE, and multiple IEF bands with pI’s between
4.9 and 5.1 (Figure 1). These three electrophoretic assays
are indicative of glycosylation in native CEase.
Both the mutant N187Q/N361Q and native CEases exhibit
essentially identical functional activity (Figure 2). Both
enzymes show similar cholesteryl [1-14C]oleate hydrolytic
activities as a function of enzyme concentration (Figure 2a),
with peak hydrolytic activity at pH 7.0 (Figure 2b). Both
enzymes show similar bile salt taurocholate (TC) depend-
ences of enzymatic activity, with half-maximal activity
around 5.0 mM TC (Figure 2c). Using the water soluble
ester p-NPB, both enzymes display similar profiles as a
function of pH and TC concentration, with half-maximal
activity at ∼0.4 mM TC at pH 7.0 (Figure 2d). Taken
together, these experiments show that the presence of the
N-linked sugars is not required for ester hydrolysis.
Further characterization of the purified mutant CEase by
DLS yields a sample with a narrow, monomodal distribution
(baseline of 0.999-1.000) of monomers, with molecular
masses of 61.2 ( 1.2 and 65.5 ( 1.5 kDa, based on two
independent samples of 14 and 20 measurements, respec-
tively. This is in agreement with the calculated 63.5 kDa
molecular mass of the mutant, unglycosylated enzyme. In
addition, characterization of native bovine CEase also shows
a homogeneous, monomodal distribution of the monomer,
with an average molecular mass of 64.0 ( 3.0 kDa.
Crystal Structure. The structure was solved by molecular
replacement based on a truncated model of T. californica
acetylcholinesterase (Table 1) (23). CEase is a member of
Table 1: Crystallographic Data
Data Collection Statistics
a ) 100.42 Å, b ) 54.25 Å,
c ) 106.34 Å, and ? ) 104.12°
unit cell dimensions
resolution limit (Å)
completeness (%) (last shell)
Refinement and Model Statistics
532 (1-112, 120-533, and 574-579)
water molecules 199
average B-factor (Å2)14.044
rms ∆ bond lengths (Å) 0.007
rms ∆ bond angles (deg)1.377
Ramachandran geometry89.4% most favored regions
10.2% allowed regions
0.2% generously allowed regions
0.2% disallowed regions
FIGURE 1: Electrophoretic characterization of mutant and native
CEases: (a) SDS-PAGE, (b) IEF gel, and (c) native PAGE.
Molecular weight and IEF standards are indicated by column S,
and mutant and native CEases are indicated by columns M and N,
Structure of Pancreatic Cholesterol Esterase at 1.6 ÅBiochemistry, Vol. 37, No. 15, 1998 5109
the R-? hydrolase superfamily, the general fold found in
esterases and other triglyceride lipase structures, typically
fungal. The structure is composed of a system of 11
?-strands forming the core of the protein surrounded by 15
R-helices (Figure 3), with very clear electron density
throughout the molecule (Figure 4). The catalytic triad of
Asp320, His435, and Ser194 is located roughly at the center
of the molecule. The unique environment around the active
site residues distinguishes pancreatic CEase from other
lipases (Figure 5). On the basis of homology to residues in
acetylcholinesterase, Ala195, Gly107, and Ala108 of CEase
are likely involved in coordinating the oxyanion intermediate
in the reaction. Ala195 forms the N terminus of an
11-residue R-helix, placing it at the positive end of the helix
dipole, contributing to the stabilization of the negatively
charged reaction intermediate (27).
hydrogen bonds to the amino group of Ala195 in the
apoenzyme structure. Strikingly, the extreme C terminus
with its hydrophobic sequence PVVIGF is lodged in the
active site (Figures 5 and 6a). The C terminus physically
A water molecule
displaces the putative oxyanion binding residues Gly107 and
Ala108 from the catalytic serine. Compared to acetylcho-
linesterase, the side chain of conserved Tyr105 is flipped,
and the residues following veer away from the active site
(Figure 6a). The hydrophobic C terminus also functions as
a plug, and a calculation of accessible surface area upon
removal of this hexapeptide shows that 37 mostly hydro-
phobic residues and 504 Å2of surface area are exposed.
Importantly, the catalytic serine and histidine and the putative
oxyanion binding residues are made more accessible to
substrate (Table 2). Therefore, this C-terminal hexapeptide
clearly plays a role in activation of this lipase.
CEase also lacks an amphipathic, helical lid region
important for interfacial activation found in most triglyceride
lipases, and instead, it is truncated into a pair of antiparallel
?-strands. The N-terminal disulfide loop in CEase, contain-
ing residues 64-80, makes up this truncated lid region. A
superposition of the structures of pancreatic CEase and
Candida cylindracea CEase (28) clearly shows the truncation
of the lid region (Figure 6b).
FIGURE 2: Functional analysis of mutant and native CEase. (a) Cholesteryl [1-14C]oleate hydrolysis vs added enzyme. (b) Cholesteryl
[1-14C]oleate hydrolysis vs pH. (c) Cholesteryl [1-14C]oleate hydrolysis vs TC. (d) p-NPB hydrolysis vs TC and pH. The CEase concentration
was 25 ng/275 µL for cholesteryl [1-14C]oleate hydrolysis experiments, in the presence of 4 mM TC. p-NPB hydrolysis experiments were
performed at 0.02 mg/mL CEase. Squares and diamonds represent native and mutant CEase, respectively, and buffer blanks are indicated
5110 Biochemistry, Vol. 37, No. 15, 1998 Chen et al.
There are three distinct systems of ?-sheets, two core sys-
tems conserved in the R-? hydrolase fold, and a unique
third system lying near the active site. This third set of
?-sheets is composed of four strands, amino acids 66-68
and 74-76 in the truncated lid region, amino acids 108-110,
and amino acids 575-577 in the extreme C terminus (Figure
3). Immediately preceding the extreme C terminus is a pro-
line-rich repeat region, composed of the consensus repeat
sequence (PVPPTGDSGAP)n, where n ranges from 3 in cows
to 16 in humans (2). This proline-rich repeat is disordered
in the electron density and presumably very flexible.
Residues 113-119 in CEase are also disordered in the
electron density, corresponding to a hydrophilic loop near
the active site that is conserved in all CEase sequences, but
absent in other lipases and esterases. This was earlier
proposed to be a bile salt binding site, and was found to be
similarly disordered in the 2.8 Å apoenzyme structure (15).
An electrostatic potential map of the protein surface reveals
unusually large patches of negative potential, with a small
number of neutral and positive potential areas (Figure 7a).
A conglomerate of positive charge is centered around
residues 56-63, which was earlier determined to be the site
of interaction with heparin-like molecules on the cell surface
(29) (Figure 3). The primary sequence in this region,
KAKSFKKR, is similar to other heparin binding sequences
(2, 29). This patch is located 17-30 Å from the catalytic
Ser194, consistent with its proposed role of anchoring to the
cell surface (4-6).
A large lobe of negative potential is proximal to the ac-
tive site, in a helical domain composed of residues 322-
376. In addition to this segregated distribution of charge,
CEase also has an unusual distribution of temperature factors
(Figure 7b). CEase shows a patch of systematically higher
temperature factors also located in the helical domain
FIGURE 3: Topology and structure of bovine pancreatic CEase. The catalytic triad of Asp320, His435, and Ser194 is shown in ball-and-
stick representation at the center, and the putative heparin binding domain is shown at the lower right, consisting of residues 56-63. This
figure was generated using MOLSCRIPT (40).
FIGURE 4: Representative N-terminal electron density. A 2Fo- Fcmap was calculated to 1.6 Å resolution and contoured at 1σ. The map
shows clear electron density in an N terminus ?-sheet region of CEase. This figure was generated using MOLSCRIPT (40).
Structure of Pancreatic Cholesterol Esterase at 1.6 Å Biochemistry, Vol. 37, No. 15, 1998 5111
containing residues 322-376. This region of higher tem-
perature factors is strikingly coincident with the area of
negative potential, suggesting a possible role for electrostatic
interactions between the mixed micelle and the enzyme.
FIGURE 5: Structure of the active site region of CEase. Active site residues Ser194, Asp320, and His435 are labeled. Putative oxyanion
coordinating residues Ala195, Gly107, and Ala108 are highlighted in lowercase, and C-terminal residues 574-579 are labeled. Water
molecules close to Ser194 are indicated as spheres. This figure was generated using MOLSCRIPT (40) and Raster3D (41).
FIGURE 6: (a) R-Carbon alignment of the residues in the active site region of CEase (black) and T. californica acetylcholinesterase (gray).
Residues in lowercase correspond to acetylcholinesterase, and residues in uppercase correspond to pancreatic CEase. Conserved Tyr105 in
CEase is flipped with respect to Tyr116 in acetylcholinesterase, and the C terminus of CEase diverts the oxyanion binding loop away from
catalytic Ser194 in the apoenzyme structure. The orientation of the oxyanion loop in acetylcholinesterase may represent the structure of
CEase in this region in the activated state of the enzyme. (b) R-Carbon alignment of the N terminus of CEase and C. cylindracea CEase.
Amino acids 61-210 of CEase (black) were aligned with the corresponding residues of C. cylindracea CEase (gray). Comparison of the
regions enclosed by the N-terminal disulfide loop regions shows a truncated loop in CEase, and a helical lid in C. cylindracea CEase. The
disulfide bridge (Cys64-Cys80) is indicated, and the active site Ser194 is shown in ball-and-stick representation. This figure was generated
using MOLSCRIPT (40), Raster3D (41), and O (42).
5112 Biochemistry, Vol. 37, No. 15, 1998Chen et al.
Purification and Characterization of CEase. The mutation
of the two N-linked glycosylation sites and the subsequent
expression, purification, and functional characterization show
that no N-linked oligosaccharides are involved in the
hydrolytic activity of the enzyme. Electrophoretic charac-
terization of native CEase shows that the increase in the
number of isoelectric forms and average pI relative to those
of mutant CEase, and the change in mass-to-charge ratio in
native CEase, is due to the presence of the N-linked sugars.
Even though the function of the N-linked sugars is not well
understood, rat pancreatic CEase lacking the N-linked sugars
was secreted at 50% lower levels in Chinese hamster ovary
and pancreatoma cells, showing a role for glycosylation in
the secretion of the enzyme in vivo (30, 31). The N-linked
sugars are also implicated in protein folding and the thermal
stability of the enzyme (31). Although the amino acids in
the region surrounding residues 187 and 361 are neutral to
hydrophilic, the N-linked sugars may also play a role in
improved solubility. We show by DLS and crystal packing
that bovine CEase is a functional monomer as is human
CEase (32). Therefore, the native CEase dimers, present in
both the uncomplexed and TC-complexed structures of Wang
et al., are due to crystallization conditions and crystal
packing. Crucial to this study was the increased homogeneity
in the double mutant. Previously, we were unable to obtain
atomic resolution data from native CEase crystals, and the
improvements in homogeneity brought about by the removal
of the N-linked sugars may have contributed to the growth
of well-ordered crystals.
A Possible Role for the C Terminus in Lipase ActiVation.
The structure of CEase at 1.6 Å resolution shows an R-?
hydrolase with an unusual active site region, where the last
six amino acids of the enzyme, PVVIGF, splay the putative
oxyanion hole away from the active site (Figures 5 and 6a).
Defining a biological role for the C terminus has been
difficult (13); however, the last six amino acids are highly
conserved among mammalian CEases (Table 3). Therefore,
we expect that the C terminus plays a similar role in the
function of these enzymes. The length of the proline-rich
Table 2: Solvent Accessibility of Residues upon Removal of
C-Terminal Residues 574-579a
accessibility after removal
of peptide (Å2)
aCatalytic residues are indicated in bold, and the putative oxyanion
binding residues are in bold italics. Calculation of the solvent accessible
area was carried out using ASC (39).
FIGURE 7: Electrostatic potential and temperature factor profiles
of CEase. (a) Potential diagram ramped from -10/kT (red) to +10/
kT (blue). The putative heparin binding site lies at the area of
positive potential at the lower left, with Arg63 indicated. (b)
Temperature factor diagram of the active site face of bovine
pancreatic CEase. The catalytic His435 is indicated, with temper-
ature factors ramped from blue (low temperature factors) to red
(high temperature factors). The areas of high temperature factors
and negative potential are coincident, notably in the helical region
of amino acids 322-376 at the upper right. This figure was
generated using GRASP (43).
Structure of Pancreatic Cholesterol Esterase at 1.6 Å Biochemistry, Vol. 37, No. 15, 1998 5113
repeats preceding the C terminus varies across the mam-
malian CEases, with three repeats in the bovine enzyme, four
in the rat enzyme, and 16 in the human enzyme. Structurally,
the proline-rich repeats may act as a tether to the C terminus
and may exhibit similar disorder in the structures of other
The removal of the final six amino acids (CEase-C) ex-
poses a predominantly hydrophobic 504 Å2of surface area
(Table 2). Manual docking of cholesteryl linoleate (PDB
code 1cle) into the active site of CEase-C indicates that the
alkyl chain of the fatty acid fits in the hydrophobic pocket
left by the removal of the terminal amino acids, with the
proper tetrahedral geometry dictated by the oxyanion inter-
mediate (Figures 8 and 9) (28). Given that this probable
binding mode of the cholesteryl ester occupies the same
hydrophobic pocket as the C terminus, we propose that the
C-terminal peptide must move to accommodate the substrate.
The structure shows that this hexapeptide plays an important
role in the function of this enzyme not only by displacing
the oxyanion hole from the active site but also by occupying
a deep hydrophobic pocket that is the proper size for accom-
modating the fatty acid of a cholesterol ester (Figure 8).
The truncation of the lid region in pancreatic CEase and
the presence of the C-terminal peptide in the active site
suggest a unique set of structural rearrangements required
for lipase activation. Whether the movement of the peptide
is driven by substrate, by bile salt, or by a combination of
the two is unclear. The movement of the C-terminal peptide
may be driven by the interaction with bile salt. As the
C-terminal residues are hydrogen bonded to the mobile loop
implicated in bile salt activation, binding of bile salt and
the subsequent ordering of residues 113-119 may contribute
to the movement of the C-terminal peptide (33). Further-
more, enzymatic data on native and truncated forms of
CEase, where the proline-rich repeats and the hydrophobic
C terminus were deleted, show lowered Kmand elevated Vmax
values for the truncated CEases relative to those of the native
CEase at low, nonsaturating concentrations of bile salt (33).
At saturating concentrations of bile salt, the truncated and
native CEases are kinetically indistinguishable, and are
therefore structurally indistinguishable, indicative of a dis-
placed C terminus in the presence of bile salt (33).
Therefore, as the enzyme binds to the bile salt, the resulting
displacement of the C terminus allows the loop containing
the oxyanion binding site to move into a position that can
coordinate the intermediate (Figures 5 and 6a).
The movement of the C-terminal peptide may also be
driven by substrate binding, as implicated by the crystal
structure of the apoenzyme of CEase in the presence of
detergent. Wang et al., in their structure of the apoenzyme
determined independently at 2.8 Å resolution, show a
preformed oxyanion hole with no density seen for the C
terminus. While our construct involved mutation of two
N-linked consensus glycosylation sequences, the differences
in the apoenzyme structures are not due to the lack of
N-linked glycosylation in our structure.
glycosylation sites are distal to the active site and C terminus,
and furthermore, activity assays on the native and mutant
CEases show identical enzymatic activity (Figure 2); there-
fore, glycosylation does not play a role in the conformational
differences between our structure and that of Wang et al. In
light of their structure, it appears that crystallization condi-
tions may account for the notable differences between their
apoenzyme structure and ours. The authors use the detergent
zwittergent 3-12 as an additive in the crystallization. Despite
the lack of interpretable detergent density in the putative fatty
acid binding pocket in their structure, it is possible that the
long alkyl chain of the detergent may have helped displace
the C terminus from the active site, allowing the oxyanion
hole to move into its productive binding position. This is
consistent with the known sensitivity of the conformation
of lipase structures to crystallization conditions and suggests
a role for substrate in the displacement of the C terminus,
as mimicked by detergent. Our crystals lack any detergent
and therefore may more accurately reflect the conformation
of the enzyme in the absence of bile salt and substrate.
Nevertheless, the enzymatic data and the crystal structure
of the apoenzyme in the presence of detergent coupled with
our high-resolution apoenzyme structure point toward a clear
role for the displacement of the C terminus during lipase
Bile Salt Binding and Enzymatic ActiVity. Characterization
of native and mutant CEases shows different bile salt
concentration dependence curves of enzymatic activity for
water soluble versus hydrophobic esters, with an observed
half-maximal activation for TC of around 400 µM in the
case of water soluble ester hydrolysis and half-maximal
activation at a TC concentration of 5.0 mM in the case of
cholesteryl oleate hydrolysis. While these numbers cannot
be directly compared due to the unknown concentration of
substrate in the micelle and the dynamic physical chemical
properties of the mixed micelle, 5.0 mM is in the range of
the critical micelle concentration of TC, suggesting a role
of TC as a surfactant in addition to being an activator of
CEase. From the independently determined 2.8 Å structure
of the enzyme complexed with bile salt, two binding sites
were found, a proximal site and a distal site. The proximal
site near the catalytic triad orders residues 116-125 and
swings the binding loop away from the active site, thus
allowing easier access of substrate (15). On the basis of
our apoenzyme structure, we argue that the second, distal
site may be involved in opening the substrate binding pocket
to accommodate the bulkier cholesterol esters. The distal
bile salt binding site is located in a cleft between the he-
lical bundle domain of residues 322-376 and the core of
the enzyme. This is consistent with the temperature fac-
tor distribution of our structure, where this same helical
bundle domain has systematically higher temperature fac-
tors than the remainder of the molecule (Figure 7b). This
domain may be ordered by the binding of bile salt to the
An Electrostatic Role in Lipase ActiVation? Pancreatic
CEase shows a patch of systematically higher temperature
factors located in the helical domain containing residues
322-376 (Figure 7b), coincident with the area of negative
Table 3: Alignment of C-Terminal Residues of Mammalian CEases
5114 Biochemistry, Vol. 37, No. 15, 1998 Chen et al.
potential (Figure 7a). Because of the charged nature of the
more mobile regions of the molecule, we propose a role for
electrostatic interactions in binding the mixed micelle.
Interactions of CEase with mixed micelles of bile salt and
phospholipids, containing free cholesterol, cholesterol esters,
and triglycerides, may be mediated by the positive charge
on phospholipids such as phosphatidylcholine. As the phos-
pholipids and cholesterol esters are hydrolyzed, changes to
the physical chemical properties of the mixed micelle lead
to dissociation from the protein. This region may thus have
evolved to properly orient the enzyme face relative to the
mixed micelle or guide the micelle toward the active face
of the molecule.
EVolutionary Aspects. Though evolutionarily unrelated,
the mechanism of hydrolysis by CEase is profoundly
homologous to the mechanisms of hydrolysis of esters by
the serine proteases. Therefore, the comparisons of the active
site structures provide another critical assessment of the fea-
tures most important to the chemical mechanisms of hy-
drolysis. The docking of cholesteryl linoleate into the active
center allows for an assessment of contributions that may
pertain to the mechanism of hydrolysis (Figure 8). From
this model, the ester bond can be brought against the γO of
Ser194 and the ?N of His435 in a manner that suggests the
stereochemistry of nucleophilic attack by the γO of Ser194
on the carbonyl carbon of the ester. This model predicts
FIGURE 8: Stereodiagram of proposed Michaelis complex between CEase (black) and cholesteryl linoleate (gray). Active site residues
Ser194, His435, and Asp320 are indicated, and putative oxyanion residues Ala195, Gly107, and Ala108 are in lowercase. The cholesteryl
linoleate molecule spatially overlaps with C-terminal residues 574-579. Docking was done using MIDAS (25). This figure was generated
using MOLSCRIPT (40) and Raster3D (41).
FIGURE 9: Active site hydrogen bond network and proposed mechanism and stereochemistry of the tetrahedral intermediate in the ester
hydrolysis reaction. The NH of Ala195 is in a position to coordinate the oxyanion, and the loop containing Gly107 may move into a
productive binding position to act as a second hydrogen bond donor to the oxyanion.
Structure of Pancreatic Cholesterol Esterase at 1.6 ÅBiochemistry, Vol. 37, No. 15, 1998 5115
that the nucleophilic attack takes place from the opposite
side of the ester, leading to the inversion of chirality of the
tetrahedral intermediate with respect to the trypsin-like serine
proteases (Figure 9). This predicted stereochemical inversion
is consistent with the structures of other triglyceride lipases
in complex with inhibitors (34-37). From our docking, the
hydrogen bonds from the imidazole to the γO of Ser194 and
from the Nδ2 of His435 to the γO of Asp320 are essentially
coplanar, making excellent hydrogen bonded geometry
between them. The geometry of the catalytic triad is ideally
suited for the ?N of His435 to act first as a base for the
serine proton and then as an acid to the oxygen of the
cholesterol leaving group, presenting both oxygens in a
coplanar manner with the plane of the imidazole.
additional similarity is the doubly hydrogen bonded support
by two donors to the other oxygen of Asp320. These have
direct analogues in the serine proteases; however, this oxygen
lies in the opposite relative orientation with respect to the
imidazole in the proteases. The oxyanion in the tetrahedral
intermediate can be readily stabilized by the NH of Ala195;
however, a second hydrogen bond would require the active
site to close over the substrate. A primary candidate for a
second bond is the NH of Gly107, once oriented in the
CEase hydrolyzes both water soluble and hydrophobic
esters, and its structure places it evolutionarily between the
triglyceride lipases and the esterases. The triglyceride lipases
preferentially hydrolyze hydrophobic esters and lipids, while
esterases such as acetylcholinesterase function mainly on
water soluble substrates. For CEase to accommodate larger,
more hydrophobic esters, bile salt is required, in a dual role
as a molecule binding specifically to CEase to open the active
site to these bulkier molecules and as a surfactant to
solubilize the hydrophobic esters, phospholipids, and tri-
glycerides. Therefore, the bile salt dependence of CEase
allows the enzyme to accommodate both hydrophilic and
The structure of the CEase apoenzyme suggests a new set
of structural rearrangements involved in lipase activation.
To date, most triglyceride lipase structures are distinguished
by an amphipathic, helical lid covering the active site
catalytic triad, shielding the hydrophobic binding pocket from
substrate (35, 38). During interfacial activation, the helical
lid swings open upon contact with the hydrophobic interface,
revealing a binding pocket readily accessible to substrate.
There is no amphipathic lid in the structure of CEase; rather,
a broader rearrangement of the tertiary structure is necessary
for its physiological activity, mediated by bile salt and sub-
strate binding. The displacement of the C terminus lodged
in the active site is also necessary for substrate binding, as
this reveals the hydrophobic binding site for the acid portion
of the ester and allows the oxyanion hole to form properly
to bind the reaction intermediate. Furthermore, the bile salt
dependence of enzymatic activity coupled with these unusual
structural features distinguishes CEase as a novel lipase with
an unusual mode of lipase activation.
We thank S. L. LaPorte and Dr. Janet Finer-Moore for
critical comments on the manuscript and Drs. Finer-Moore
and Earl Rutenber for assistance with the structure refine-
1. Hui, D. Y. (1996) Biochim. Biophys. Acta 1303, 1-14.
2. Wang, C. S., and Hartsuck, J. A. (1993) Biochim. Biophys.
Acta 1166, 1-19.
3. Mackay, K., Starr, J. R., Lawn, R. M., and Ellsworth, J. L.
(1997) J. Biol. Chem. 272, 13380-13389.
4. Lopez-Candales, A., Bosner, M. S., Spilburg, C. A., and Lange,
L. G. (1993) Biochemistry 32, 12085-12089.
5. Bosner, M. S., Gulick, T., Riley, D. J. S., Spilburg, C. A.,
and Lange, L. G. (1988) Proc. Natl. Acad. Sci. U.S.A. 85,
6. Bosner, M. S., Gulick, T., Riley, D. J. S., Spilburg, C. A.,
and Lange, L. G. (1989) J. Biol. Chem. 264, 20261-20264.
7. Howles, P. N., Carter, C. P., and Hui, D. Y. (1996) J. Biol.
Chem. 271, 7196-7202.
8. Li, F., Huang, Y., and Hui, D. Y. (1996) Biochemistry 35,
9. Kyger, E. M., Wiegand, R. C., and Lange, L. G. (1989)
Biochem. Biophys. Res. Commun. 164, 1302-1309.
10. Wang, C. S., Dashti, A., Jackson, K. W., Yeh, J. C.,
Cummings, R. D., and Tang, J. (1995) Biochemistry 34,
11. Hansson, L., Blackberg, L., Edlund, M., Lundberg, L.,
Stromqvist, M., and Hernell, O. (1993) J. Biol. Chem. 268,
12. Blackberg, L., Stromqvist, M., Edlund, M., Juneblad, K.,
Lundberg, L., Hansson, L., and Hernell, O. (1995) Eur. J.
Biochem. 228, 817-821.
13. Downs, D., Xu, Y. Y., Tang, J., and Wang, C. S. (1994)
Biochemistry 33, 7979-7985.
14. Sugo, T., Mas, E., Abouakil, N., Endo, T., Escribano, M.-J.,
Kobata, A., and Lombardo, D. (1993) Eur. J. Biochem. 216,
15. Wang, X., Wang, C.-S., Tang, J., Dyda, F., and Zhang, X. C.
(1997) Structure 5, 1209-1218.
16. Chen, J. C.-H., Miercke, L. J. W., Krucinski, J., Starr, J. R.,
Saenz, G., Wang, X., Spilburg, C. A., Lange, L. G., Ellsworth,
J. L., and Stroud, R. M. (1997) FASEB J. 11, A1064.
17. Good, L., and Nazar, R. N. (1992) Nucleic Acids Res. 20,
18. Spilburg, C. A., Cox, D. G., Wang, X., Bernat, B. A., Bosner,
M. S., and Lange, L. G. (1995) Biochemistry 34, 15532-
19. Cox, D. G., Leung, C. K., Kyger, E. M., Spilburg, C. A., and
Lange, L. G. (1990) Biochemistry 29, 3842-3848.
20. Navaza, J. (1994) Acta Crystallogr. A50, 157-163.
21. Sack, J. S. (1988) J. Mol. Graphics 6, 224-225.
22. Brunger, A. T. (1992) X-PLOR: A System for X-ray Crystal-
lography and NMR, Yale University Press, New Haven, CT.
23. Raves, M. L., Harel, M., Pang, Y. P., Silman, I., Kozikowski,
A. P., and Sussman, J. L. (1997) Nat. Struct. Biol. 4, 57-63.
24. Kleywegt, G. J., and Brunger, A. T. (1996) Structure 4, 897-
25. Ferrin, T. E., Huang, C. C., Jarvis, L. E., and Langridge, R.
(1988) J. Mol. Graphics 6, 13-27.
26. Laskowski, R. A., Moss, D. S., and Thornton, J. M. (1993) J.
Mol. Biol. 231, 1049-1067.
27. Hol, W. G. J., van Duijnen, P. T., and Berendsen, H. J. C.
(1978) Nature 273, 443-446.
28. Ghosh, D., Wawrzak, Z., Pletnev, V. Z., Li, N., Kaiser, R.,
Pangborn, W., Jornvall, H., Erman, M., and Duax, W. L.
(1995) Structure 3, 279-288.
29. Baba, T., Downs, D., Jackson, K. W., Tang, J., and Wang, C.
S. (1991) Biochemistry 30, 500-510.
30. Morlock-Fitzpatrick, K. R., and Fisher, E. A. (1995) Proc.
Soc. Exp. Biol. Med. 208, 186-190.
31. Abouakil, N., Mas, E., Bruneau, N., Benajiba, A., and
Lombardo, D. (1993) J. Biol. Chem. 268, 25755-25763.
32. Loomes, K. M., and Senior, H. E. J. (1997) FEBS Lett. 405,
5116 Biochemistry, Vol. 37, No. 15, 1998 Chen et al.
33. DiPersio, L. P., Carter, C. P., and Hui, D. Y. (1994) Bio- Download full-text
chemistry 33, 3442-3448.
34. Brzozowski, A. M., Derewenda, U., Derewenda, Z. S., Dodson,
G. G., Lawson, D. M., Turkenburg, J. P., Bjorkling, F., Huge-
Jensen, B., Patkar, S. A., and Thim, L. (1991) Nature 351,
35. Grochulski, P., Li, Y., Schrag, J. D., Bouthillier, F., Smith,
P., Harrison, D., Rubin, B., and Cygler, M. (1993) J. Biol.
Chem. 268, 12843-12847.
36. van Tilbeurgh, H., Egloff, M.-P., Martinez, C., Rugani, N.,
Verger, R., and Cambillau, C. (1993) Nature 362, 814-
37. Rubin, B. (1994) Nat. Struct. Biol. 1, 568-572.
38. Grochulski, P., Li, Y., Schrag, J. D., and Cygler, M. (1994)
Protein Sci. 3, 82-91.
39. Eisenhaber, F., Lijnzaad, P., Argos, P., Sander, C., and Scharf,
M. (1995) J. Comput. Chem. 16, 273-284.
40. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946-950.
41. Merritt, E. A., and Murphy, M. E. P. (1994) Acta Crystallogr.
42. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M.
(1991) Acta Crystallogr. A47, 110-119.
43. Nicholls, A., Sharp, K. A., and Honig, B. (1991) Proteins 11,
Structure of Pancreatic Cholesterol Esterase at 1.6 ÅBiochemistry, Vol. 37, No. 15, 1998 5117