Molecular Basis for the Recognition and Cleavages of IGF-II, TGF-α, and Amylin by Human Insulin-Degrading Enzyme

Article abstract of DOI:10.1016/j.jmb.2009.10.072

Insulin-degrading enzyme (IDE) is involved in the clearance of many bioactive peptide substrates, including insulin and amyloid-β, peptides vital to the development of diabetes and Alzheimer's disease, respectively. IDE can also rapidly degrade hormones that are held together by intramolecular disulfide bond(s) without their reduction. Furthermore, IDE exhibits a remarkable ability to preferentially degrade structurally similar peptides such as the selective degradation of insulin-like growth factor (IGF)-II and transforming growth factor-α (TGF-α) over IGF-I and epidermal growth factor, respectively. Here, we used high-accuracy mass spectrometry to identify the cleavage sites of human IGF-II, TGF-α, amylin, reduced amylin, and amyloid-β by human IDE. We also determined the structures of human IDE-IGF-II and IDE-TGF-α at 2.3 A and IDE-amylin at 2.9 A. We found that IDE cleaves its substrates at multiple sites in a biased stochastic manner. Furthermore, the presence of a disulfide bond in amylin allows IDE to cut at an additional site in the middle of the peptide (amino acids 18-19). Our amylin-bound IDE structure offers insight into how the structural constraint from a disulfide bond in amylin can alter IDE cleavage sites. Together with NMR structures of amylin and the IGF and epidermal growth factor families, our work also reveals the structural basis of how the high dipole moment of substrates complements the charge distribution of the IDE catalytic chamber for the substrate selectivity. In addition, we show how the ability of substrates to properly anchor their N-terminus to the exosite of IDE and undergo a conformational switch upon binding to the catalytic chamber of IDE can also contribute to the selective degradation of structurally related growth factors.

Full text of DOI:10.1016/j.jmb.2009.10.072

doi:10.1016/j.jmb.2009.10.072
J. Mol. Biol. (2010) 395, 430–443
Available online at www.sciencedirect.com
Molecular Basis for the Recognition and Cleavages
of IGF-II, TGF-α, and Amylin by Human
Insulin-Degrading Enzyme
Qing Guo1, Marika Manolopoulou1, Yao Bian1,
Alexander B. Schilling2 and Wei-Jen Tang1
⁎
1Ben-May Department
for Cancer Research,
The University of Chicago,
929 East 57th Street, Chicago,
IL 60637, USA
2Proteomics and Informatics
Services Facility, University of
Illinois at Chicago, 835 South
Wolcott Avenue, Chicago,
IL 60612, USA
Insulin-degrading enzyme (IDE) is involved in the clearance of many
bioactive peptide substrates, including insulin and amyloid-β, peptides
vital to the development of diabetes and Alzheimer's disease, respectively.
IDE can also rapidly degrade hormones that are held together by intra-
molecular disulfide bond(s) without their reduction. Furthermore, IDE
exhibits a remarkable ability to preferentially degrade structurally similar
peptides such as the selective degradation of insulin-like growth factor
(IGF)-II and transforming growth factor-α (TGF-α) over IGF-I and
epidermal growth factor, respectively. Here, we used high-accuracy mass
spectrometry to identify the cleavage sites of human IGF-II, TGF-α, amylin,
reduced amylin, and amyloid-β by human IDE. We also determined the
structures of human IDE–IGF-II and IDE–TGF-α at 2.3 Å and IDE–amylin at
2.9 Å. We found that IDE cleaves its substrates at multiple sites in a biased
stochastic manner. Furthermore, the presence of a disulfide bond in amylin
allows IDE to cut at an additional site in the middle of the peptide (amino
acids 18–19). Our amylin-bound IDE structure offers insight into how the
structural constraint from a disulfide bond in amylin can alter IDE cleavage
sites. Together with NMR structures of amylin and the IGF and epidermal
growth factor families, our work also reveals the structural basis of how the
high dipole moment of substrates complements the charge distribution of
the IDE catalytic chamber for the substrate selectivity. In addition, we show
how the ability of substrates to properly anchor their N-terminus to the
exosite of IDE and undergo a conformational switch upon binding to the
catalytic chamber of IDE can also contribute to the selective degradation of
structurally related growth factors.
Received 8 June 2009;
received in revised form
26 October 2009;
accepted 27 October 2009
Available online
5 November 2009
© 2009 Elsevier Ltd. All rights reserved.
Keywords: metalloprotease; X-ray crystallography; substrate specificity;
insulin-degrading enzyme; mass spectrometry
Edited by I. Wilson
Introduction
Insulin-degrading enzyme (IDE) is a 110-kDa zinc
*Corresponding author. E-mail address:
wtang@uchicago.edu.
metalloprotease that belongs to the M16A family
(EC 3.4.24.56, insulysin, or insulinase).1,2 Similar to
other M16 family proteases, IDE has a conserved
inverted zinc-binding sequence HEXXH in which
the two histidines are involved in zinc ion coordi-
nation and one glutamate residue is involved in
catalytic water activation.1–4 Recent structural ana-
lysis revealed that human IDE consists of two
roughly equally sized N- and C-terminal domains
(IDE-N and IDE-C).3 IDE-N and IDE-C are homo-
logous to each other and have an αβαβα-roll
Abbreviations used: IDE, insulin-degrading enzyme;
IGF, insulin-like growth factor; IGF-II, insulin-like growth
factor-II; TGF-α, tumor growth factor-α; EGF, epidermal
growth factor; Aβ, amyloid-β; MS, mass spectrometry;
LC, liquid chromatography; TOF, time of flight; Q-TOF,
quadruple time of flight; MALDI, matrix-assisted laser
desorption/ionization; FT-ICR, Fourier transform ion
cyclotron resonance; TCEP, tris(2-carboxyethyl)
phosphine; PDB, Protein Data Bank.
0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.

Substrate Recognition by Human IDE
431
structure, shared among proteases within the M16
family, which include mitochondrial processing
peptidase, pitrilysin, and mitochondrial prese-
quence peptidase.5–7 IDE-N and IDE-C are joined
by an extended 28-aa loop, and together, they form
an enclosed chamber to selectively enclose and
cleave certain peptides. IDE was recently described
as a prototypical cryptidase because of its ability to
use a sizable catalytic chamber (crypt) to selectively
recognize peptides targeted for proteolysis.1
the binding of the exosite with a shorter peptide
substrate could reduce the apparent size of the
catalytic chamber.20 This is postulated to facilitate
the cleavage of short substrates of IDE such as 9-aa-
long bradykinin. However, it remains unclear how
IDE utilizes the properties of the catalytic chamber
and exosite to recognize and selectively degrade
substrates such as IGF-II, TGF-α, and amylin.
IDE exhibits preferential cleavage of the structur-
ally related IGF and epidermal growth factor (EGF)
proteins, peptide hormones that contain three disul-
fide bonds each.1,15–17,22 IGF-I and IGF-II are struc-
turally and functionally related to insulin.23,24
Together with insulin, this IGF family is involved in
regulation of multiple biological processes including
carbohydrate metabolism, growth, life span, devel-
opment, and neoplasia.25,26 NMR analysis reveals
that the structure of IGF-II is similar to that of insulin
and IGF-I.23 Similar to insulin, IGF-II is also a high-
affinity (∼100 nM) substrate of IDE and is rapidly
degraded by this enzyme.1 Paradoxically, IDE binds
IGF-I with reduced affinity compared to IGF-II and
does not effectively degrade this hormone. EGF and
TGF-α are structurally related growth factors that
work in the endocrine, paracrine, and autocrine
systems for various biological activities.27–29 Similar
to the IGF family, IDE also selectively degrades TGF-α
over EGF.17 The sites where IDE cleaves IGF-II and
TGF-α are mostly unknown, as is the molecular basis
of substrate selectivity among members within the
IGF family and the EGF family.
IDE prefers to degrade amyloidogenic peptides
such as amylin, or islet amyloid polypeptide,13 a 37-aa
peptide containing a disulfide bond near the N-
terminus, between cysteine 2 and cysteine 7. Amylin
is normally produced by pancreatic β cells along with
insulin to control glucose homeostasis.30 Similar to
Aβ, amylin can induce cell death of pancreatic β cells
and may thus be involved in the development of
diabetes.30,31 We have previously identified the
cleavage sites of amylin by IDE, as well as solved
the structure of human IDE in complex with amylin in
the presence of a reducing agent.3 The reducing agent
was required for the crystallization of human IDE
protein. However, since a reduced environment
breaks disulfide bonds, the molecular basis for the
interaction of IDE with amylin that retains the
disulfide bond remains elusive.
Since the discovery of IDE based on its ability to
rapidly degrade insulin, IDE was shown to play a
major role in the degradation and clearance of
insulin in vivo.1,2,8,9 This is consistent with the
observation that IDE gene knockout mice have an
elevated level of insulin and develop symptoms of
diabetes, while single nucleotide polymorphisms of
the IDE gene are associated with type 2 diabetes in
humans.10,11 IDE also plays a role in preventing
formation of amyloid deposits by degrading
amyloid-β (Aβ), a key peptide for the development
of Alzheimer's disease.1,10,12–14 Outside of insulin
and Aβ, several short peptides with molecular
masses of 3–10 kDa have been shown to serve as
substrates of IDE, such as insulin-like growth factor
(IGF)-II,15,16 amylin,3 glucagon, transforming growth
factor-α (TGF-α),17 and atrial natriuretic peptide.18
The structural analysis of substrate-bound IDE has
revealed several unique features of how IDE selec-
tively degrades certain bioactive peptides.1–3,19–21
IDE has a catalytic chamber with a total volume of
approximately 16,000 Å3. The unique size of this
chamber allows the enzyme to preferentially cleave
peptides smaller than 80 aa long. In addition, the
complementarity in charge and shape of this chamber
with the substrate also determines the substrate
selectivity as exemplified by the insulin-bound IDE
structure.21 Apart from these unique properties of the
IDE catalytic chamber, IDE also has a highly
conserved exosite located approximately 30 Å away
from the zinc ion bound at the catalytic center.19 The
structures of IDE in complex with Aβ, glucagon,
insulin, and bradykinin reveal that this exosite serves
as an anchoring site for the N-termini of IDE
substrates.3,20,21 Such an interaction is hypothesized
to serve as a molecular ruler, directing IDE to per-
form the initial cleavage at least 10 aa away from the
N-terminal end of its longer substrates.1 In addition,
Fig. 1. Characterization of the degradation of IGF-II and TGF-α by IDE. (a) Primary sequences of human IGF-II, TGF-α,
and insulin. Disulfide bonds are shown by a line connecting two cysteines. Arrows on IGF-II and TGF-α depict the cleavage
sites by IDE from this analysis while the cleavage sites of insulin are as reported previously.21 The underlined amino acids
are observed in the crystal structures of substrate-bound IDE. (b) Comparison of secondary structure (left) and electrostatic
surface representation (right) of IGF-II, TGF-α, and insulin. PDB accession codes of IGF-II, TGF-α, and insulin are 1IGL,
1YUF, and 1G7A, respectively. The molecular surface is colored as calculated by APBS32 (b−6 kT in red, 0 kT in white, and
N+6 kT in blue). (c) Inhibition of the IDE-mediated degradation of the fluorogenic substrate V (a bradykinin-mimetic) by
insulin, TGF-α, and IGF-II. IDE (1 μg) was mixed with substrate V (450 nM) in the presence of indicated concentrations of
insulin, TGF-α, and IGF-II, and fluorescence intensity was monitored for 10 min at 37 °C. Results (means SD) are
representative of three independent experiments performed in duplicate. (d) and (e) show the representative MS/MS
spectrum for N-terminal regions of IGF-II and TGF-α, respectively. The ESI-tandem mass spectrum of the 529.229 ion from
IDE-digested TGF-α at m/z 1584.668 (left) and the 976.469 ion from IDE-digested IGF-II at m/z 1950.936 (right). The amino
acid sequence and ion identification of the tandem mass spectrum are shown, and experimentally observed ions are labeled
in the sequence.

432
Substrate Recognition by Human IDE
IDE's ability to degrade a wide variety of substrates
human body. In order to understand how IDE diffe-
rentiates and degrades IGF-II, TGF-α, and amylin
without breaking their disulfide bonds, we per-
formed structural and biochemical studies on these
IDE substrates. We took advantage of the high-
accuracy mass spectrometry (MS) to identify the
products of IGF-II, TGF-α, and amylin digestion by
IDE. To eliminate the need for a reducing agent
that differ in length, structure, and biological pro-
perties is of great interest in the fields of diabetes and
Alzheimer's research. Determining the mechanism of
how IDE recognizes and degrades certain substrates
over others as well as how it degrades amyloidogenic
peptides is a crucial step in understanding how IDE
affects the homeostasis of these peptides in the
Fig. 1 (legend on previous page)

Substrate Recognition by Human IDE
433
Table 1. Summary of IDE-degraded fragments of IGF-II
and TGF-α
during the crystallization of IDE, we mutated all 13
cysteines in human IDE to construct a fully func-
tional, cysteine-free IDE.20,21 Using this construct, we
determined the structures of human IDE in complex
with three disulfide-containing substrates: IGF-II,
TGF-α, and amylin. Together with the NMR struc-
tures of these three substrates and other peptides
within the same families, new insights regarding how
IDE uses its exosite and catalytic chamber to bind,
unfold, and degrade these disulfide bond-containing
substrates emerge.
Observed (M+H)
Calculated (M+H)
Δppm
IGF-II
1–16
1709.796
1950.936
1140.543
1303.602
1450.670
1209.528
1453.692
3505.657
1297.590
3349.556
1654.769
1725.807
2217.045
2069.976
1709.793
1950.935
1140.539
1303.603
1450.675
1209.533
1453.690
3505.659
1297.591
3349.573
1654.770
1725.809
2217.052
2069.980
1.67
0.42
3.64
0.51
3.49
1.65
1.28
0.57
0.19
5.07
0.11
1.40
3.36
2.13
1–18
17–26
17–27
17–28
19–28
37–48
37–67
38–48
38–67
49–63
49–64
48–67
49–67
Results
Mass spectroscopy analysis of the degradation
TGF-α
1–11
of IGF-II and TGF-α by IDE
1218.497
1584.663
1731.731
1981.809
1179.459
1697.794
1560.735
1097.529
1218.498
1584.668
1731.738
1981.811
1179.458
1697.795
1560.739
1097.534
0.53
3.64
4.04
0.94
0.85
0.59
2.56
4.56
1–14
IGF-II is a 67-aa growth-promoting hormone that
plays a vital role in embryo development and epi-
genetic imprinting. TGF-α is a 50-aa-long peptide
that can induce epithelial cell development. These
two peptides differ significantly in their primary
sequences, secondary structure, shape, and surface
charge distribution (Fig. 1a and b).33–35 However,
like insulin, these two peptides are high-potency
inhibitors (60–180 nM, Fig. 1c) for the degradation of
a fluorogenic bradykinin-mimetic peptide, substrate
V by human IDE. The IC50 values are in agreement
with the reported Ki values.4,16,17,22 Thus, IDE binds
IGF-II and TGF-α, two structurally diverse disulfide-
bond-containing peptides, with high affinity com-
parable to that of insulin.
IDE can effectively cleave both IGF-II and TGF-α
but little is known regarding cleavage sites for each
substrate. Both IGF-II and TGF-α are stabilized by
three intramolecular disulfide bonds. IDE is known
to degrade its substrates without the assistance of
disulfide bond isomerase activity, and its proteolytic
activity does not require the breakage of any
disulfide bonds. To decipher the cleavage sites of
IGF-II and TGF-α, we used liquid chromatography
coupled with quadruple time-of-flight (LC-Q-TOF)
MS. We identified 14 IGF-II fragments and 8 TGF-α
fragments that were generated by 5-min incubations
with IDE followed by treatment with a reducing
agent, DTT (Table 1). The identification of these
fragments is based on the accurate match of the
observed mass with the predicted mass of the
precursor and b/y product ions. Figure 1d and e
show sample mass spectra obtained for IGF-II and
TGF-α, respectively. We also performed matrix-
assisted laser desorption/ionization (MALDI)-TOF
tandem MS to gain better MS/MS coverage of these
precursor ions and confirm the identity of these
fragments (Table S1).
1–15
1–17
12–21
35–50
36–50
41–50
Assignment of the most prominent peaks generated after 5-min
incubation with IDE using Q-TOF MS/MS.
amino acids 63–64/64–65, and three pairs of cut sites
on TGF-α, amino acids 14–15/15–16, amino acids
20–21/21–22, and amino acids 34–35/35–36. In one
case, we also observed three cleavage sites that are
adjacent to each other (amino acids 26–27, 27–28,
and 28–29) on IGF-II. This is consistent with the
probabilistic mode of binding of these regions to the
catalytic cleft for cleavage. Such biased stochastic
mode is a commonly found feature of IDE substrate
cleavage sites.1,3,4,21 We did not observe cleavages
near the N-terminus of either IGF-II or TGF-α
(Fig. 1a). This is consistent with the notion that
IDE uses the exosite to anchor the N-termini of both
substrates; thus, no cleavage can occur at the first
10 aa residues of the N-terminal end. Interestingly,
the cleavage sites of TGF-α defined using the
accurate mass determination differ significantly
from those determined by the blockage of Edman
degradation of rat IDE-degraded [125I]TGF-α.36
Such a discrepancy may be due to the methodology
and/or source of IDE.
MS analysis of the degradation of amylin and
Aβ(1–40) by IDE
Previously, we used MALDI-TOF MS to assess the
cleavage sites of amylin and Aβ(1–40) by IDE.3 To
more accurately determine the cleavage sites, we
used Fourier transform ion cyclotron resonance (FT-
ICR) MS for its high mass accuracy and sensitivity
(Table 2 and Table S2). Based on the excellent match
of the precursor ions to the predicted mass as well as
the extensive coverage of b and y product ions from
each precursor ion, we identified seven major
Similar to the cleavage pattern of insulin by IDE,
we found that IDE preferentially cleaves several
discrete regions of IGF-II and TGF-α (Fig. 1a).
Several of these sites are clustered together. These
include three pairs of cleavage sites on IGF-II, amino
acids 36–37/37–38, amino acids 47–48/48–49, and

434
Substrate Recognition by Human IDE
Table 2. Ion identification of major IDE-degraded fragments of amylin and Aβ(1–40)
Observed (M+H)
Calculated (M+H)
Δppm
Observed b/y ionsa
Amylinb
1–15 (C–C)c
16–27
1639.7677
1271.6722
1358.7054
1554.7352
1152.6000
1367.6867
1639.7873
1271.6936
1358.7257
1154.7588
1152.6089
1367.6995
13.2
16.9
14.9
20.5
7.7
b7–b13, y5–y6
b3–b5, b7–b11, y7–y9
b3, b5–b11, y6–y7, y9–y10
b4–b14, y6, y11–y13
b2–b11, y4–y10
16–28
19–34
23–34
23–36
9.3
b3–b6, b8–b13, y2–y11
Amylinb (+TCEP)
16–27
1271.6741
1358.7127
2281.1128
1152.6000
1367.6866
1271.6936
1358.7257
2281.1207
1152.6089
1367.6995
15.3
9.6
3.5
7.7
9.4
b2–b5, b7–b11, y5, y7–y9
b3–b11, y6–y7, y11
16–28
16–37
b5, b7–b10, y5–y10, y12–y17
b3–b11, y6–y7, y9
23–34
23–36
b3–b13, y2–y11
Aβ(1–40)d
19–40
21–40
20–40
1–28
2180.1456
1886.0096
2033.0727
3261.5354
2648.4593
2785.5235
2180.1420
1886.0052
2033. 0736
3261.5328
2648.4480
2785.5069
1.6
2.3
0.4
0.8
4.3
6.0
b5–b6, b9, b11–b18, y6–y10, y16–18
b9–b17, y5–y9, y13, y16
b9, b11–b17, y5–y10, y12–y14, y16–y17
b5–b8, b10–b13, y4–y6, y8–y13
15–40
14–40
b6, b8–b10, b17–b18, y7–y11, y13, y16–y18
b5–b12, b14–b15, b17–b18, y8–y10, y13, y16–y20
a
The b and y product ions of the most prominent fragment peaks of IDE-degraded amylin, reduced amylin, and Aβ(1–40) are
generated by LC-ESI-FT-ICR-MS-CID-MS/MS, and their identity is analyzed using MassMatrix webserver (http://searcher.rrc.uic.edu/
cgi-bin/mm-cgi/search_form.py).
b
Five-minute incubation of amylin with IDE in 50:1 ratio.
c
This fragment has an intact disulfide bond.
d
One- to five-second incubation of Aβ(1–40) with IDE.
cleavage sites when the non-reduced amylin was
used (Table 2 and Fig. 2c). We also performed the
cleavage reaction in the presence of a reducing
agent, tris(2-carboxyethyl)phosphine (TCEP), to
assess whether the presence of disulfide bond in
amylin affects the cleavage site of amylin by IDE.
While six out of seven amylin cleavage sites were
also cut by IDE when amylin was reduced by TCEP,
we did not observe the cleavage between residues 18
and 19, which is located in the middle of amylin
(Table 2; Table S2; Fig. 2c). This indicates that the
presence of disulfide bond can alter the cleavage site
of amylin by IDE.
We have analyzed the initial cleavage sites of
insulin by IDE to gain the mechanistic under-
standing of insulin degradation by IDE.21 While
IDE is known to cleave Aβ(1–40) at multiple sites,
the preferred initial cleavage sites have not been
defined yet.3,37 To address this, we performed the
digestion of Aβ(1–40) by IDE by varying the time of
incubation and then assessed the cleavage products
by FT-ICR MS/MS (Table 2). As expected, IDE
cleaves Aβ(1–40) at multiple sites even only at brief
incubation (1–5 s). Our MS analysis reveals the
preferred initial cleavage sites of Aβ(1–40) by IDE
to be in the middle of Aβ(1–40) (amino acids 18–19,
19–20, and 20–21). The cleavages at amino acids
14–15, amino acids 15–16, and amino acids 28–29
occurred less frequently (Table 2; Fig. 2c). These data
are consistent with our proposed yet unproven
model that, upon the anchoring of N-terminus of
Aβ to the IDE exosite, the middle segment of Aβ is
preferably cut in a probabilistic manner for its initial
cleavage, which would prevent the amyloid fiber
formation by Aβ.1
Structures of amylin-, Aβ(1–42)-, IGF-II-, and
TGF-α-bound IDE
Previous structural studies of IDE with various
substrates have required the use of a reducing agent
in the crystallization conditions, due to the presence
of 13 cysteine residues in human IDE.3 This reduced
environment also breaks any disulfide bond(s) in the
substrates, resulting in IDE–substrate complexes
that may not reflect the native enzyme–substrate
binding conformation. We have recently con-
structed a catalytically inactive, cysteine-free IDE
(IDE-CF-E111Q) that can be crystallized without the
reducing agent.1,21 Using this mutant, we re-
examined the structural basis of the interactions of
IDE with amylin, a substrate containing one
disulfide bond. We have solved the crystal structure
of IDE-CF-E111Q in complex with amylin at 2.9 Å
resolution (Fig. 2a; Table 2). While we were unable
to observe the disulfide bond in the previous IDE–
amylin structure,3 the current structure shows a
density that fits well with residues K1–N3, C7–A8,
and L12–L16, with a disulfide bond present between
C2 and C7 (Fig. 2b). However, we could not see
density for the majority of the loop formed by amino
acids 4–6 of amylin. Thus, we do observe two discreet
segments of amylin, an N-terminal portion in the
exosite and another segment in catalytic site of the
enzyme; this spatial separation of the substrate is a
common feature in IDE–substrate structures. It is
worth noting that, similar to the structures of pre-
vious IDE–substrate complex, the averaged thermal
B-factors of the amylin is higher than that of IDE
(Table 3).3,21 This is presumably due to the structural
heterogeneity of the substrates and/or less than 100%

Substrate Recognition by Human IDE
435
Fig. 2. Structures of IDE in complex with amylin. (a) Global view of the structure of amylin-bound IDECF-E111Q
monomer. IDE-N and IDE-C are colored green and cyan, respectively. Amylin is in the stick representation colored
orange. (b) Stereo view of composite omit map (purple) of the IDE-bound amylin is contoured at 1.5 σ. Oxygen, nitrogen,
and carbon atoms of amylin are shown in red, blue, and orange, respectively. (c) Primary sequences of amylin, reduced
amylin, and Aβ(1–40). The arrows on the amylin depict the cleavage sites by IDE as experimented and the arrows on
Aβ(1–40) depict the cleavage sites by IDE after short (1–5 s) incubation. The minor cleavage sites on Aβ(1–40) are marked
with the shorter arrows. The underlined amino acids with the scissor bond marked by a red arrow are observed in the
crystal structures of substrate-bound IDE. (d) Detailed interaction of the N-terminus of non-reduced amylin (left), reduced
amylin (middle), and Aβ(1–42) (right) with the exosite of IDE. (e) Detailed interaction of amylin with the IDE catalytic site.
The color of IDE residues corresponds to the respective color of IDE domain in (a). The PDB codes for structures of amylin-
bound IDE-CF-E111Q, reduced amylin-bound IDE-E111Q, and Aβ(1–42)-bound IDE-CF-E111Q are 3HGZ, 2G48, and
2WK3, respectively.
occupancy (despite the extensive effort in loading
amylin to IDE in order to ensure the high occupancy,
see Methods). The same phenomenon was observed
in the structures of IDE bound to Aβ(1–42), IGF-II,
and TGF-α described below (Table 3).
Amylin and Aβ, another IDE substrate, are both
amyloidogenic peptides and both can be effectively
cleaved at multiple locations (Fig. 2c). The structure
of Aβ(1–40) in complex with IDE has previously
been solved, as have the cleavage sites for amylin
and Aβ by IDE.3 Aβ(1–42) is a form of Aβ that forms
amyloid plaque much faster than Aβ(1–40) and
serves as a better marker for the progression of
Alzheimer's disease. To learn whether IDE recog-
nizes these two peptides differently, we also solved
the structure of IDE in complex with Aβ(1–42)
(Table 2). We found the binding of Aβ(1–42) to the
exosite and catalytic site of IDE in this structure to be
virtually identical with that of Aβ(1–40) in our
Aβ(1–40)-bound IDE structure. Compared to the
Aβ-bound IDE structure, we found a novel inter-
action of amylin with the exosite of IDE (Fig. 2d).
The disulfide bond formed by amylin C2 and C7
changes the local hydrogen bond network to the IDE
β12-strand compared to Aβ (Fig. 2d). Most notice-
ably, the side chain of the N-terminus of amylin (K1)
forms a salt bridge with that of IDE E341. This is
different from other substrates of IDE, where the
amino-terminus of the substrates (i.e., Aβ) forms
hydrogen bonds with the side chain of E341 and the
main chain of L359.
The structure of amylin at the catalytic site of IDE
shows that amylin is embedded into several
hydrophobic pockets, and amylin residues F15 and

436
Substrate Recognition by Human IDE
L16, positioned nearby the catalytic metal ion, are
coordinated by H108, H112, and E189 of IDE
(Fig. 2e). These hydrophobic interactions include
the contacts of amylin L12 with IDE F202 and W199,
that of amylin F15 with IDE F141, and that of amylin
L16 with IDE F115. This corresponds to the bio-
chemical data available for amylin, which indicate
that the cleavage between residues 15 and 16 is
indeed one of the major cleavage sites (Fig. 2c and e).3
The crystal structures of IDE-CF-E111Q com-
plexed with IGF-II and TGF-α were determined at
2.3 Å resolution to further understand how IDE
binds different substrates (Table 2). The overall fold
of IDE in both cases is nearly identical with amylin-
bound IDE (Fig. 3a), with two discrete substrate
segments clearly visible: one at the exosite and the
other at the catalytic site of the respective structures
(Fig. 3b). Both main chains and side chains of the
first 5 aa of IGF-II are visible, while only the main
chains of the first 3 aa of TGF-α are visible (Fig. 3c).
At the catalytic site, several residues of IDE located
in domains 1 and 4 form a largely polar cavity with
hydrophobic patches that interact with cleavage
sites in both substrates (Fig. 3d). At the catalytic site
of IDE, IGF-II amino acids 13–18 and TGF-α amino
acids 10–14 fit nicely into the electron density map,
and the observed primary cleavage sites of these two
substrates (T16–L17 of IGF-II and S11–H12 of TGF-α)
appear poised for cleavage (Fig. 3d).
Discussion
IDE cleaves multiple substrates that have various
biological functions, including amylin, Aβ, IGF-II,
and TGF-α. These substrates are diverse in their
sequence, size, and structure. Based on accumulated
structural and biochemical evidence, it has been
proposed that the catalytic cycle and substrate
specificity of IDE revolve around a conformational
switch between the open and closed states of this
enzyme, with interactions between potential sub-
strates and the IDE catalytic chamber stabilized by
complementary electrostatic properties (Fig. 4).1,3
We propose that IDE normally exists in an equilib-
rium between an open (IDEO) and a closed (IDEC)
state.3 In the open state, the negatively charged
interior of the IDE-N domain and the positively
Table 3. Data collection and structure refinement statistics of substrate-bound IDE
IDE–IGF-II
IDE–TGF-α
IDE–amylin
IDE-Aβ(1–42)a
Data collection
Beamline
Space group
Cell dimension (Å)
a
b
c
APS 19ID
APS 19ID
APS 19ID
APS 19ID
P65
P65
P65
P65
263.0
263.0
90.8
50–2.28
13.4 (50.7)
20.5 (2.1)
5.4 (3.1)
99.7 (96.9)
162,414
262.2
262.2
90.5
50–2.3
9.4 (52.0)
23.6 (3.6)
7.1 (6.8)
99.8 (100.0)
157,230
262.9
262.9
90.9
50–2.9
8.0 (33.2)
25.1 (3.8)
5.7 (4.3)
99.9 (99.2)
79,043
261.6
261.6
90.7
50–2.6
9.1 (45.3)
17.6 (3.0)
3.6 (3.4)
99.8 (99.9)
109,059
Resolution (Å)
Rsym (%)b
I/σ
Redundancyc
Completeness (%)
Unique reflections
Refinement
d
Rwork
0.206
0.239
0.191
0.231
0.178
0.223
0.186
0.232
e
Rfree
No. of atoms
Protein
15,710
598
15,644
653
15,818
346
15,590
233
Water
B-factors
IDE
35.5
54.2
44.5
28.6
52.6
46.5
38.6
67.8
48.5
35.2
71.2
36.5
Substrate
Water
r.m.s.d.
Bond lengths (Å)
Bond angles (°)
Ramachandran plot (%)f
Favorable region
Allowed region
Generously allowed region
Disallowed region
PDB code
0.015
1.363
0.016
1.463
0.012
1.406
0.022
1.959
91.9/91.6
8.1/8.4
0/0
0/0
3E4Z
92.3/91.2
7.3/8.8
0/0
0/0
3E50
89.6/90.0
10.4/10.0
0/0
0/0
3HGZ
90.7/100.0
9.3/0.0
0/0
0/0
2WK3
The outer resolution shell. Values in parentheses indicate the highest-resolution shell.
a
The backbone nitrogen of Gly29 of Aβ(1–42) is substituted with 2-nitrobenzyl group to improve the solubility of Aβ(1–42) (Erik
Johnson, University of Chicago, PhD thesis).
b
Rmerge =∑(I−〈I〉)/∑〈I〉.
c
Nobs/Nunique
.
d
e
f
Rwork =∑hkl
Fo −k Fc
/∑hkl Fo .
|| | | ||
| |
Rfree, calculated the same as for Rwork but on the 5% data excluded from the refinement calculation.
Values for IDE protein/substrate.

Substrate Recognition by Human IDE
437
Fig. 3. Structures of IDE in complex with IGF-II and TGF-α. (a) Global view of the structure of IGF-II-bound (top) and
TGF-α-bound (bottom) IDE-CF-E111Q monomer. IDE-N and IDE-C are colored green and cyan as Fig. 1. IGF-II and TGF-α
are colored orange. (b) Composite omit maps (blue) of IGF-II and TGF-α are contoured at 1.5 σ. The substrates are colored
orange. (c) The detailed interaction of the N-terminus of IGF-II and TGF-α with the exosite site of IDE. (d) Detailed
interaction of IDE catalytic chamber with IGF-II and TGF-α. Oxygen, nitrogen, and carbon atoms of substrates are shown in
red, blue, and orange, respectively.
charged interior of the IDE-C domain result in
selective binding of ligands with charge distribu-
tions complementary to the electrostatic properties
of the chamber (Fig. 4a). IDE substrates IGF-II, TGF-
α, and amylin also exhibit such a charge distribution
(Fig. 4b). Modeling the N-terminal end of IGF-II,
TGF-α, or amylin in the exosite of IDE reveals charge
complementarity between the substrate and the IDE
chamber, as the negatively charged surfaces of IGF-II,
TGF-α, and amylin fit well with the positively
charged chamber wall of IDE-C.
The interactions between the substrate and the IDE
chamber would appropriately position the flexible
N-terminus of the substrate to bind at the exosite
(Fig. 4a). The binding of the substrate N-terminus to
the IDE exosite and/or the charge complementarity
of the catalytic chamber of IDE with the substrate
would facilitate the switch of the open conformation
to the closed conformation, which is the preferred
state of IDE.1,3 By comparing the native NMR
structures of IGF-II, TGF-α, and amylin with those
in the IDE-bound structures, we find that IDE
binding also results in a substantial substrate con-
formational change (Fig. 4c and d). The native con-
formation of the substrates does not position the
putative cleavage sites close enough to the zinc-
coordinated catalytic center of IDE, suggesting that a
change in secondary structure must occur prior to
catalysis. For three peptides, the most noticeable
change was the conversion of IGF-II amino acids
13–18, TGF-α amino acids 10–14, and amylin amino
acids 16–23 from α-helices to β-strands. These
required conformational changes for IGF-II, TGF-α,
and amylin are similar to those found in IDE in
complex with Aβ and glucagons.3 In addition, while
IGF-II and amylin fit well in the catalytic cavity, the
native conformation of TGF-α has an elongated
shape and would extend beyond the walls of the
chamber, potentially interfering with the switch
from IDEO to IDEC (Fig. 4b). Since IDE does cleave
TGF-α, this offers further evidence that an induced-
fit structural rearrangement of substrates would be
required for binding and cleavage within the
catalytic chamber of IDE. While speculative, struc-
tural rearrangement of the substrate, at least in some
cases, must occur prior to the formation of IDEC.
The structural rearrangement would allow certain
sections of IDE substrates to fit into the catalytic
cleft of IDE for cleavage in a biased stochastic
manner. This is consistent with the multiple
cleavage sites found by our MS analysis on the
degradation products of IGF-II, TGF-α, and amylin
as well as the previous analysis of cleavage sites
and products of IDE substrates. With the anchoring

438
Substrate Recognition by Human IDE
Fig. 4. Analysis of IDE's binding with IGF-II, TGF-α, and amylin. (a) A model of how IDE binds, unfolds, and
degrades its substrate. IDE has two conformational states: the open state, IDEO, and the closed state, IDEC. The IDE-N and
IDE-C are depicted as green and cyan, respectively. IGF-II (PDB code: 1IGL) is depicted as red cartoon. IDEO is
theoretically modeled based on the substrate free E. coli pitrilysin structure (1Q21); IDEC corresponds to the atomic
coordinate of IDE-CF-E111Q-IGFII (3E4Z). The detailed description of the mechanism is in Discussion. (b) Shape and
surface charge distribution of IGF-II (left), TGF-α (middle), and amylin (right) modeled in the catalytic chamber of IDE. (c)
Comparison of IGF-II (left), TGF-α (middle), and amylin (right) in their free forms (transparent gray) with IDE-bound
forms (red). The segments in the free forms of IGF-II and TGF-α corresponding to IDE-bound forms are colored
transparent red. The arrows indicate the cleavage sites, and the disulfide bonds are colored yellow. (d) Comparison of
IDE-bound IGF-II (red), TGF-α (green), amylin (blue), and Aβ(1–42) (orange) in stick model.
of the N-terminus of substrates to the exosite that is
30 Å away from the catalytic center, the given
section would likely to be at least 10 residues away
from its N-terminus for the initial cleavage. Indeed,
we found that the preferred initial cleavages of
Aβ(1–40) occur in the middle of the peptide (amino
acids 18–21). In addition, this anchor-dependent
cleavage could also explain our structural and MS
observations that the presence of a disulfide bond
in amylin could constrain and alter the binding of
amylin to the catalytic chamber of IDE so that an
additional cleavage site between residues 18 and 19
of amylin can be cut. The initial cleavage can induce
further conformational changes in the substrate,
allowing for further cleavage inside the catalytic
chamber. Such processive cleavages for the initial

Substrate Recognition by Human IDE
439
digestion of insulin allows IDE to cut it into two
pieces and inactivate its function without breaking
the disulfide bonds of insulin.21 Ultimately, the
cleavage could induce IDE to switch to the open
conformation, leading to release of the proteolytic
fragments and allowing for incorporation of other
substrates or intermediate degraded fragments to
start a new degradation cycle.
Our structures, together with the secondary
structures and amino acid composition of the IGF
and EGF family hormones, reveal that charge
complementarity and exosite anchoring are used
by IDE to selectively degrade certain members of
those hormone families. IGF-II and IGF-I share 64%
sequence identity while TGF-α and EGF share 36%
sequence identity (Fig. 5a). Accordingly, their
structures are highly homologous within each family
(Fig. 5b). The r.m.s.d. of TGF/EGF is 2.536 Å, while
that of IGF-I/IGF-II is 1.298 Å. However, the surface
charge distributions between peptides in the same
family differ. TGF-α has a much higher dipole
moment (200 Debyes) than that of EGF (90 Debyes)
and, compared to EGF, has a better charge comple-
mentarity for the catalytic chamber of IDE (Fig. 5b).
IGF-II has a clear dipole charge distribution
(330 Debyes) that could complement the charged
catalytic chamber of IDE (Fig. 5b).38 On the other
hand, despite having only a slightly lower dipole
moment than IGF-II (300 Debyes, Fig. 5c), the surface
charge distribution of IGF-I does not coordinate well
with the IDE chamber. Thus, within this growth
family, the dipole moment and charge complemen-
tarity could be used as the parameters to predict
susceptibility to IDE.
Currently, the role played by the specific substrate
sequences in determining IDE selectivity is unclear.
However, it is worth noting that IGF-I residues R55
and R56 would form a positively charged region
that would contact the positively charged chamber
of IDE-C, decreasing the binding affinity of IGF-I.
The corresponding residues in IGF-II, A55 and L56,
are nonpolar. The same situation can be observed
from the comparison of TGF-α and EGF. EGF
contains positively charged R45 and R48, which
Fig. 5. Structural comparison of IGF, EGF, and amyloidogenic peptides. (a) Sequence alignment of the IGF-II/IGF-I
and TGF-α/EGF. (b) Comparison of the electrostatic surface of IGF-II (1IGL), IGF-I (3GF1), TGF-α (1YUF), EGF (1EPH),
amylin (2KB8), and Aβ(1–40) (1AML). The molecular surface is colored as calculated by APBS32 (b−6 kT in red, 0 kT in
white, and N+6 kT in blue). The dipole moment of these peptides was the average from distinct NMR solution structures
[10 for IGF-I, 20 for IGF-II, 16 for TGF-α, 10 for EGF, 20 for Aβ(1–40), and 30 for amylin].38 (c) NMR solution structures of
IGF-II, IGF-I, TGF-α, EGF, amylin, and Aβ(1–40).

440
Substrate Recognition by Human IDE
would contact IDE-C and can interfere with the first
step in IDE–substrate binding as outlined in our
model.
Methods
Protein expression and purification
Amylin and Aβ are both amyloidogenic peptides,
exhibit bipolar surface charge distributions, and
share 36% sequence similarity. Comparison of NMR
structures of these peptides reveals that amylin has a
high dipole moment (500 Debyes), allowing it to
complement well with the charge distribution of the
catalytic chamber of IDE (Fig. 5b and c). However,
such charge complementarity is not observed for
Aβ. Furthermore, these two peptides are much
smaller compared to the IGF and EGF family
peptides that they are likely to have more flexibility
in their interaction with the IDE chamber. Thus, it is
difficult to assess the relative contribution of charge
distribution and conformational flexibility for the
binding and degradation of Aβ and amylin by IDE.
A common feature between the amylin-, IGF-II-,
and TGF-α-bound IDE structures is the interaction
of the substrate's N-terminus with the IDE exosite.
Indeed, this feature has also been observed in our
previously reported IDE–substrate structures, indi-
cating that this is an important interaction for IDE–
substrate binding. In comparing the NMR structures
of IGF-II and IGF-I, we see that both have a
relatively stable helix domain in the middle, while
the conformations of the N- and C-termini vary
greatly (Fig. 5c). A significant difference can be seen
at the N-terminus of IGF-II, which is longer and
more flexible than that of IGF-I. A longer and more
flexible N-terminus can also be observed for TGF-α
compared to EGF and Aβ compared to amylin.
We hypothesize that a longer and more flexible
N-terminus increases the possibility that the substrate
will make the appropriate contacts for anchoring to
the IDE exosite. In addition to the charge repulsion
from the positive patches found in IGF-I and EGF, the
failure of the N-termini of these growth factors to
reach the exosite of IDE would also help to explain
why IDE preferentially degrades IGF-II over IGF-I
and TGF-α over EGF. Differences in the N-terminus
may also explain why Aβ has a higher binding
affinity than amylin. However, their structural
dissimilarity and the bulkiness of the loop formed
by the disulfide bond of amylin may also play a role
in determining binding affinity.
The expression vectors for cysteine-free human IDE
(IDE-CF; C110L, C171S, C178A, C257V, C414L, C573N,
C590S, C789S, C812A, C819A, C904S, C966N, and C974A)
and the catalytically inactive IDE-CF-E111Q mutants were
created as described previously.21 Wild-type human IDE
and IDE mutants were expressed in Escherichia coli Rosetta
(DE3) cells (at 25 °C and 19 h, IPTG induction) and
purified by Ni-NTA, Source-Q, and Superdex S-200
columns as described previously.21 Insulin was purchased
from RayBiotech, IGF-II and TGF-α were from Peprotech,
and amylin amide was from Bachem.
MS analysis for TGF-α and IGF-II
Enzyme reactions were carried out at 37 °C by mixing
5 μl of 1 mg/ml TGF-α and IGF-II in 20 mM Hepes buffer
(pH 7.2) with 5 μl IDE protein in an enzyme-to-substrate
molar ratio of 1:50. Reactions were stopped by the
addition of 30 μl stop solution (170 mM ethylenediamine-
tetraacetic acid and 0.07% trifluoroacetic acid). For
MALDI-TOF MS, mass spectra were obtained in either
linear or positive reflector mode using a Voyager 4700
MALDI-TOF mass spectrometer (Applied Biosystems).
For Q-TOF-MS2, a 0.3-μl aliquot of IDE-digested IGF-II
and TGF-α was first incubated with 300 mM DTT for
30 min and then injected onto an Agilent 1100 LC system
attached to an Agilent 6520 Q-TOF-MS equipped with a
Chip-Cube nanoflow interface. A chip packed with
Zorbax C18SB (75 μm internal diameter×43 mm length,
Agilent Technologies G4240-62001) was used to trap and
elute the peptides, which were run using a gradient as
follows: 0.1% formic acid in water is held at 100% for 5 min
and then changed to 45% of a 10:90 mix of 0.1% formic
acid in water:acetonitrile over 20 min and then changed to
75% of this mix over 10 min. The flow rate of the
chromatography segment of the chip is 200 nl/min and
that of the enrichment (trapping) segment is 3 μl/min.
Enrichment is conducted for 5 min at 100% 0.1% formic
acid in water. The Q-TOF-MS is run with the following
parameters: MS1 scan range, 300–2000 m/z; MS/MS scan
range, 50–3000 m/z; MS1 scan rate, 8.4 scans/s; MS/MS
scan rate, 2 scans/s. Positive ion mode was used
throughout the run with a source Vcap of −1875 V, a
fragmentor voltage of 175 V, a drying gas temperature of
325 °C, and a gas flow rate of 5 L/min. For MS/MS, an
isolation width of 4 amu was used to filter selected
precursor ions into the collision cell, which had a ramped
collision energy using a slope of 3 V/100 amu and an
offset of 2 V. All data were collected as centroided to
conserve disk space, and a reference compound simulta-
neously infused into the spray chamber with an m/z of
1221.99 was used to automatically recalibrate each
spectrum as it was recorded.
Our biochemical assays and structures expand the
repertoire of substrates that IDE is known to degrade
and our knowledge of the IDE–substrate binding
mechanism. Our comparative analyses of the IGF
and EGF hormone families revealed that charge dis-
tribution and flexibility of the substrate N-terminus
are important factors in IDE–substrate recognition
and cleavage. Furthermore, proper placement of
substrates in relation to the IDE catalytic site requires
the requisite structural change of the substrates.
Additional work, such as mutagenesis studies and
N-terminal truncation/elongation, will be needed to
verify our hypotheses. However, this study provides
key information for the engineering of substrates
with altered binding affinities to IDE for therapeutic
applications.
The data acquired this way were subsequently qualita-
tively analyzed by using the molecular feature extractor
tool built into the MassHunter (Agilent Technologies
version B.03), resulting in the detection of all chromato-
graphed peptides listed by monoisotopic mass. The results
of these extractions were exported to Microsoft Excel.
Masses detected were compared with lists of possible
peptides derived from subsequences of the original
protein subject to a minimum mass agreement of 5 ppm
between detected peptides and predicted peptides.

Substrate Recognition by Human IDE
441
MS analysis for amylin and reduced amylin
of human IGF-II or TGF-α (Peprotech). The reactions were
initiated by the addition of 5 μl of 0.2 mg/ml IDE protein.
The substrate V degradation was assessed by monitoring
fluorescence intensity for 10 min every 20 s on a Tecan
Safire2 microplate reader (lex =327 nm, lem =395 nm).
Enzyme reactions were carried out at 37 °C by mixing
5 μl of 1 mg/ml amylin in 20 mM Hepes buffer (pH 7.2)
with 5 μl IDE protein in an enzyme-to-substrate molar
ratio of 1:50 with or without 10 mM TCEP added.
Reactions were stopped after 5-min incubation by the
addition of 30 μl stop solution (170 mM ethylenediamine-
tetraacetic acid and 0.07% trifluoroacetic acid). The
digested samples were then analyzed using FT-ICR MS.
To do so, we injected amylin samples (12 μl) into a nano
RP-HPLC system (Dionex), with a C18 analytical column
(Agilent). Peptides were eluted from the nano column
with a linear gradient of 5–95% acetonitrile in 0.1% formic
acid and sprayed into an LTQ-FT tandem MS instrument
(Thermo Scientific). Spectra were acquired using positive
ion nano ESI mode with the FT-ICR acquiring precursor
spectra from 200 to 2000 m/z. For tandem MS, precursor
ions were fragmented by collision-induced dissociation.
MS/MS spectra were acquired in a data-dependent
manner from the five most intense precursor ions of
each FT-ICR MS scan. The RAW data files are processed
by Xtract™ function in Xcalibur™ (Thermo Fisher
Scientific) to generate reduced data files containing the
deconvoluted masses and intensities for MS spectra.
Accession codes
The coordinates for IGF-II-, TGF-α-, amylin-, and
Aβ(1–42)-bound IDE have been deposited in Research Col-
laboratory for Structural Bioinformatics PDB under acces-
sion codes 3E4Z, 3E50, 3HGZ, and 2WK3, respectively.
Acknowledgements
We are grateful to the staff of the Structural
Biology Center and BioCars at the Advanced
Photon Source, Argonne National Laboratory, for
help in data collection; Erik Johnson for the
synthetic Aβ(1–42) peptide; and Ray Hulse for the
technical assistance. This work was supported by
National Institutes of Health grant GM81539 to W.-J.
Tang. Use of the Advanced Photon Source was
supported by the U.S. Department of Energy, Office
of Basic Energy Sciences, under contract no. W-31-
109-ENG-38. Use of the proteomics and informatics
services facility was supported by the Chicago
Biomedical Consortium.
Protein crystallization and structure determination
IDE-CF-E111Q in complex with IGF-II, TGF-α, amylin,
or Aβ(1–42) was made after five cycles of protein–
substrate complex formation and separation by gel
filtration to ensure the high occupancy as described
previously.3,21 The substrate-bound IDE was crystallized
by hanging drop vapor diffusion at 18 °C, using 1 μl of
protein (16–20 mg/ml) and 1 μl of mother liquor (10–13%
polyethylene glycol monomethyl ether 5000, 100 mM
Hepes, pH 7.0, 4–14% Tacsimate, and 10% dioxane).
Clusters of needle crystals appeared in 3 to 5 days and
were equilibrated in cryo-protective buffer containing 30%
glycerol and mother liquor and were flash frozen in liquid
nitrogen. Diffraction data were collected at 100 K at the
Advance Photon Source 14-BM-C and 19-ID beamlines at
Argonne National Laboratory. The data sets were pro-
cessed using HKL2000.39 The structures were solved by
molecular replacement (Phaser)40 using the IDE portion of
Aβ-bound IDE-E111Q structure as a search model
[Protein Data Bank (PDB) code: 2G473]. Structure refine-
ment and rebuilding were performed using REFMAC and
Coot.39,41 The extra electron density at the catalytic
chamber of IDE in the structures of IDE in complex with
TGF-α, IGF-II, amylin, and Aβ(1–42) were clearly visible
based on σA-weighted Fo −Fc map calculated by CNS42
and manually built. The refinement statistics are summa-
rized in Table 2. Figures were generated using PyMOL43
and CCP4MG.40 The volumes of the catalytic chamber of
IDE and that of insulin were calculated using VOIDOO44
and shape complementarity was calculated as described
previously.45
Supplementary Data
Supplementary data associated with this article
can be found, in the online version, at doi:10.1016/
j.jmb.2009.10.072
References
1. Malito, E., Hulse, R. E. & Tang, W. J. (2008). Amyloid
beta-degrading cryptidases: insulin degrading
enzyme, presequence peptidase, and neprilysin. Cell.
Mol. Life Sci. 65, 2574–2585.
2. Hulse, R. E., Ralat, L. A. & Tang, W. J. (2009). Structure,
function, and regulation of insulin-degrading enzyme.
Vitam. Horm. 80, 635–648.
3. Shen, Y., Joachimiak, A., Rosner, M. R. & Tang, W. J.
(2006). Structures of human insulin-degrading enzyme
reveal a new substrate recognition mechanism. Nature,
443, 870–874.
4. Duckworth, W. C., Bennett, R. G. & Hamel, F. G.
(1998). Insulin degradation: progress and potential.
Endocr. Rev. 19, 608–624.
5. Johnson, K. A., Bhushan, S., Stahl, A., Hallberg, B. M.,
Frohn, A., Glaser, E. & Eneqvist, T. (2006). The closed
structure of presequence protease PreP forms a unique
10,000 Å3 chamber for proteolysis. EMBO J. 25,
1977–1986.
6. Taylor, A. B., Smith, B. S., Kitada, S., Kojima, K.,
Miyaura, H., Otwinowski, Z. et al. (2001). Crystal
structures of mitochondrial processing peptidase
reveal the mode for specific cleavage of import signal
sequences. Structure, 9, 615–625.
Enzymatic competition assay
Enzyme activities were assayed using a fluorogenic
bradykinin-mimetic substrate of IDE, substrate V (7-
methoxycoumarin-4yl-acetyl-RPPGFSAFK-2,4-dinitro-
phenyl; R&D Systems).3,19–21 Competition reactions were
carried out at 37 °C by mixing 90 μl of 0.5 μM substrate V
in 50 mM potassium phosphate buffer (pH 7.3) and 10 μl

442
Substrate Recognition by Human IDE
7. Maskos, K. (2004). Pitrilysins/inverzincins. In Hand-
book of Metalloproteins (Messerschmidt, A., Dode, W. &
Cygler, M., eds), vol. 3, pp. 190–198. John Wiley &
Sons, Hoboken, NJ.
8. Fawcett, J., Permana, P. A., Levy, J. L. & Duckworth,
W. C. (2007). Regulation of protein degradation by
insulin-degrading enzyme: analysis by small inter-
fering RNA-mediated gene silencing. Arch. Biochem.
Biophys. 468, 128–133.
9. Mirsky, I. A. & Broh-Kahn, R. H. (1949). The
inactivation of insulin by tissue extracts; the distribu-
tion and properties of insulin inactivating extracts.
Arch. Biochem. 20, 1–9.
10. Farris, W., Mansourian, S., Chang, Y., Lindsley, L.,
Eckman, E. A. & Frosch, M. P. (2003). Insulin-
degrading enzyme regulates the levels of insulin,
amyloid beta-protein, and the beta-amyloid precursor
protein intracellular domain in vivo. Proc. Natl Acad.
Sci. USA, 100, 4162–4167.
11. Sladek, R., Rocheleau, G., Rung, J., Dina, C., Shen, L. &
Serre, D. (2007). A genome-wide association study
identifies novel risk loci for type 2 diabetes. Nature,
445, 881–885.
12. Leissring, M. A., Farris, W., Chang, A. Y., Walsh, D. M.,
Wu, X., Sun, X. et al. (2003). Enhanced proteolysis of
beta-amyloid in APP transgenic mice prevents plaque
formation, secondary pathology, and premature death.
Neuron, 40, 1087–1093.
13. Kurochkin, I. V. (2001). Insulin-degrading enzyme:
embarking on amyloid destruction. Trends Biochem.
Sci. 26, 421–425.
14. Kim, M., Hersh, L. B., Leissring, M. A., Ingelsson, M.,
Matsui, T. & Farris, W. (2007). Decreased catalytic
activity of the insulin-degrading enzyme in chromo-
some 10-linked Alzheimer disease families. J. Biol.
Chem. 282, 7825–7832.
and II by a human insulin degrading enzyme. Endocr.
Res. 10, 101–112.
23. Terasawa, H., Kohda, D., Hatanaka, H., Nagata, K.,
Higashihashi, N. & Fujiwara, H. (1994). Solution
structure of human insulin-like growth factor II;
recognition sites for receptors and binding proteins.
EMBO J. 13, 5590–5597.
24. Cooke, R. M., Harvey, T. S. & Campbell, I. D. (1991).
Solution structure of human insulin-like growth
factor 1: a nuclear magnetic resonance and re-
strained molecular dynamics study. Biochemistry, 30,
5484–5491.
25. Broughton, S. & Partridge, L. (2009). Insulin/IGF-like
signalling, the central nervous system and aging.
Biochem. J. 418, 1–12.
26. Pollak, M. (2008). Insulin and insulin-like growth
factor signalling in neoplasia. Nat. Rev., Cancer, 8,
915–928.
27. Harvey, T. S., Wilkinson, A. J., Tappin, M. J., Cooke,
R. M. & Campbell, I. D. (1991). The solution structure
of human transforming growth factor alpha. Eur. J.
Biochem. 198, 555–562.
28. Kohda, D. & Inagaki, F. (1992). Three-dimensional
nuclear magnetic resonance structures of mouse
epidermal growth factor in acidic and physiological
pH solutions. Biochemistry, 31, 11928–11939.
29. Schneider, M. & Wolf, E. (2009). The epidermal
growth factor receptor ligands at a glance. J. Cell.
Physiol. 218, 460–466.
30. Haataja, L., Gurlo, T., Huang, C. J. & Butler, P. C.
(2008). Islet amyloid in type 2 diabetes, and the toxic
oligomer hypothesis. Endocr. Rev. 29, 303–316.
31. Höppener, J. W., Ahrén, B. & Lips, C. J. (2000). Islet
amyloid and type 2 diabetes mellitus. N. Engl. J. Med.
343, 411–419.
32. Baker, N. A., Sept, D., Joseph, S., Holst, M. J. &
McCammon, J. A. (2001). Electrostatics of nanosystems:
application to microtubules and the ribosome. Proc.
Natl Acad. Sci. USA, 98, 10037–10041.
33. Chothia, C., Lesk, A. M., Dodson, G. G. & Hodgkin,
D. C. (1983). Transmission of conformational change
in insulin. Nature, 302, 500–505.
34. Moy, F. J., Li, Y. C., Rauenbuehler, P., Winkler, M. E.,
Scheraga, H. A. & Montelione, G. T. (1993). Solution
structure of human type-alpha transforming growth
factor determined by heteronuclear NMR spectro-
scopy and refined by energy minimization with
restraints. Biochemistry, 32, 7334–7353.
15. Keller, S., Schmid, C., Zapf, J. & Froesch, E. R. (1989).
Inhibition of insulin degradation by insulin-like
growth factors I and II in human hepatoma (HepG2)
cells. Acta Endocrinol. (Copenh.), 121, 279–285.
16. Misbin, R. I., Almira, E. C., Duckworth, W. C. & Mehl,
T. D. (1983). Inhibition of insulin degradation by
insulin-like growth factors. Endocrinology, 113,
1525–1527.
17. Gehm, B. D. & Rosner, M. R. (1991). Regulation of
insulin, epidermal growth factor, and transforming
growth factor-alpha levels by growth factor-degrading
enzymes. Endocrinology, 128, 1603–1610.
18. Muller, D., Baumeister, H., Buck, F. & Richter, D.
(1991). Atrial natriuretic peptide (ANP) is a high-
affinity substrate for rat insulin-degrading enzyme.
Eur. J. Biochem. 202, 285–292.
35. Torres, A. M., Forbes, B. E., Aplin, S. E., Wallace, J. C.,
Francis, G. L. & Norton, R. S. (1995). Solution structure
of human insulin-like growth factor II. Relationship to
receptor and binding protein interactions. J. Mol. Biol.
248, 385–401.
36. Hamel, F. G., Gehm, B. D., Rosner, M. R. & Duckworth,
W. C. (1997). Identification of the cleavage sites of
transforming growth factor alpha by insulin-degrading
enzymes. Biochim. Biophys. Acta, 1338, 207–214.
37. Leissring, M. A., Lu, A., Condron, M. M., Teplow, D. B.,
Stein, R. L., Farris, W. & Selkoe, D. J. (2003). Kinetics of
amyloid beta-protein degradation determined by
novel fluorescence- and fluorescence polarization-
based assays. J. Biol. Chem. 278, 37314–37320.
38. Felder, C. E., Prilusky, J., Silman, I. & Sussman, J. L.
(2007). A server and database for dipole moments of
proteins. Nucleic Acids Res. 35, W512–521.
19. Im, H., Manolopoulou, M., Malito, E., Shen, Y., Zhao,
J. & Neant-Fery, M. (2007). Structure of substrate-free
human insulin-degrading enzyme (IDE) and bio-
physical analysis of ATP-induced conformational
switch of IDE. J. Biol. Chem. 282, 25453–25463.
20. Malito, E., Ralat, L. A., Manolopoulou, M., Tsay, J. L.,
Wadlington, N. L. & Tang, W. J. (2008). Molecular
bases for the recognition of short peptide substrates
and cysteine-directed modifications of human insulin-
degrading enzyme. Biochemistry, 47, 12822–12834.
21. Manolopoulou, M., Guo, Q., Malito, E., Schilling, A. B.
& Tang, W. J. (2009). Molecular basis of catalytic
chamber-assisted unfolding and cleavage of human
insulin by human insulin degrading enzyme. J. Biol.
Chem. 284, 14177–14188.
39. Murshudov, G. N., Vagin, A. A. & Dodson, E. J.
(1997). Refinement of macromolecular structures by
the maximum-likelihood method. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 53, 240–255.
22. Roth, R. A., Mesirow, M. L., Yokono, K. & Baba, S.
(1984). Degradation of insulin-like growth factors I

Substrate Recognition by Human IDE
443
40. Potterton, E., McNicholas, S., Krissinel, E., Cowtan, K.
& Noble, M. (2002). The CCP4 molecular-graphics
project. Acta Crystallogr., Sect. D: Biol. Crystallogr. 58,
1955–1957.
41. Emsley, P. & Cowtan, K. (2004). Coot: model-building
tools for molecular graphics. Acta Crystallogr., Sect. D:
Biol. Crystallogr. 60, 2126–2132.
for macromolecular structure determination. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 54, 905–921.
43. Delano, W. L. (2002). PyMOL. Pymol.sourceforge.net.
44. Kleywegt, G. J. & Jones, T. A. (1994). Detection,
delineation, measurement and display of cavities in
macromolecular structures. Acta Crystallogr., Sect. D:
Biol. Crystallogr. 50, 178–185.
42. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano,
W. L., Gros, P. & Grosse-Kunstleve, R. W. (1998).
Crystallography & NMR system: a new software suite
45. Lawrence, M. C. & Colman, P. M. (1993). Shape
complementarity at protein/protein interfaces. J. Mol.
Biol. 234, 946–950.