Grassystatins A-C from marine cyanobacteria, potent cathepsin E inhibitors that reduce antigen presentation

Article abstract of DOI:10.1021/jm9009394

In our efforts to exploremarine cyanobacteria as a source of novel bioactive compounds, we discovered a statine unit-containing linear decadepsipeptide, grassystatin A (1), which we screened against a diverse set of 59 proteases. We describe the structure

Full text of DOI:10.1021/jm9009394

5
732 J. Med. Chem. 2009, 52, 5732–5747
DOI: 10.1021/jm9009394
Grassystatins A-C from Marine Cyanobacteria, Potent Cathepsin E Inhibitors That Reduce Antigen
Presentation
†
Jason C. Kwan, Erika A. Eksioglu, Chen Liu, Valerie J. Paul, and Hendrik Luesch*
‡
‡
§
,†
†
‡
Department of Medicinal Chemistry and Department of Pathology, Immunology, and Laboratory Medicine, University of Florida, Gainesville,
§
Florida 32610, and Smithsonian Marine Station, 701 Seaway Drive, Fort Pierce, Florida 34949
Received June 25, 2009
In our efforts to explore marine cyanobacteria as a source of novel bioactive compounds, we discovered a
statine unit-containing linear decadepsipeptide, grassystatin A (1), which we screened against a diverse set
of 59 proteases. We describe the structure determination of 1 and two natural analogues, grassy-
statins B (2) and C (3), using NMR, MS, and chiral HPLC techniques. Compound 1 selectively inhibited
cathepsins D and E with IC s of 26.5 nM and 886 pM, respectively. Compound 2 showed similar potency
5
0
and selectivity against cathepsins D and E (IC s of 7.27 nM and 354 pM, respectively), whereas the
5
0
truncated peptide analogue grassystatin C (3), which consists of two fewer residues than 1 and 2, was less
potent against both but still selective for cathepsin E. The selectivity of compounds 1-3 for cathepsin E
over D (20-38-fold) suggests that these natural products may be useful tools to probe cathepsin E
function. We investigated the structural basis of this selectivity using molecular docking. We also show
that 1 can reduce antigen presentation by dendritic cells, a process thought to rely on cathepsin E.
Although all proteases share in common the ability to
cleave peptidebonds, their variousregulatory roleshave made
them interesting targets for drug discovery. They are involved
in such diverse processes as blood coagulation, the cell cycle,
nonselective inhibition of metalloproteases is thought to be the
reason for musculoskeletal side effects seen in early matrix
a
metalloprotease (MMP) inhibitors that were evaluated for
cancer treatment.
2
1
infection, and neurodegenerative disorders, among others.
Modified peptides have potential for use as protease in-
hibitors. Their peptidic nature may allow them to bind in a
substrate-like manner at the active site or to interact with
other sites on the enzyme. Modifications made to the normal
peptide structure could increase resistance to proteolytic
cleavage (so that they act as inhibitors and not substrates)
or confer onthem moredrug-like properties, such as increased
lipophilicity. It is with this in mind that we have been engaged
in a systematic search for protease inhibitors among natural
products produced by marine cyanobacteria. This ancient
group of organisms is known to produce a vast array of
secondary metabolites. These are often lipophilic modified
However, because of the ubiquity of proteolytic signaling,
potential therapeutic inhibitors must be selective in order to
reduce the chance of undesirable off-target effects. For example,
*Corresponding author. E-mail: luesch@cop.ufl.edu. Tel: 352-273-
7
738. Fax: 352-273-7741.
a
Abbreviations: 7-AAD, 7-aminoactinomycin D; ACE, angiotensin
converting enzyme; APC, antigen presentation cell; APT, attached
proton test; ADAM, protein with a disintegrin and a metalloprotease
domain; BSA, bovine serum albumin; CAD, collisionally activated
decomposition; CE, collission energy; CFSE, carboxyfluorescein diace-
tate succinimidyl ester; COSY, correlation spectroscopy; CUR, curtain
gas; CXP, collision-cell exit potential; DC, dendritic cell; DMEM,
Dulbecco’s modified Eagle medium; DP, declustering potential; EP,
entrance potential; ESIMS, electrospray ionization mass spectrometry;
FDLA, 1-fluoro-2,4-dinitro-5-leucinamide; GS1, gas 1; GS2, gas 2;
GM-CSF, granulocyte macrophage colony-stimulating factor; HE-
TLOC, hetero half-filtered TOCSY for measurement of long-range
coupling constants; HMBC, heteronuclear multiple-bond correlation
spectroscopy; HMQC, heteronuclear multiple-quantum correlation
spectroscopy; HRESI/APCIMS, high-resolution electrospray ioniza-
tion/atmospheric pressure chemical ionization mass spectrometry
3
peptides that possess potent cytotoxicity. Such metabolites
have presumably been optimized by millions of years of
natural selection to be potent and specific to their intended
target. In some cases, the ecological target may be protease
enzymes. Already, we have identified several lyngbyastatins
4
that potently inhibit the serine protease elastase. Cyanobac-
teria produce modified peptides through the nonribosomal
peptidesynthetase (NRPS) pathway or throughcombinations
(dual probe); HSQC, heteronuclear single-quantum correlation spec-
5
of the NRPS and polyketide synthase (PKS) pathways. Both
troscopy; IFN-γ, interferon-γ; IL-4, interleukin-4; Ii, invariant chain;
IL-17, interleukin-17; IS, ionspray voltage; MHC, major histocompat-
ibility complex; MLR, mixed lymphocyte reaction; MMP, matrix
metalloprotease; MoDC, monocyte-derived dendritic cell; MRM, mul-
tiple reaction monitoring; NRPS, nonribosomal peptide synthetase;
PBMC, peripheral blood mononuclear cell; PBS, phosphate-buffered
saline; PDB, Protein Data Bank; PKS, polyketide synthase; PE, phy-
coerythrin; PMA, phorbol 12-myristate 13-acetate; rmsd, root mean
square deviation; ROESY, rotating frame nuclear Overhauser effect
spectroscopy; TACE, tumor necrosis factor R-converting enzyme;
of these pathways are highly modular, presumably allowing
evolution of bioactive compounds through combinatorial
alterations. The modular architecture of these pathways has
led several groups to pursue combinatorial biosynthesis of
6
novel “non-natural” products.
Herein, we describe the isolation, structure determination,
and biological evaluation of three linear modified peptides,
grassystatins A-C (1-3, Figure 1). All three contain a statine
unit [(3S,4S)-4-amino-3-hydroxy-6-methylheptanoic acid, Sta],
TEM, temperature; T
H
, T helper cell; TOCSY, total correlation spec-
troscopy; TTc, tetanus toxin C-fragment.
r
pubs.acs.org/jmc
Published on Web 08/28/2009
2009 American Chemical Society

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5733
Figure 1. Structures of grassystatins A-C (1-3).
inhibition of cathepsins in a cellular system and also the
disruption of antigen presentation by dendritic cells (DCs),
11
a process in which cathepsin E has been recently implicated.
Results
Isolation and Structure Determination. Samples of the
cyanobacterium, identified as Lyngbya cf. confervoides, were
12
collected off Grassy Key as described previously, and off
Key Largo, FL. The nonpolar extract (MeOH-EtOAc, 1:1)
of each collection was subjected to silica chromatography
and reversed-phase HPLC to furnish 1 and 2. Compound 3
was found only in the Key Largo collection that previously
1
2
yielded grassypeptolide.
HRESI/APCIMS and NMR data for 1 suggested a mo-
lecular formula of C H N O (m/z 1196.6812 for [M þ
5
8
95
9
16
þ
þ
Na] , 1174.6988 for [M þ H] , 598.8455 for [M þ H þ
2
þ
2þ
1
Na] , and 587.8544 for [M þ 2H] ). Perusal of the H and
1
3
C NMR spectra revealed that it was a depsipeptide
Table 1), with several exchangeable proton signals chara-
(
cteristic of amides (δ ∼6 to ∼8), R-protons (δ ∼4 to ∼5),
H
H
1
13
and some deshielded signals in both the H and C NMR
spectra indicative of methines adjacent to an ester linkage
(
δH/C 5.13/78.1 and 4.70/77.5). There were also several
N-methyl signals (δH 3.01 and 2.30) and one O-methyl
apparent (δ 3.72). In addition, two conformers were present
H
Figure 2. Structures of pepstatin A (including binding site
nomenclature), tasiamide, and tasiamide B.
1
3
1
13
in the ratio 15:1. Analysis of the H NMR, C NMR, APT,
COSY, edited HSQC, HMBC, ROESY, and TOCSY spec-
tra in CDCl of 1 (Table 1 and Supporting Information)
3
revealed the presence of four proteinogenic amino acid units
(Ala, Thr, Asn, and Leu) and two hydroxyisovaleric acid
(Hiva) moieties. In addition, O-Me-Pro, N-Me-Phe, N,N-
which was first described in the broad-spectrum natural
7
aspartic protease inhibitor pepstatin A (Figure 2). In the
latter, statine arises from a mixed NRPS/PKS pathway that
8
condenses leucine and malonate units. The closest structural
relatives among cyanobacterial natural products are tasi-
Me
that there were two terminal groups (O-Me-Pro and N,
N-Me -Val), it was clear from the degree of unsaturation
-Val, and statine (Sta, C-25-C-32) were deduced. Given
2
9
10
amide and tasiamide B (Figure 2). Tasiamide does not
contain a statine unit, and there are also some differences in
configuration of several amino acid residues (Figure 2). To
test aspartic and other protease inhibitory activity, we
screened compound 1 against 59 diverse proteases and found
selective inhibition of the aspartic proteases cathepsins D and
E. Notably, compounds 1-3 discriminate between these two
enzymes, while pepstatin A does not. We demonstrate the
2
that the compound was linear (all 16 double bond equiva-
lents were accounted for). The hydroxyl protons for Sta and
1
4
Thr units were evident and thus precluded branching of the
chain through ester linkages at these positions. The frag-
ments Sta-Thr-Ala-(N-Me-Phe)-(O-Me-Pro) and (N,N-Me
-
2
Val)-Hiva-Hiva-Leu-Asn were readily constructed with the

5
734 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
Kwan et al.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5735

5
736 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
Kwan et al.
Figure 3. ESIMS fragmentation pattern of grassystatins A (1) and B (2).
help of HMBC and ROESY data. The continuous sequence
of the two fragments was confirmed by ESIMS fragmenta-
tion (Figure 3). Without other evidence, however, it was still
unclear whether the Sta and Asn units were joined through
C-1 or C-4 of Asn (C-33 and C-36 in 1), as no correlations
shown that the relative configuration of statine and statine-
like units derived from other amino acids can be easily
determined by examination of the coupling constants of
the R-methylene signals. The downfield R-methylene signal
(H-26a) shows a large coupling to H-27, and the upfield
H-26 (H-26b) shows a small coupling to H-27 (8.7 and 5.4
Hz, respectively), thus indicating that the configuration is
3S,4S.
2
0
were observed through the NH group or from the NH or
2
H-28 in the Sta unit. Collection of NMR data for 1 in
DMSO-d (see Table S1, Supporting Information) revealed
6
an extra HMBC correlation from one of the NH amide
2
protons to the β-carbon of Asn, thus determining the chain
proceeded through C-1.
HRESI/APCIMS of 2 suggested a molecular formula of
þ
C H N O (m/z 1226.6687 for [M þ K] , 1210.6936 for
5
9
97
9
16
þ
þ
þ
[M þ Na] , 1188.7119 for [M þ H] , and 605.8516 for [M þ
2
1
A portion of 1 was hydrolyzed (6 N HCl, 110 °C, 24 h) and
analyzed by chiral HPLC-MS. This revealed the presence of
H þ Na] ), and the H NMR spectrum indicated a striking
similarity to 1, including the same conformational ratio.
1
5
1
Examination of the H NMR, COSY, HMQC, HMBC,
L-Pro, N-Me-D-Phe, L-Ala, L-Thr, L-Asp, L-Leu, and N,
N-Me -L-Val. In addition, peaks corresponding to both
ROESY, and TOCSY spectra of 2 (Table 1 and Supporting
Information) revealed the presence of the same units found
in 1, except for 2-aminobutyric acid (Aba) in place of Ala.
The close similarity of proton and carbon chemical shifts
between 1 and 2 indicated that 2 had the same sequence and
relative configuration as 1. Compounds 1 and 2 exhibited
2
L- and D-Hiva were detected, indicating two units of opposite
configuration were present. To assign their order, another
portion of 1 was subjected to base hydrolysis (0.5 N NaOH/
MeOH 1:1, rt, 72 h) to selectively hydrolyze the ester bonds
and liberate the two terminal units (Hiva-2 and N,N-Me2-
Val). Chiral analysis of the base hydrolysate indicated the
presence of L-Hiva only, thus determining the configuration
shown for 1 (Figure 1).
2
0
D
very similar optical rotation ([R]
-4.4 and -5.0, respec-
tively, c 0.08 and 0.1, both in MeOH), indicating that they
have the same absolute configuration. The sequence of 2 was
confirmed by ESIMS fragmentation (Figure 3).
To establish the configuration of the Sta, a portion of the
acid hydrolysate of 1 was derivatized with L-FDLA and
þ
The m/z peak at 1009.5941, for [M þ Na] , in the HRESI/
1
6
subjected to modified Marfey’s analysis. Peaks corre-
sponding to both (3S,4S)- and (3R,4S)-Sta-L-FDLA were
detected, probably due to epimerization at C-3 resulting
from dehydration/rehydration. An attempt to confirm the
APCIMS and NMR data for 3 suggested a molecular for-
1
mula of C H N O . Analysis of the H NMR spectrum
5
0
82
8
12
suggested that the compound was a peptide (amide signals at
δH 6-8, R-proton signals at δH ∼4-5.5) with at least two
conformers, of which the most prominent ones exist in the
ratio of 2.45:1. Aromatic signals (δH 7.2-7.3), putative N-
methyl singlets (δH 3.090, 3.087, 3.05, and 2.77) and an O-
1
7
relative configuration of this unit in situ by J-based analysis
failed, probably because the small H-27-H-28 coupling
(2.7 Hz) precluded measurement of heteronuclear coupling
constants by HETLOC
1
8,19
across this bond. It was recently
methyl singlet (δ 3.75) were also observed. Analysis of the
H

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5737
1
H NMR, COSY, edited HSQC, HMBC, ROESY, and
TOCSY spectra of 3 recorded in CDCl revealed the pre-
lysate, using the same fluorogenic substrate used in previous
experiments involving cathepsins D and E (see Figure 7A
and Experimental Section). MCF7 cells express both cathe-
3
sence of four proteinogenic R-amino acids (Pro, Gly, Ile,
Leu), two N-methylated R-amino acids (N-Me-Phe, N-Me-
Gln), one hydroxy acid (2-hydroxy-3-methylpentanoic acid,
Hmpa), and Sta (C-25-C-32, Table 2). The sequence (N-Me-
Phe)-Gly-Ile-Sta-Gln-Leu-Hmpa could be determined by
HMBC analysis. A ROESY correlation between H-5a and
H-8 allowed the joining of O-Me-Pro to N-Me-Phe. ROESY
data also confirmed the previously determined HMBC
sequence (Table 2). It was unambiguously established that
C-5 of N-Me-Gln (C-37 in 3) was the primary amide carbon,
by virtue of the HMBC correlation of H-28 to C-33 and
correlations from H-34 and H -38 to C-39. Additionally,
3
there was a [M - 128] peak at m/z 858.5322 in the HRESI/
APCIMS which was consistent with loss of O-Me-Pro (calcd
for C H N O , 858.5341). By default, an OH group was
10
proposed at C-46, and this was supported by the proton
chemical shift at this position (δ 4.15), which suggested OH
2
2,23
psins D and E,
assess the cell permeability of pepstatin A derivatives using
and they have previously been used to
2
4
enzymatic experiments. However, relative expression levels
of cathepsins D and E in this cell line have not been directly
compared. The apparent IC s of inhibition in this system
5
0
were ∼0.5 μM for 1 and ∼5 nM for pepstatin A. For both
compounds, this represents roughly a 20-fold decrease in
potency compared to experiments with purified cathepsin D
(and cathepsin E in the case of pepstatin A) but for 1 a 564-
fold decrease in potency compared to experiments with
purified cathepsin E. If one examines the relative potency
of 1 and pepstatin A in experiments using purified cathepsins
D and E (Table 3), it is apparent that the IC50 of 1 is 153-fold
higher than that of pepstatin A for cathepsin D, and against
cathepsin E the IC50 of 1 is 4.9-fold higher than that of
pepstatin A. In the MCF7 lysate experiment (Figure 7A), 1
exhibits a 100-fold difference in IC50 compared to pepstatin
A (∼0.5 μM versus ∼5 nM, respectively). This fact, together
with similar (20-fold) decrease in potency of pepstatin A
and grassystatin A (1) of protease inhibition in MCF7
lysate compared with purified cathepsin D, suggests that
cathepsin D predominates in MCF7 lysate, as the potency
difference is closer to 153-fold than 4.9-fold. Pepstatin A
exhibits reduced potency in MCF7 lysate compared to both
purified cathepsins D and E, as mentioned previously. This
suggests the presence of other enzymes in the lysate which
can cleave the substrate, that are either not inhibited by
pepstatin A or only inhibited at high concentrations of this
compound.
þ
4
4
72
7
H
rather than an acyloxy group. The sequence was further
confirmed by ESIMS fragmentation (Figure 4).
A portion of 3 was hydrolyzed (6 N HCl, 110 °C, 24 h) and
analyzed by chiral HPLC-MS. Peaks corresponding to
2
1
L-Pro, N-Me-D-Phe, L-Ile, N-Me-L-Glu, and L-Leu were
detected. The four stereoisomers of Hmpa eluted very closely
together, but a putative assignment of (2R,3S)-Hmpa was
made. This was later confirmed by analysis of the hydro-
lysate by conventional chiral HPLC with a different column,
under conditions where the four stereoisomers eluted further
apart (see Experimental Section). A portion of the hydro-
lysate was then derivatized with L-FDLA as with 1, and once
again, two peaks were detected corresponding to (3R,4S)-
Sta-L-FDLA and (3S,4S)-Sta-L-FDLA. The further down-
We then treated intact living MCF7 cells with both
compound 1 and pepstatin A, using a protocol previously
field of the CH protons at C-26 showed a large coupling
2
constant to H-27 (9.3 Hz), indicating that the configuration
2
of this unit is 3S,4S.
24
used to study pepstatin A conjugates (see Figure 7B).
0
Following treatment, cells were washed and treated with
trypsin, which has previously been shown to reduce surface-
bound inhibitors which can interfere with subsequent enzy-
Protease Profiling and Cellular Activity Studies. The struc-
tures of compounds 1-3, in particular the presence of the
statine unit and similarity to pepstatin A, led us to suspect
that these compounds may be aspartic protease inhibitors.
To test activity and to probe selectivity for certain aspartic
and other proteases, we screened 1 against a panel of
proteases to identify inhibitory activity at 10 μM (Figure 5
and Table S2, Supporting Information). It was found to be
active against a subset of aspartic proteases: cathepsin D and
cathepsin E. The only other proteases with compromised
activities were the metalloproteases ADAM9, ADAM10,
and TACE (Figure 5). Subsequent validation of these hits
revealed that the greatest activity was against cathepsin E
2
5
matic assays. After lysis, the enzymatic activity of the lysate
was measured as before with a fluorogenic substrate. Con-
sistent with previous reports that pepstatin A has poor cell
permeability, its apparent IC50 (∼5 μM) is now almost
29000-fold higher than with purified cathepsins D and E
and 1000-fold higher than with MCF7 lysate. Conversely,
the apparent IC50 for compound 1 (also ∼5 μM) is almost
200-fold higher than with purified cathepsin D and 10-fold
higher than with MCF7 lysate (see Figure 7B). The selectivity
of pepstatin A is of course different to that of compound 1
(vide supra), and the subcellular targeting of 1 is unknown,
including local concentration at the site of cathepsin D/E
localization. Another uncertainty is the stability of 1 and its
route of metabolism within cells and in lysates. However,
collectively these results show that the barrier of the cell
membrane reduces the apparent potency of pepstatin A to a
much greater extent than that of compound 1 (1000- versus
10-fold), and they suggest that 1 can successfully enter the
cell to inhibit cellular enzymes. Results of subsequent experi-
ments (vide infra) supported this notion.
(IC50 886 pM). Compounds 1-3 all showed selectivity for
cathepsin E over cathepsin D compared to pepstatin A
(Table 3).
Of the metalloproteases, only TACE inhibition was vali-
dated in the second round of assays (Table 3). IC s against
ADAM9 and ADAM10 were in the high micromolar range
or above 100 μM. The IC s of TACE inhibition were in the
low micromolar range, and in contrast to the dose-dependent
but time-independent inhibition of cathepsins (Figure 6A
and Figure S1A, Supporting Information), analysis of the
progress curves revealed concentration- and time-dependent
inhibition (Figure 6B and Figure S1B, Supporting In-
formation).
5
0
5
0
Investigation of the Effect of Grassystatin A (1) on Antigen
Presentation. Cathepsin E is thought to have a functional
role in the proteolysis of antigenic peptides, which are
subsequently presented as antigens on the surface of anti-
gen-presenting cells (APCs) in the major histocompatibility
To assess whether grassystatin A (1) was able to inhibit
cellular enzymes, we investigated the effect of 1 on MCF7 cell
2
4,26
complex (MHC) class II pathway.
Antigen presentation

5
738 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
Kwan et al.
Table 2. NMR Spectral Data for Grassystatin C (3) at 600 MHz in CDCl
a
3
a
conformer 1
conformer 2
b
, mult
b
, mult
1
1
c
C/H no.
δ
H
(J in Hz)
δ
C
δ
H
(J in Hz)
δ
C
H- H COSY
key HMBC
key ROESY
O-Me-Pro
1
2
3
3
4
4
5
5
6
7
8
9
9
1
1
1
1
1
173.1, s
58.8, d
28.8, t
173.1, s
58.8, d
28.8, t
4.35, dd (7.9, 5.4)
2.02, m
1.82, m
1.88, m
1.72, m
3.45, m
3.05, m
3.72, s
4.35, dd (7.9, 5.4)
2.02, m
1.82, m
1.88, m
1.72, m
3.45, m
3.05, m
3.72, s
3a, 3b
1
a
b
a
b
a
b
2, 3b, 4a, 4b
2, 3a, 4a, 4b
3a, 3b, 4b, 5a, 5b
3a, 3b, 4a, 5a, 5b
4a, 4b, 5b
1
1
6
24.8, t
46.7, t
24.8, t
46.7, t
H-8
3b
4a, 4b, 5a
52.3, q
168.6, s
56.2, d
35.4, t
52.3, q
168.6, s
56.4, d
35.4, t
1
N-Me-Phe
5.45, dd (8.8, 6.4)
3.27, m
2.83, m
5.40, dd (8.8, 6.3)
3.27, m
2.83, m
9a, 9b
8, 9b
8, 9a
16, 17
H-5a, 16
a
b
0
7
7
16
16
136.7, s
136.7, s
1/15 7.21, m
2/14 7.25, m
129.3, d 7.21, m
128.5, d 7.25, m
126.8, d 7.19, m
30.0, q
168.8, s
129.3, d 12/14
128.5, d 11/15, 13
126.8, d 12/14
30.0, q
168.8, s
41.1, t
3
6
7.19, m
3.057, s
3.061, s
8, 17
8, 9a, 9b, 18a, 18b
Gly
Ile
17
1
1
8a
8b
4.11, dd (-17.4, 4.7) 41.2, t
3.94, dd (-17.4, 3.7)
7.49, dd (4.7, 3.7)
171.8, s
4.23, dd (-17.1, 5.5)
3.82, dd (-17.1, 3.8)
7.70, dd (5.5, 3.8)
18b, NH
18a, NH
18a, 18b
17, 19
17, 19
19
16
16
20
NH
19
171.8, s
57.8, d
36.5, d
11.8, q
24.3, t
2
2
2
2
2
2
0
1
2
3a
3b
4
4.50, dd (8.0, 5.7)
2.01, m
0.91, m
1.48, m
1.12, m
57.9, d
36.6, d
15.7, q
24.3, t
4.56, dd (9.2, 4.4)
2.13, m
0.93, m
1.46, m
1.12, m
21, NH
20, 22, 23b
21
23b, 24
22, 23a, 24
23a, 23b
20
19, 25
NH (Gly)
0.88, m
7.25, m
11.7, q
0.88, m
7.40, d (8.9)
11.7, q
NH
25
25
26a, 26b
Sta
172.6, s
2.48, dd (-13.9, 9.3) 38.7, t
2.35, m
173.0, s
38.2, t
2
2
2
2
2
3
3
3
6a
6b
7
8
9
0
1
2
2.56, dd (-13.3, 9.6)
2.39, m
3.95, m
26b, 27
26a, 27
26a, 26b, 28, OH
27, 29, NH
28, 30
25
25
NH (Ile)
NH (Ile)
d
3.92, m
3.98, m
1.35, m (2H)
1.45, m
0.88, m
0.83, m
4.78, b
72.3, d
52.0, d
38.9, t
24.7, d
23.7, q
21.2, q
73.7, d
52.5, d
38.9, t
24.7, d
23.3, q
21.2, q
4.07, m
33
1.35, m (2H)
1.45, m
0.84, m
0.83, m
4.78, b
7.92, d (8.5)
29, 31, 32
30
30
38
34
d
OH
NH
27
28
6.49, b
33
N-Me-Gln 33
170.2, s
55.8, d
23.7, t
170.2, s
59.5, d
25.9, t
3
3
3
3
3
3
3
4
5.09, m
2.36, m
1.85, m
2.30, m
2.14, m
5.07, m
2.60, m
1.64, m
2.51, m
35a, 35b
34, 35b, 36a, 36b
34, 35a, 36a, 36b 33, 37
33, 38, 39
38, 40, NH (Sta)
38
38
5a
5b
6a
6b
7
32.5, t
33.1, t
35a, 35b, 36b
35a, 35b, 36a
37
37
2.22, ddd (-14.1, 8.5, 5.6)
38
175.0, s
30.3, q
175.0, s
29.1, q
8
3.02, s
2.74, s
34, 39
30, 32, 34, 35a, 35b,
3
6b, 40, 41b, 43/44
NH
NH
39
2
a
b
6.45, b
6.15, b
6.43, b
6.14, b
NH
NH
2
2
b
a
2
Leu
Hmpa
a
173.5, s
47.9, d
40.8, t
172.3, s
46.9, d
41.0, t
4
4
4
4
4
4
0
4.85, m
1.67, m
1.32, m
1.66, m
0.97, m
0.93, m
7.27, m
4.87, m
1.78, m
1.53, m
1.66, m
0.94, m
0.97, m
7.19, m
41a, 41b, NH
40, 41b, 42
40, 41a, 42
41a, 41b, 43, 44
42
39 (w), 45 (w) 34, 38
39
1a
1b
2
3
4
39
38
24.7, t
21.2, q
23.4, q
24.7, t
21.9, q
23.2, q
38
38
42
40
NH
45
45
45
174.3, s
74.1, d
38.5, d
12.8, q
26.2, t
175.1, s
74.1, d
38.5, d
12.4, q
26.2, t
4
4
4
4
4
5
6
7
8
9a
9b
0
4.11, m
1.83, m
0.80, d (6.8)
1.45, m
1.31, m
4.11, m
1.83, m
0.76, d (6.8)
1.45, m
1.31, m
47
46, 48, 49a, 49b
47
47, 49b, 50
47, 49a, 50
49a, 49b
0.92, m
11.8, q
0.92, m
11.8, q
OH
b
Conformers 1 and 2 were the most prominent, and they were evident in approximately a 2.45:1 ratio. Multiplicity derived from APT and HMQC
c
d
spectra. Protons showing long-range correlation to indicated carbon. Correlation apparent for conformer 2 only.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5739
Figure 4. ESIMS fragmentation pattern for grassystatin C (3).
Figure 5. Protease panel treated with grassystatin A (1), 10 μM. Values represent % enzyme activity compared to solvent control and are
additionally represented by a continuous color scale (0%, red; 100%, green).
Table 3. IC50s of Grassystatins A-C (1-3) against Aspartic Proteases and Metalloproteases Identified in the Primary Screen
a
a
protease
grassystatin A (1)
grassystatin B (2)
grassystatin C (3)
pepstatin A
GM6001
cathepsin D
cathepsin E
ADAM9
ADAM10
TACE
26.5 ( 5.4 nM
886 ( 135 pM b
46.1 ( 21.7 μM
7.27 ( 0.90 nM
354 ( 192 pM
1.62 ( 0.3 μM
42.9 ( 1.7 nM
173 ( 9.9 pM
181 ( 8.5 pM
b
b
85.5 ( 4.0 μM
87.2 ( 17.1 μM
>100 μM
>100 μM
56.3 ( 6.4 nM
263 ( 9.2 nM
13.1 ( 1.8 nM
b
b
b
>100 μM
1.23 ( 0.21 μM
b
b
b
2.23 ( 0.23 μM
28.6 ( 3.2 μM
a
b
IC50s of positive controls. These IC50s were calculated from the later part of the progress curves, as time-dependent inhibition was evident (see text).
to T cells stimulates their proliferation and the release of
DCs were chosen for this study as they are the most potent
antigen-presenting cells and have much higher cathepsin E
2
7
certain inflammatory cytokines (vide infra). We therefore
investigated the effect of compound 1 on human peripheral
blood mononuclear cells (PBMCs). PBMCs are a mixture
containing various APCs (dendritic cells, B cells, and macro-
phages) and T cells. We examined the effect of 1 on T cells
2
8
expression than other APCs. The main targets of antigen
presentation are TH cells, which go on to orchestrate the
ensuing immune response. We therefore decided to use an
enriched population of these cells as the responders in our
assay. Differentiated DCs were cultured in the presence of
antigen (TTc), phorbol 12-myristate 13-acetate (PMA), and
TH cells for 5 days. Compound 1 was able to reduce T cell
proliferation in a dose-dependent manner (Figure 8B). TTc
and PMA alone (i.e., in the absence of DCs) were unable to
increase T cell proliferation, and thus this effect of 1 is
dependent on DCs. In these experiments, 1 was also able to
inhibit upregulation of interleukin-17 (IL-17) (Figure 8C)
and interferon-γ (IFN-γ) (Figure 8D) in response to antigen
presentation.
þ
using flow cytometry to gate for CD3 lymphocytes (T cells).
Compound 1 (10 μM) was able to significantly reduce T cell
proliferation in response to exogenous antigen (tetanus toxin
C-fragment, TTc, Figure 8A). In the same experiment, T cell
viability was unaffected (data not shown).
We then investigated the effect of 1 on the interaction
þ
between monocyte-derived dendritic cells (DCs) and CD4
T cells (T helper cells, T ) in a mixed lymphocyte reaction
H
(MLR). The first experiments were autologous MLRs,
where DCs and T cells came from the same human donor.

5
740 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
To determine whether 1 had any effect on T cell recogni-
tion of foreign MHC II proteins, we carried out the same
Kwan et al.
pepstatin A, input structures of the ligands had all amide
bonds trans, except the proline amide, for which separate cis
and trans structures were produced. For cathepsin D, the
crystal structure of pepstatin A bound to this enzyme was
experiment with DCs and T cells from different donors. We
H
found that 1 had no effect on DC-stimulated proliferation in
an allogeneic MLR (data not shown). This is likely because T
cells were recognizing nonself MHC II proteins on the sur-
face of DCs. Even though proliferation was not reduced,
we still observed a significant downregulation of IL-17 and
IFN-γ production (panels A and B of Figure 9, respectively).
Molecular Docking. To gain some insight into the struc-
tural basis for the selectivity of the grassystatins for cathepsin
E over D, compounds 1 and 3 were docked into these two
enzymes (Figure 10). For both proteins, compounds 1 and 3
3
1
used for docking (PDB code 1LYB). We were able to
successfully redock pepstatin A into this structure (Figure
S2A, Supporting Information) prior to docking our ligands.
For cathepsin E, homology modeling was used to obtain an
appropriate starting structure, because the only crystal
structure published is of an early activation intermediate
2
9
3
2
(PDB code 1TZS; see Experimental Section). For the
homology modeling, the structure of mature porcine pepsin
3
3
(PDB code 4PEP), a protein with highly homologous
primary sequence, was used as a template for the primary
sequence of cathepsin E, taken from 1TZS (see Experimental
Section). Pepstatin A was successfully docked into the
resulting structure (Figure S2B, Supporting Information)
before docking of 1 and 3 was attempted.
3
0
were successfully docked using AutoDock Vina 1.0, with
the ligand treated as fully flexible (see Experimental Section).
To be consistent with the mode of cathepsin inhibition by
Many of the putative hydrogen bond interactions sug-
gested by the crystal structure of pepstatin A bound to
3
1
cathepsin D (PDB code 1LYB) are also present in the
model of compound 1 bound to this enzyme (Figure 10A).
The reduced affinity of 1 versus pepstatin A may be due to
the presence of a polar residue at P2 (Val in pepstatin A and
Asn in 1). Cathepsin D has an established preference for
hydrophobic residues in this position, although it is some-
3
4
what tolerant of polar residues here. In our docked con-
formation of 1, the Asn side chain is curled down in order to
interact with Ser-80 in the flap and to avoid the hydrophobic
residues Met-307 and Met-309. The docked structure of
grassystatin A (1) in cathepsin E (Figure 10B) shows this
unit interacting with the polar residue Gln-303, which re-
places Met-307 in cathepsin D. This could be one reason for
an increased affinity for cathepsin E versus cathepsin D.
Another factor could be the numerous hydrogen bond
interactions possible between the O-Me-Pro unit of 1 with
Gln-85 in cathepsin E (Figure 10B). It is probably not
possible to form so many hydrogen bonds with the equiva-
lent residue in cathepsin D (His-77).
Grassystatin C (3) was less potent than grassystatins A and
B (1 and 2, respectively) against both cathepsins D and E.
One potential reason for this is the absence of the terminal N,
N-Me -Val, which could act as either a hydrogen bond donor
2
(
if protonated) or acceptor (if unprotonated). N,N-Dimethyl
Figure 6. Dose-response curves for grassystatin A (1) against
groups have in the past been used to improve the affinity of
ligands that interact with acidic groups, such as a key Asp
residue in the H1 receptor (the native ligand histamine
(
%
A) cathepsins D and E and (B) ADAM9, ADAM10, and TACE.
activity figures are derived from initial reaction slope for (A) and
end slope for (B), due to apparent time-dependent inhibition (see
Figure S1, Supporting Information).
3
5
instead has a primary amine). In cathepsin D, there are
Figure 7. Activities of grassystatin A (1) and pepstatin A against MCF7 cellular proteases as determined with a cathepsin D/E substrate (see
text). (A) MCF7 cell lysate was directly treated with grassystatin A (1) or pepstatin A. (B) MCF7 cells were treated with grassystatin A (1) or
pepstatin A for 1 h. Cells were lysed, and the protease activity of the lysates was assessed. The asterisk denotes significance of P < 0.05 using a
two-tailed t test. Data points are shown ( SD.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5741
Figure 8. Downregulation of antigen presentation of T cells and T
(
H
cells after treatment with grassystatin A (1) on activated PBMC and DC.
A) Downregulation of the activation of CD3 T cells in whole PBMCs after treatment with different concentrations of 1. (B) Downregulation
þ
of T
H
activation (proliferation) by the addition of different concentrations of 1. (C, D) Effect of 1 on the production of intracellular IFN-γ
cells induced by autologous activated DC. “Ctrl” refers to T cells that did not have DCs added to them. The asterisk
denotes significance of P < 0.05 using a two-tailed t test. Data points are shown ( SEM.
(
C) and IL-17 (D) by T
H
similar structure was produced, where the basic nitrogen was
close to Tyr-20 (not shown). Thus, this unit may serve to
anchor the inhibitor in the correct position within the bind-
ing cleft. Indeed, we observed that when docking into the
36
same protein structure, more spurious structures were
produced for grassystatin C (3) than for grassystatin A (1).
In these situations, because 3 has fewer rotatable bonds than
1, it is unlikely that the search parameters would be insuffi-
cient for 3 and not 1.
Discussion
Consistent with the notion that natural products are com-
pounds preoptimized by evolution to act on specific biological
targets, the marine cyanobacterium Lyngbya confervoides has
yielded potent and selective protease inhibitors. Grassystatins
10
A-C (1-3) and the related compound tasiamide B
Figures 1 and 2) all contain a statine or statine-like unit with
(
the same configuration (3S,4S). This is also the same configu-
rationasthe statineunitsinpepstatinA. It has previouslybeen
shown that the configuration of C-3 in the central statine is
important to pepstatin A’s inhibitory activity against aspartic
3
7
proteases.
Like pepstatin A, 1-3 inhibited cathepsins D and E,
although they were selective for cathepsin E (∼20- to ∼38-
fold), while pepstatin A did not discriminate between these
proteases. In pepstatin A, the central statine unit is the
pharmacophore of inhibition that binds to cathepsin D at
Figure 9. Effect of 1 on the production of (A) intracellular IFN-γ
and (B) IL-17 by T cells induced by allogeneic activated DC in an
MLR. “Ctrl” refers to T cells that did not have DCs added to them.
The asterisk denotes significance of P < 0.05 using a two-tailed
t test. Data points are shown ( SEM.
H
0
31
the P1-P1 site (see Figure 2 for sites). If the binding mode is
the same for 1-3, then the units flanking the statine unit
confer the differential activity for cathepsins D and E. Cathe-
psin D strongly favors hydrophobic amino acids in the P2
position, compared to polar units such as Asn in 1 and 2, or N-
several polar residues within reach of this unit (Tyr-10, Gln-
1
4, Thr-125, Lys-130, Gln-258, and Gln-260). The situation
is similar with cathepsin E, where there are a number of polar
residues in the same pocket (Tyr-20, Glu-24, Glu-27, Glu-
3
8
34
121, Asp-125, Glu-256, and Tyr-257). In the docked struc-
ture of 1 shown in Figure 10B, the basic nitrogen of N,
N-Me -Val is close to Gln-121. In the same run, another
2
Me-Gln in 3, which are more tolerated by cathepsin E.
This could explain why these compounds are less potent
inhibitors against cathepsin D compared to pepstatin A,

5
742 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
Kwan et al.
Figure 10. Docked structures of grassystatins A (1) and C (3) with cathepsins D and E. For each the protein is shown in green, and possible
hydrogen bonds are shown as dotted yellow lines. (A) Docked conformation of grassystatin A (1, yellow) with cathepsin D. (B) Docked
conformation of grassystatin A (1) with cathepsin E. (C) Docked conformation of grassystatin C (3, pink) with cathepsin D. (D) Docked
conformation of grassystatin C (3) with cathepsin E. For docked conformations of pepstatin A, see Figure S2, Supporting Information.
which has valine at P2 (Figure 2). Both cathepsins D and E
0
bend in the progress curve of the reaction within a time scale
40
where the uninhibited reaction is still linear. Statine-based
slow-binding inhibitors of aspartic proteases have been de-
allow polar (but not charged) units at position P2 , and
4,39
3
hydrophobic units such as leucine are also allowed.
There-
4
1
fore, the changefromThr in1 and 2 toIle in 3may not account
for its lower activity. The putative hydrogen bond between
Asn-NH and Ser-80-OH may be particularly important to
binding, and this interaction is not possible in 3 because the
R-nitrogen of Gln is methylated. Compound 3 does not
scribed. There are several examples of slow-binding inhibi-
tors of zinc metalloproteases; for example, the antihyper-
tensive drugs captopril and enalapril are both slow-binding
42
inhibitors of angiotensin converting enzyme (ACE). There
43
are also some slow-binding inhibitors of MMPs. The reason
for slow binding in these cases may be the expulsion of a
tightly bound, catalytically active water molecule from the
possess terminal units N,N-Me -L-Val-L-Hiva. The basic
2
nitrogen of N,N-Me -Val is probably able to interact with
2
4
0
acidic residues in both cathepsins D and E (see Results). It has
previously been shown that occupation of the S5 subsite of
active site. With slow-binding inhibitors, the onset of inhibi-
tion depends on the preincubation time of the test compound
with the enzyme. In the large-scale screen, this time may have
been longer than desirable, leading to an apparent lower IC50
for compound 1 against ADAM9 and ADAM10.
3
4
cathepsin E with Lys increases substrate turnover. Occupa-
tion of this site by positively chargedresidues may thereforebe
key to inhibitor binding.
Three metalloproteases in the ADAM family were identi-
fied in the primary screen of compound 1 (ADAM9,
ADAM10, and TACE). Only one of the hits (TACE), how-
ever, could be replicated in a dose-response assay (Table 3).
Inhibition of TACE by compounds 1-3 was concentration-
and time-dependent (Figure 6B), with IC s of 1.23, 2.23, and
Consistent with previously reported experiments using
2
4,26a,b,d
pepstatin A,
we found that 1 was able to reduce
antigen-stimulated T cell proliferation in PBMCs (Figure 8A)
and antigen presentation by DCs to TH cells (Figure 8B).
Concurrently, we found that 1 reduced production of IL-17
and IFN-γ by T cells (panels C and D of Figure 8, re-
spectively). IFN-γ is a proinflammatory molecule and the
5
0
28.6 μM, respectively. A slow onset of inhibition indicates
slow binding of the inhibitor and is apparent by a noticeable
signature cytokine produced by T 1 cells that, among other
H

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5743
2
7
effects, activates macrophages. These cells are strongly
involved in cellular immunity against cancer and intracellular
pathogens such as viruses but are also involved in the etiology
a SpectraMax M5 (Molecular Devices), and IR data were
obtained on a Bruker Vector 22 instrument. The purities of
compounds 1-3 were determined to be greater than 95% by
1
HPLC, MS, and NMR. H and 2D NMR spectra in CDCl for 1
4
4
3
of transplant rejection. IL-17 is another proinflammatory
cytokine, produced by a recently described subset of T cells,
1
3
and 2 were recorded on a Bruker 500 MHz spectrometer.
and APT spectra for 1 were recorded on a Bruker 600 MHz
C
4
5
T 17 cells. T 17 cells and IL-17 have been implicated in a
H
H
1
Avance spectrometer. H and 2D NMR spectra in CDCl
3
for 3,
for 1, were
number of autoimmune and allergic diseases such as rheuma-
1
and also H and 2D NMR spectra in DMSO-d
4
5
6
toid arthritis and asthma. Because of their involvement
in proinflammatory disorders and the association of such
pathologies with the activation of T cells by antigen presenta-
tion, we decided to investigate the contribution of 1 to
the modulation or downregulation of proinflammatory
cytokines.
collected on a Bruker Avance II 600 MHz spectrometer using a 1
mm triple-resonance high-temperature superconducting cryo-
5
2
genic probe. Spectra were referenced to residual solvent
signals [δH/C 7.26/77.0 (CDCl ) and δH/C 2.49/39.5 (DMSO-
)]. HMQC and HSQC experiments were optimized for 145 Hz,
3
d
6
and HMBC experiments were optimized for 7 Hz. HRESI/
APCIMS data were recorded on an Agilent LC-TOF mass
spectrometer equipped with an APCI/ESI multimode ion source
detector in positive ion mode. LC-MS data were obtained using
an API 3200 (Applied Biosystems) equipped with a Shimadzu
LC system. ESIMS fragmentation data were obtained on an
API 3200 by direct injection with a syringe driver. Flow cyto-
metry was carried out on a FACSCalibur flow cytometer using
CellQuest software (BD Biosciences, Heidelberg, Germany).
Figures of docked ligands were prepared using PyMol.
Exogenous antigens are internalized by APCs and proteo-
lytically cleaved within endosomes, before they are presented
2
7
on the cell surface bound to MHC class II proteins. The
invariant chain (Ii) is a chaperone that prevents endogenous
peptides from binding to MHC class II proteins while they are
4
6
transported from the endoplasmic reticulum to endosomes.
Ii undergoes several cleavage steps both before and after
entering the endosome. Its eventual removal allows antigens
to bind MHC II for subsequent presentation. Cysteine
proteases, such as cathepsin S, have a well-established role
Extraction and Isolation. Samples of L. confervoides were
collected off Grassy Key on May 26, 2004, and fractionated as
1
2
4
7
previously described. The silica gel fraction eluting with 100%
methanol was purified by preparative reversed-phase HPLC
in Ii cleavage. There have been conflicting reports, however,
of whether an aspartic protease-dependent cleavage is also
(
Phenomenex Luna 10u C18 AXI, 100 ꢀ 21.2 mm, 10.0 mL/min;
UV detection at 220 and 254 nm), using a MeOH-H O linear
gradient (60-100% over 30 min, then 100% MeOH for 10 min),
to give impure 1 and 2 at t 24.3 and 24.9 min, respectively.
4
8
required for MHC II maturation. Mari ꢀc et al. found that
aspartic protease inhibitors reduced cleavage of MHC II-Ii
complexes in B-lymphoblastoid cells. Similarly, Zhang et
2
R
26d
al. found that pepstatin A induced an accumulation of Ii-
containing cleavage intermediates both in a murine B cell line
These were purified using a different column [Phenomenex
Ultracarb 5u ODS (30), 250 ꢀ 10.0 mm, 2.0 mL/min; UV
detection at 220 and 254 nm] using the same linear gradient to
furnish compounds 1, tR 33.8 min (5.7 mg), and 2, tR 35.1 min
26b,c
and in live mice. Two other reports
found that pepstatin A
could inhibit presentation of antigens when cells (murine
microglia and human DCs) were treated with intact antigens
but not when treated with precleaved epitopic peptides.
(
1.3 mg).
Samples of the same species were collected off Key Largo on
4
9
May 8, 2003. A voucher specimen is maintained at the Smith-
sonian Marine Station. The freeze-dried organism was extracted
with EtOAc-MeOH (1:1) to afford the nonpolar extract, which
was directly fractionated by silica gel column and eluted with
Recently, Costantino et al. have presented results suggesting
that the role of different enzymes in Ii cleavage is highly
variable and there is a large degree of redundancy. Our own
results suggest that 1 is not able to inhibit MHC II-Ii
cleavage, as DCs treated with 10 μM 1 in the allogeneic (but
not autologous) MLR were still able to stimulate T cell
proliferation. Furthermore, 1 was abletodownregulate proin-
flammatory cytokines in both types of assays. This indicates
that either presentation of TTc is inhibited in both cases, with
inhibition of cytokine production being a consequence of this,
or that 1 has a direct effect on cytokine expression. While
further studies are needed, it opens the possibility for this
compound to be used in the study of antigen presentation and
also of the role of cathepsin E in inducing proinflammatory
responses.
2 2
increasing concentrations of 2-propanol in CH Cl . The frac-
tion eluting with 100% 2-propanol (665.8 mg) was subjected to
preparative reversed-phase HPLC [column, Luna C18(2) 100A
AXI, 10 μm (100 ꢀ 21.20 mm), Phenomenex; flow rate 10.0 mL/
min; detection by UV at 220 and 254 nm] using a linear
MeOH-H O gradient (60-100% MeOH over 30 min, then
2
1
compounds, a minor peak eluted at t 17.2 min, which was then
00% MeOH for 5 min). In addition to previously isolated
R
deconvoluted using different conditions [column, ODS-AQ
(
10 ꢀ 250 mm), YMC; flow rate, 2.0 mL/min; detection by
UV at 220 and 254 nm] using the same linear MeOH-H O
2
gradient, to furnish pure 3, t 27.3 min (1.0 mg).
Grassystatin A (1). Colorless amorphous solid; [R]
R
2
0
-4.4 (c
D
In conclusion, we believe that compounds 1-3 could prove
to be valuable probes for the study of cathepsin E function.
Previously, only one inhibitor selective for cathepsin E was
max
0.08, MeOH); UV (MeOH) λ (log ε) 206 (4.93), 258 (4.03),
324 (3.2); IR (film) νmax 3291 (br), 3068 (w), 3054 (w), 3019 (w),
2955, 2937, 2925, 2914, 2851, 2360, 2342, 1733, 1646, 1540, 1457,
-
1
374 (w), 1265, 1109 (w), 1023 (w), 896 (w), 739 cm ; NMR
5
0
1
data, H NMR, C NMR, APT, COSY, HMQC, HMBC,
known, a protein from the roundworm Ascaris lumbricoides.
1
13
Because this inhibitor is not widely available, studies into the
function of cathepsin E have had to rely on nonselective
inhibitors such as pepstatin A or cathepsin D knockout
1
3
ROESY, TOCSY in CDCl , see Table 1, H NMR, COSY,
edited HSQC, HMBC, ROESY in DMSO-d , see Table S1,
6
þ
þ
2þ
2
Supporting Information; HRESI/APCIMS m/z [M þ Na]
5
1
animals/cells. To the best of our knowledge, grassystatins
A-C (1-3) are the only other natural protease inhibitors that
are selective for cathepsin E over cathepsin D.
1
1
5
5
196.6812 (calcd for C58
174.6988 (calcd for C58
H
95
96
9
N O
16Na, 1196.6794), [M þ H]
H
9
N O
16 1174.6975), [M þ H þ Na]
þ
16
98.8455 (calcd for C H N O Na, 598.8436), [M þ 2H]
5
8
96
9
87.8544 (calcd for C58
Grassystatin B (2). Colorless amorphous solid; [R]
H
97
N
9
O
16, 587.8527).
2
0
D
-5.0 (c
Experimental Section
0
.1, MeOH); UV (MeOH) λ (log ε) 202 (4.47), 266 (2.96), 320
max
General Experimental Procedures. Optical rotation was mea-
sured on a Perkin-Elmer 341 polarimeter. UV was measured on
(2.45); IR (film) νmax 3276 (br), 3079 (w), 3054 (w), 3017 (w),
2961, 2927, 2874, 2360, 2342, 1752, 1732, 1690, 1627, 1549, 1493

5
744 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
Kwan et al.
(
w), 1463 (w), 1436 (w), 1389 (w), 1369 (w), 1267 (w), 1207 (w),
standards were as follows: L-Ile (12.3; 132f86), L-allo-Ile (13.4),
D-allo-Ile (57.5), D-Ile (70.5), L-Leu (13.1; 132f86), and D-Leu
(51.7). The MS parameters used were as follows: DP 31.0, EP
8.0, CE 17.3, CXP 3.1, CUR 35, CAD medium, IS 4500, TEM
750, GS1 65, and GS2 65.
-
1
179 (w), 1124 (w), 1023 (w) cm ; NMR data, H NMR,
1
1
COSY, HMQC, HMBC, ROESY, TOCSY in CDCl
3
, see
þ
Table 1; HRESI/APCIMS m/z [MþK] 1226.6687 (calcd for
þ
þ
C
59
H
97
N
9
O
16K, 1226.6690), [M þ Na] 1210.6936 (calcd for
C H N O Na, 1210.6951), [M þ H] 1188.7119 (calcd for
Hmpa in the hydrolysate of 3 was detected in negative ion
mode [column, Chirobiotic TAG (4.6 ꢀ 250 mm), Supelco;
5
9
97
9
16
2
þ
C
59
H
98
N
9
O
N
16, 1188.7131), [M þ H þ Na] 605.8516 (calcd
16Na, 605.8515).
Grassystatin C (3). Colorless amorphous solid; [R]
for C59
H
98
9
O
solvent, MeOH-10 mM NH OAc (40:60, pH 5.35); flow rate,
4
2
0
-21.9
0.5 mL/min; detection by ESIMS in negative ion mode (MRM
D
(
c 0.04, MeOH); UV (MeOH) λ
(log ε) 203 (4.74), 260 (2.89),
scan)]. The MS parameters used were as follows: DP -35.0, EP
-8.0, CE -17.9, CXP -1.7, CUR 40, CAD medium, IS -4500,
TEM 750, GS1 65, and GS2 65. (2R,3S)-Hmpa from the
R
hydrolysate eluted at tR 6.4 min. The retention times (t , min;
MRM ion pair, parentfproduct) of the authentic standards
were as follows: (2S,3R)-Hmpa (6.0; 131f85), (2S,3S)-Hmpa
(6.2; 131f85), (2R,3S)-Hmpa (6.4; 131f85), and (2R,3R)-
Hmpa (7.0; 131f85). The hydrolysate was examined under
different HPLC conditions in order to confirm this assignment
max
3
2
1
20 (2.17); IR (film) νmax 3307 (br), 3078 (w), 3054 (w), 3016 (w),
659, 2927, 2904, 2874, 2361, 2340, 1742, 1635, 1531, 1462, 1442,
-
1
410, 1368, 1285 (w), 1199 (w), 1047 (w) cm ; NMR data, H
1
NMR, COSY, edited HSQC, HMBC, ROESY, TOCSY in
CDCl
1
8
þ
þ
3
, see Table 2; HRESI/APCIMS m/z [M þ Na]
009.5941 (calcd for C50
58.5322 (calcd for C H N O , 858.5341).
H
82
8
N O
12Na, 1009.5950), [M - 128]
4
4
72
7
10
Acid Hydrolysis and Chiral Amino Acid Analysis by LC-MS
and HPLC. A sample of 1 (100 μg) was treated with 6 N HCl at
10 °C for 24 h. The hydrolysate was concentrated to dryness,
reconstituted in 100 μL of H O, and then analyzed by chiral
HPLC [column, Chirobiotic TAG (4.6 ꢀ 250 mm), Supelco;
solvent, MeOH-10 mM NH OAc (40:60, pH 5.23); flow rate,
.5 mL/min; detection by ESIMS in positive ion mode (MRM
scan)]. L-Thr, L-Leu, L-Pro, N,N-Me -L-Val, and N-Me-D-Phe
eluted at tR 7.2, 9.0, 14.4, 27.0, and 45.4 min, respectively. The
retention times (t , min; MRM ion pair, parentfproduct) of the
[
column, Chiralpak MA (þ) (4.6 ꢀ 50 mm), Daicel Chemical
1
Industries, Ltd.; solvent, 2 mM CuSO -CH CN (85:15); flow
4
3
2
rate, 1.0 mL/min; detection by UV absorption at 254 nm].
2R,3S)-Hmpa from the hydrolysate eluted at tR 15.4 min.
The retention times (t , min) of the authentic standards were
as follows: (2R,3S)-Hmpa (15.4), (2R,3R)-Hmpa (17.9),
2S,3R)-Hmpa (22.7), and (2S,3S)-Hmpa (27.5). Under these
(
4
R
0
2
(
conditions, all other units eluted at tR <6.5 min.
R
Base Hydrolysis to Determine Configuration of Hiva Units.
The acid hydrolysate of 1 was analyzed by chiral HPLC [column,
Chirobiotic TAG (4.6 ꢀ 250 mm), Supelco; solvent, MeOH-
authentic amino acids were as follows: L-Thr (7.2; 120f74),
L-allo-Thr (7.5), D-Thr (8.6), D-allo-Thr (11.9), L-Pro (14.4;
1
16f70), D-Pro (39.5), L-Leu (9.0; 132f86), D-Leu (20.6), N-
Me-L-Phe (25.0; 180f134), N-Me-D-Phe (45.4), N,N-Me -L-Val
27.0; 146f100), and N,N-Me -D-Val (69.8). The assignment of
4
10 mM NH OAc (60:40, pH 5.63); flow rate, 0.5 mL/min; detec-
2
tion by ESIMS in negative ion mode (MRM scan)]. Both L-Hiva
and D-Hiva were detected at t 6.0 and 6.4 min, respectively. The
(
2
R
L-Thr was confirmed by coinjection of the hydrolysate with
L-allo-Thr and L-Thr. The MS parameters used were as follows:
DP 31.0, EP 8.0, CE 17.3, CXP 3.1, CUR 35, CAD medium, IS
R
retention times (t , min; MRM ion pair, parentfproduct) of the
authentic standards were as follows: L-Hiva (6.0; 117f71) and
D-Hiva (6.4). A sample of 1 (100 μg) was suspended in 80 μL of
MeOH-0.5 N NaOH (1:1) and left to stand at room temperature
for 72 h. The solution was neutralized by the addition of 20 μL of 1
N HCl and was then analyzed by chiral HPLC-MS as before. Only
4
500, TEM 750, GS1 65, and GS2 65. L-Ala was also detected in
positive ion mode, at t 8.0 min, but with slightly different MS
R
conditions. The retention times (t , min; MRM ion pair,
R
parentfproduct) of the authentic standards were as follows:
L-Ala (8.0; 90f44), D-Ala (14.6). The MS parameters used were
as follows: DP 21.0, EP 8.0, CE 15.0, CXP 5.0, CUR 50, CAD
medium, IS 4500, TEM 750, GS1 65, and GS2 65. Asp was only
detected weakly in positive ion mode, and consequently negative
ion mode was used with the same LC conditions. L-Asp eluted
L-Hiva was detected at tR 6.0 min. The retention times (t , min;
R
MRM ion pair, parentfproduct) of the authentic standards were
as follows: L-Hiva (6.0; 117f71) and D-Hiva (6.4). The MS
parameters used were as follows: DP -30.0, EP -3.0, CE -17.3,
CXP -2.0, CUR 45, CAD medium, IS -4500, TEM 650, GS1 50,
and GS2 25.
at t
R
6.1 min, indicating that the configuration of the Asn unit
was L. The retention times (t , min; MRM ion pair,
Modified Marfey’s Analysis to Determine Configuration of
Statine Units. Samples of both 1 and 3 (35 μg) were subjected to
acid hydrolysis, derivatized with L-FDLA as described pre-
viously, and analyzed by reversed-phase HPLC [column, Allti-
ma HP C18 HL (4.6 ꢀ 250 mm), 5 μm, Alltech; flow rate,
0.5 mL/min; detection by ESIMS in negative ion mode (MRM
scan, 468f408)], using a linear gradient of MeOH in H O (both
2
containing 0.1% HCOOH, 40-100% MeOH over 50 min). Two
peaks, corresponding to (3S,4S)-Sta-L-FDLA and (3R,4S)-Sta-
R
parentfproduct) of the authentic standards were as follows:
L-Asp (6.1; 132f88) and D-Asp (6.8). The MS parameters used
were as follows: DP -30.0, EP -5.0, CE -18.5, CXP -13.0,
CUR 30, CAD high, IS -4500, TEM 750, GS1 65, and GS2 65.
A sample of 3 was treated with 6 N HCl at 110 °C for 24 h. The
hydrolysate was concentrated to dryness, reconstituted in
1
00 μL of H O, and then analyzed by chiral HPLC [column,
2
Chirobiotic TAG (4.6 ꢀ 250 mm), Supelco; solvent, MeOH-
L-FDLA, were observed in both samples in a 1:1 ratio at t
R
and 35.9 min, respectively. The retention times (t , min) of the
R
35.5
1
detection by ESIMS in positive ion mode (MRM scan)].
0 mM NH OAc (40:60, pH 5.33); flow rate, 0.5 mL/min;
4
authentic standards were as follows: (3S,4S)-Sta-L-FDLA
(35.5), (3R,4S)-Sta-L-FDLA (35.9), (3S,4S)-Sta-D-FDLA
[corresponding to (3R,4R)-Sta-L-FDLA, 45.7], and (3R,4S)-
Sta-D-FDLA [corresponding to (3S,4R)-Sta-L-FDLA, 46.4].
The MS parameters used were as follows: DP -60.0, EP -7.0,
CE -28.0, CXP -7.4, CUR 40, CAD high, IS -4500, TEM 750,
GS1 40, and GS2 40.
N-Me-L-Glu, L-Ile, L-Leu, L-Pro, and N-Me-D-Phe eluted at tR
6
.0, 8.3, 8.6, 13.3, and 41.7 min, respectively. The retention times
(t , min; MRM ion pair, parentfproduct) of the authentic
R
standards were as follows: N-Me-L-Glu (6.0; 162f44), N-Me-D-
Glu (15.8), L-Ile (8.3; 132f86), L-allo-Ile (8.5), D-allo-Ile (19.6),
D-Ile (22.2), L-Leu (8.6; 132f86), D-Leu (19.8), L-Pro (13.3;
1
16f70), D-Pro (35.2), N-Me-L-Phe (23.2; 180f134), and
Protease Inhibition Screen. Compound 1 was added into the
reaction buffer containing enzyme by acoustic droplet ejection
(Echo 550; Labcyte Inc., Sunnyvale, CA) such that the final
concentration was 10 μM. After incubation at room tempera-
ture for 10-15 min, the substrate was added, after which
fluorescence at each relevant ex/em wavelength was measured
every 5 min for 2 h. The substrate alone in the reaction buffer
served as background. The activity of 1 was evaluated by
N-Me-D-Phe (41.7). To further separate the isobaric Ile and
Leu units, different LC conditions were employed [column,
Chirobiotic TAG (4.6 ꢀ 250 mm), Supelco; solvent, MeOH-
1
4
0 mM NH OAc (90:10, pH 5.65); flow rate, 0.5 mL/min;
detection by MS (MRM scan)]. L-Ile and L-Leu eluted at tR
2.3 and 13.1 min, respectively. The retention times (t , min;
MRM ion pair, parentfproduct) of the authentic amino acid
1
R

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5745
obtaining percent enzyme activity relative to the slope of no
inhibitor control. Each enzyme assay was performed in dupli-
cate by Reaction Biology Corp. (Malvern, PA). For individual
buffers, substrates, and positive controls, see Table S2, Support-
ing Information.
GM-CSF (Leukine, Berlex, WA) and 20 ng/mL IL-4 (BD
Biosciences) were added to the culture for 7 days.
Isolation of CD4 T Helper Cells from PBMCs. CD4 T cells
þ
þ
were purified from PBMCs by negative selection using the
“Human CD4 T Cell Enrichment” kit (EasySep, StemCell
þ
Protease Inhibition Assays to Determine IC50 Values. Assays
were carried out in the same way as was in the protease screen,
with 3-fold dilution series of 1-3 starting at 10 or 100 μM as
appropriate. At least 10 concentrations were used in each assay
series to sufficiently define IC50 values (solvents: DMSO for 1
and 2, EtOH for 3). Assays were carried out by Reaction Biology
Corp. (Malvern, PA). Enzyme activity (in percent), calculated
as above, was used to determine IC50 values with nonlinear
regression in GraphPad Prism (GraphPad Software, Inc., La
Jolla, CA).
Technologies, Vancouver, BC, Canada) by following the manu-
facturer’s protocol. Briefly, cells were resuspended in magne-
sium-free 1ꢀ PBS with 2% fetal bovine serum in 12 ꢀ 75 mm
7
polystyrene tubes at 5 ꢀ 10 cells/mL. Enrichment cocktail
(50 μL/mL) was added, incubated for 10 min at room tempera-
ture, and then followed by 100 μL/mL nanoparticle cocktail and
extra incubation. Afterward, the total volume was brought to
2.5 mL, and the tube was put into an EasySep magnet. After
5 min incubation to allow for magnetic beads to attach to the
side of the tube, the contents were decanted into a clean tube.
The beads were washed once more to increase cell purity. Cells
were subsequently used in MLR reactions as described below.
Mixed Lymphocyte Reactions (MLRs). Autologous or allo-
Inhibition of Cellular Cathepsins D and E by Grassystatin A
1). To measure the in vitro inhibition of cellular cathepsins by 1
(
and pepstatin A, kinetic assays were carried out using lysate
prepared from untreated MCF7 cells. MCF7 cells were cultured
in Dulbecco’s modified Eagle medium (DMEM; Invitrogen)
containing 10% fetal bovine serum (FBS; HyClone, Logan,
þ
geneic enriched CD4 T cells were labeled for 15 min with
200 nM carboxyfluorescein diacetate succinimidyl ester (CFSE;
Molecular Probes, Eugene OR) according to manufacturer’s
6
UT), in a humidified atmosphere containing 5% CO
2
at 37 °C.
protocol and cultured at 1 ꢀ 10 cells/mL. Monocyte-derived
DCs (MoDC), pulsed with 5 μg/mL tetanus toxin C-fragment
Cells were lysed with NP-40 lysis buffer (1% NP-40, 50 mM
NaOAc, pH 4.0), and a portion of the lysate (50 μL) was
incubated at 37 °C in the presence of varying concentrations
of the test compounds and 10 μM Mca-Gly-Lys-Pro-Ile-Leu-
(
1
TTc; Roche Diagnostics, Mannheim, Germany) and phorbol
2-myristate 13-acetate (PMA; Promega, Madison, WI), were
added to the same culture at a ratio of 1:2 in combination with
increasing concentrations of 1. The cell mixture was allowed to
incubate for 5 days in a water-jacketed incubator at 37 °C.
Afterward, culture cells were collected for flow cytometric
analysis.
Flow Cytometry. Experimental pelleted cells were incubated
for 30 min at 4 °C with antibodies and washed with staining
buffer (PBS þ 2% BSA þ 0.1% sodium azide). Subsequently
intracellular staining was carried out by first adding Cytofix/
Cytoperm to increase the permeability of cells followed by
another round of staining before fixing with 2% paraformalde-
hyde. Quantitation was carried out using FCS Express (version
Phe-Phe-Arg-Leu-Lys(Dnp)-D-Arg-NH
4
2
in 50 mM NaOAc, pH
.0 (the same substrate used for cathepsins D and E in other
assays). The total volume of the reactions was 100 μL, and the
reaction was monitored by measuring the increase in fluore-
scence (λex=320 nm, λem=405 nm).
To provide some information of cell permeability, intact live
MCF7 cells were treated directly with test compounds, using a
2
4
procedure previously used for this purpose. Cells were cul-
tured as described above and were seeded into 24-well plates.
When cells reached 80-100% confluency, varying concentra-
tions of compound 1 or pepstatin A were added. After 1 h
incubation, the medium was removed, and the cells were washed
and then trypsinized for 10 min, in order to remove surface-
3
; De Novo Software, Los Angeles, CA) by gating for lympho-
cytes based on forward and side scatter properties followed by
analysis of the percentages of positively stained quadrants. A
total of 50000 cells were analyzed for each sample, and isotype-
specific immunoglobulin controls were run for each fluoro-
chrome. Stains used in PBMC experiments were CD3-APC,
25
bound inhibitor. Cells were collected by centrifugation and
lysed with NP-40 lysis buffer. The lysate (50 μL) from each well
was incubated at 37 °C with 10 μM Mca-Gly-Lys-Pro-Ile-Leu-
Phe-Phe-Arg-Leu-Lys(Dnp)-D-Arg-NH
pH 4.0, total volume 100 μL). The reaction was monitored
as above.
2
in 50 mM NaOAc
þ
CFSE, and 7-AAD (eBioscience); those used in CD4 T cells for
MLRs were CFSE, 7-AAD, AlexaFluo 647-conjugated anti-IL-
(
1
7, and PE-conjugated anti-IFN-γ (BD Biosciences).
Molecular Docking. Compounds 1 and 3 were docked into
Isolation of PBMCs and Culture of Monocytes and DCs.
PBMCs were isolated from buffy coats (leukopac, PBL) ob-
tained from Lifesouth Community Blood Center (Gainesville,
FL) by Ficoll-Hypaque density gradient centrifugation using
Lymphoprep (Axis-Shield, Norway). Briefly, the contents of the
buffy coat were diluted to three times its volume in sterile 1ꢀ
PBS, pH 7.4 (Gibco, CA). The dilution was layered onto the
Lymphoprep in a 2:1 ratio. The sample was then centrifuged for
cathepsin D using the crystal structure of pepstatin A in
cathepsin D as a starting point (PDB code 1LYB). AutoDock
31
30
Vina 1.0 was used for all docking runs. This program is 2
orders of magnitude faster than AutoDock 4 and thus renders
docking of flexible peptides with ∼25-50 rotatable bonds
possible on normal workstations in a reasonable time frame.
The program was able to reproduce the docked conformation of
2
5 min at room temperature and 250g. The PBMCs were
˚
pepstatin A in cathepsin D, with an rmsd of 0.977 A (Figure
5
3
collected at the interface, washed twice with PBS, and centri-
fuged each time for 10 min at 4 °C and 250g. Cell viability was
assessed by trypan blue exclusion. All cultures of human
PBMCs and derived cells were maintained in RPMI 1640
medium (Sigma, St. Louis, MO) supplemented with 2 mM L-
glutamine (Life Technologies, Paisley, Scotland), 5000 units/mL
penicillin (Sigma), 5000 units/mL streptomycin sulfate (Sigma),
and 10% (v/v) fetal bovine serum (Gibco). PBMCs were either
used in an experiment with 1 at various concentrations or
separated further.
S2A). In structures of grassystatins A (1) and C (3), all bonds
were treated as rotatable, except ring and amide bonds, and the
protein was treated as rigid. For 1, the terminal amine was
protonated to reflect its likely state at physiological pH. Apart
from the Pro amide bond, all amides in the ligand were set to
trans configuration. For each compound, separate structures
were made with the Pro amide bond either cis or trans. Docking
was carried out with an exhaustiveness value of 25 and a
maximum output of 100 structures. It was observed that Auto-
Dock was always able to propose docked structures with similar
calculated affinities (∼-9 to -7 kcal/mol), and so the output
structures were examined qualitatively. The primary criterion
used in choosing the best docked structures was the position
of the statine unit relative to the active site aspartates (Asp-33
and Asp-231), with reference to the bound conformation of
6
Monocytes were obtained by adhering 5 ꢀ 10 PBMCs/mL to
a flask for 2 h at 37 °C. After removing the supernatant
containing nonadherent cells, adherent monocytes were washed
with 1ꢀ PBS, pH 7.4. Complete media were added to the
remaining cells. To induce differentiation into DCs 50 ng/mL

5
746 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18
Kwan et al.
pepstatin A. The rationale for this is found in the numerous
crystal structures of pepstatin A and analogues bound to
many different aspartic proteases.
Matthew, S.; Rocca, J. R.; Paul, V. J.; Luesch, H. Lyngbyastatins 5-7,
potent elastase inhibitors from Floridian marine cyanobacteria, Lyng-
bya spp. J. Nat. Prod. 2007, 70, 1593–1600.
5
4
55
(
¨
5) Dittman, E.; Neilan, B. A.; Borner, T. Molecular biology of peptide
and polyketide biosynthesis in cyanobacteria. Appl. Microbiol.
Docking to cathepsin E was carried out in the same manner
but with some differences. There is only one crystal structure of
Biotechnol. 2001, 57, 467–473.
3
2
cathepsin E available (PDB code 1TZS), where the inhibitory
prodomain is still resident in the active site. This structure
probably corresponds to an early intermediate in the maturation
of the enzyme. In addition to the prodomain in the active site,
the N-terminal region (Lys-14 to Asp-22) is blocking the active
site tunnel so that the enzyme is in the closed conformation. A
structure more consistent with the mature enzyme had to be
produced in order to carry out effective docking. For this
purpose, homology modeling was carried out using the
(6) Kittendorf, J. D.; Sherman, D. H. The methymycin/pikromycin
pathway: a model for metabolic diversity in natural product
biosynthesis. Bioorg. Med. Chem. 2009, 17, 2137–2146.
(
7) (a) Morishima, H.; Takita, T.; Aoyagi, T.; Takeuchi, T.; Umezawa,
H. The structure of pepstatin. J. Antibiot. 1970, 23, 263–265.
(
b) Umezawa, H.; Aoyagi, T.; Morishima, H.; Matsuzaki, M.; Hamada,
M.; Takeuchi, T. Pepstatin, a new pepsin inhibitor produced by
Actinomycetes. J. Antibiot. 1970, 23, 259–262.
(8) Morishima, H.; Sawa, T.; Takita, T.; Aoyagi, T.; Takeuchi, T.;
Umezawa, H. Biosynthetic studies on pepstatin, biosynthesis of
(3S,4S)-4-amino-3-hydroxy-6-methylheptanoic acid moiety.
J. Antibiot. 1974, 27, 267–273.
5
6
SWISS-MODEL web server. Among the protein structures
in the PDB, human cathepsin E has the highest sequence
homology with porcine pepsinogen (PDB code 2PSG) and
(
9) Williams, P. G.; Yoshida, W. Y.; Moore, R. E.; Paul, V. J.
Tasiamide, a cytotoxic peptide from the marine cyanobacterium
Symploca sp. J. Nat. Prod. 2002, 65, 1336–1339.
5
7
3
3
its mature form, pepsin (PDB code 4PEP). The activation
intermediate structure for cathepsin E (1TZS) agrees very well
(
10) Williams, P. G.; Yoshida, W. Y.; Moore, R. E.; Paul, V. J. The
isolation and structure elucidation of tasiamide B, a 4-amino-3-
hydroxy-5-phenylpentanoic acid containing peptide from the marine
cyanobacterium Symploca sp. J. Nat. Prod. 2003, 66, 1006–1009.
11) (a) Zaidi, N.; Hermann, C.; Hermann, T.; Kalbacher, H. Emerging
functional roles of cathepsin E. Biochem. Biophys. Res. Commun.
2008, 377, 327–330. (b) Zaidi, N.; Kalbacher, H. Cathepsin E: a mini
review. Biochem. Biophys. Res. Commun. 2008, 367, 517–522.
12) Kwan, J. C.; Rocca, J. R.; Abboud, K. A.; Paul, V. J.; Luesch, H.
Total structure determination of grassypeptolide, a new marine
cyanobacterial cytotoxin. Org. Lett. 2008, 10, 789–792.
˚
with that for pepsinogen (2PSG, rmsd 0.784 A); therefore, the
structure of the corresponding mature enzyme (4PEP) is most
likely a good template for homology modeling. Indeed, the
structure obtained was in excellent agreement with 1LYB
(
˚
(
rmsd 0.833 A). Docking of pepstatin A into the homology
model was successful (see Figure S2B, Supporting Information).
The conformation obtained was close to that of pepstatin A
˚
bound to cathepsin D (rmsd 1.893 A). Grassystatin A (1) was
docked using the same protocol as above. For grassystatin C (3),
a larger value of exhaustiveness was used (50).
(
(
(
13) The minor signals present in the CDCl
conformers and not impurities, as these were not observed in the
spectra obtained in DMSO-d ; see Supporting Information.
14) The threonine hydroxyl was observed only in the NMR spectra
collected in DMSO-d ; see Supporting Information.
3
NMR spectra are due to
6
Acknowledgment. This research was supported by the
National Institutes of Health, NIGMS Grant P41GM-
6
(15) The presence of L-Asp in the hydrolysate is consistent with the
presence of L-Asn in the intact molecule, the primary amide having
undergone hydrolysis.
16) Marfey, P. Determination of D-amino acids. II. Use of a bifunc-
tional reagent, 1,5-difluoro-2,4-dinitrobenzene. Carlsberg Res.
Commun. 1984, 49, 591–596.
17) Matsumori, N.;Kaneno, D.;Murata, M.;Nakamura, H.;Tachibana,
K. Stereochemical determination of acyclic structures based on
carbon-proton spin-couping constants. A method of configuration
analysis for natural products. J. Org. Chem. 1999, 64, 866–876.
0
86210, and the University of Florida College of Pharmacy.
We acknowledge the NSF for funding through the External
User Program of the National High Magnetic Field Labora-
tory (NHMFL), which supported initial NMR studies at the
Advanced Magnetic Resonance Imaging and Spectroscopy
(
(
(AMRIS) facility in the McKnight Brain Institute of the
University of Florida (UF). The 600 MHz 1 mm triple-
resonance HTS cryogenic probe used for NMR of compound
(18) A modification of the pulse sequence in ref 19 was used; see Luesch
et al., 2001 (ref 3a).
1
in DMSO-d6 and 3 in CDCl3 was developed through
(
19) Uhrin, D.; Batta, G.; Hruby, V. J.; Barlow, P. N.; Ko
¨
v ꢀe r, K. E.
collaboration between UF, NHMFL, and Bruker Biospin.
We thank J. R. Rocca (UF) and K. Horiuchi (RBC) for
technical assistance. This is contribution 793 from the Smith-
sonian Marine Station.
Sensitivity- and gradient-enhanced hetero (ω half-filtered
1
)
TOCSY experiment for measuring long-range heteronuclear cou-
pling constants. J. Magn. Reson. 1998, 130, 155–161.
20) Preciado, A.; Williams, P. G. A simple microscale method for
determining the relative stereochemistry of statine units. J. Org.
Chem. 2008, 73, 9228–9234.
21) The presence of N-Me-L-Glu in the hydrolysate is consistent with
the presence of N-Me-L-Gln in the intact molecule, the primary
amide having undergone hydrolysis.
(
(
(
Supporting Information Available: Table S1, Table S2, Figure
S1, Figure S2, and NMR spectra for compounds 1-3. This
material is available free of charge via the Internet at http://
pubs.acs.org.
¨
22) Berglund, L.; Bjorling, E.; Oksvold, P.; Fagerberg, L.; Asplund, A.;
Szigyarto, C. A.-K.; Persson, A.; Ottosson, J.; Wern ꢀe rus, H.;
˚
Nilsson, P.; Lundberg, E.; Sivertsson, A.; Navani, S.; Wester, K.;
Kampf, C.; Hober, S.; Pont ꢀe n, F.; Uhl ꢀe n, M. A genecentric human
protein atlas for expression profiles based on antibodies. Mol. Cell.
Proteomics 2008, 7, 2019–2027.
References
1) (a) L oꢀ pez-Otı
´
( n, C.; Bond, J. S. Proteases: multifunctional enzymes
in life and disease. J. Biol. Chem. 2008, 283, 30433–30437. (b) Turk,
B. Targeting proteases: successes, failures and future prospects. Nat.
Rev. Drug Discovery 2006, 5, 785–799.
(23) The relative expressions of cathepsins D and E are given as 55%
and 54%, respectively, in the Human Protein Atlas (ref 22).
However, each of these figures is relative to the tissue/cell line that
has the highest expression of that particular protein. Therefore, from
these data alone, one cannot compare the expression of cathepsins D and
E directly in this cell line.
(24) Zaidi, N.; Burster, T.; Sommandas, V.; Hermann, T.; Boehm, B.
O.; Driessen, C.; Voelter, W.; Kalbacher, H. A novel cell penetrat-
ing aspartic protease inhibitor blocks processing and presentation
of tetanus toxoid more efficiently than pepstatin A. Biochem.
Biophys. Res. Commun. 2007, 364, 243–249.
(
(
2) Coussens, L. M.; Fingleton, B.; Matrisian, L. M. Matrix metallo-
proteinase inhibitors and cancer: trials and tribulations. Science
2002, 295, 2387–2392.
3) (a) Luesch, H.; Yoshida, W. Y.; Moore, R. E.; Paul, V. J.; Corbett,
T. H. Total structure determination of apratoxin A, a potent novel
cytotoxin from the marine cyanobacterium Lyngbya majuscula.
J. Am. Chem. Soc. 2001, 123, 5418–5423. (b) Taori, K.; Paul, V. J.;
Luesch, H. Structure and activity of largazole, a potent antiproliferative
agent from the Floridian marine cyanobacterium Symploca sp. J. Am.
Chem. Soc. 2008, 130, 1806–1807.
4) (a) Matthew, S.; Ross, C.; Rocca, J. R.; Paul, V. J.; Luesch, H.
Lyngbyastatin 4, a dolastatin 13 analogue with elastase and
chymotrypsin inhibitory activity from the marine cyanobacterium
Lyngbya confervoides. J. Nat. Prod. 2007, 70, 124–127. (b) Taori, K.;
(25) Richard, J. P.; Melikov, K.; Vives, E.; Ramos, C.; Verbeure, B.;
Gait, M. J.; Chernomordik, L. V.; Lebleu, B. Cell-penetrating
peptides, a reevaluation of the mechanism of cellular uptake.
J. Biol. Chem. 2003, 278, 585–590.
(
(26) (a) Burster, T.; Reich, M.; Zaidi, N.; Voelter, W.; Boehm, B. O.;
Kalbacher, H. Cathepsin E regulates the presentation of tetanus

Article
toxin C-fragment in PMA activated primary human B cells.
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 18 5747
Schreiber, R. D. Interferon-γ and cancer immunoediting. Immu-
nologic. Res. 2005, 32, 231–245.
Biochem. Biophys. Res. Commun. 2008, 377, 1299–1303. (b) Chain,
B. M.; Free, P.; Medd, P.; Swetman, C.; Tabor, A. B.; Terrazzini, N. The
expression and function of cathepsin E in dendritic cells. J. Immunol.
(45) Tesmer, L. A.; Lundy, S. K.; Sarkar, S.; Fox, D. A. Th17 cells in
human disease. Immunological. Rev. 2008, 223, 87–113.
(46) Janeway, C. A.; Travers, P.; Walport, M.; Shlomchik, M. Antigen
presentation to T lymphocytes. Immunobiology, the Immune Sys-
tem in Health and Disease, 6th ed.; Garland Publishing: New York,
2005; pp 169-202.
(47) Riese, R. J.; Mitchell, R. N.; Villadangos, J. A.; Shi, G.-P.; Palmer,
J. T.; Karp, E. R.; de Sanctis, G. T.; Ploegh, H. L.; Chapman, H. A.
Cathepsin S activity regulates antigen presentation and immunity.
J. Clin. Invest. 1998, 101, 2351–2363.
(48) Mari ꢀc , M. A.; Taylor, M. D.; Blum, J. S. Endosomal aspartic
proteinases are required for invariant-chain processing. Proc. Natl.
Acad. Sci. U.S.A. 1994, 91, 2171–2175.
(49) Constantino, C. M.; Hang, H. C.; Kent, S. C.; Hafler, D. A.;
Ploegh, H. L. Lysosomal cysteine and aspartic proteases are
heterogeneously expressed and act redundantly to initiate human
invariant chain degradation. J. Immunol. 2008, 180, 2876–2885.
(50) Keilov ꢀa , H.; Tom ꢀa ꢁs ek, V. Effect of pepsin inhibitor from Ascaris
lumbricoides on cathepsin D and E. Biochim. Biophys. Acta 1972,
284, 461–464.
2005, 174, 1791–1800. (c) Nishioku, T.; Hashimoto, K.; Yamashita, K.;
Liou, S.-Y.; Kagamiishi, Y.; Maegawa, H.; Katsube, N.; Peters, C.; von
Figura, K.; Saftig, P.; Katunuma, N.; Yasutomo, K.; Nakanishi, H.
Involvement of cathepsin E in exogenous antigen processing in primary
cultured murine microglia. J. Biol. Chem. 2002, 277, 4816–4822. (d)
Zhang, T.; Maekawa, Y.; Yamamoto, K.; Ishikawa, H.; Nashed, B. F.;
Dainichi, T.; Hisaeda, H.; Sakai, T.; Kasai, M.; Mizuochi, T.; Asao, T.;
Katunuma, N.; Himeno, K. Pepstatin A-sensitive aspartic proteases in
lysosome are involved in degradation of the invariant chain and antigen-
processing in antigen presenting cells of mice infected with Leishmania
major. Biochem. Biophys. Res. Commun. 2000, 276, 693–701.
27) Janeway, C. A.; Travers, P.; Walport, M.; Shlomchik, M. T cell-
mediated immunity. Immunobiology, The Immune System in Health
and Disease, 6th ed.; Garland Publishing: New York, 2005; pp 319-366.
28) Zaidi, N.; Herrmann, T.; Baechle, D.; Schleicher, S.; Gogel, J.;
Driessen, C.; Voelter, W.; Kalbacher, H. A new approach for
distinguishing cathepsin E and D activity in antigen-processing
organelles. FEBS J. 2007, 274, 3138–3149.
(
(
(
(
(
29) Janeway, C. A.; Travers, P.; Walport, M.; Shlomchik, M. Auto-
immunity and transplantation. Immunobiology, The Immune Sys-
tem in Health and Disease, 6th ed.; Garland Publishing: New York,
(51) Saftig, P.; Hetman, M.; Schmahl, W.; Weber, K.; Heine, L.;
¨
Mossmann, H.; Koster, A.; Hess, B.; Evers, M.; von Figura, K.;
Peters, C. Mice deficient for the lysosomal proteinase cathepsin D
exhibit progressive atrophy of the intestinal mucosa and profound
destruction of lymphoid cells. EMBO J. 1995, 14, 3599–3608.
(52) Brey, W. W.; Edison, A. S.; Nast, R. E.; Rocca, J. R.; Saha, S.;
Withers, R. S. Design, construction, and validation of a 1-mm
triple-resonance high-temperature-superconducting probe for
NMR. J. Magn. Reson. 2006, 179, 290–293.
(53) The default value of exhaustiveness (8) was sufficient to reproduce
the bound conformation of pepstatin A, but since compounds 1
and 3 have more rotatable bonds, a higher value (25) was used for
all docking studies.
2005; pp 557-612.
30) Trott, O.; Olson, A. J. AutoDock Vina: improving the speed and
accuracy of docking with a new scoring function, efficient optimi-
zation, and multithreading. J. Comput. Chem. 2009. DOI: 10.1002/
jcc.21334 (epub ahead of print).
31) Baldwin, E. T.; Bhat, T. N.; Gulnik, S.; Hosur, M. V.; Sowder, R.
C.; Cachau, R. E.; Collins, J.; Silva, A. M.; Erickson, J. W. Crystal
structures of native and inhibited forms of human cathepsin D:
implications for lysosomal targeting and drug design. Proc. Natl.
Acad. Sci. U.S.A. 1993, 90, 6796–6800.
(
(
(
32) Ostermann, N.; Gerhartz, B.; Worpenberg, S.; Trappe, J.; Eder, J.
Crystal structure of an activation intermediate of cathepsin E.
J. Mol. Biol. 2004, 342, 889–899.
(54) (a) Bernstein, N. K.; Cherney, M. M.; Yowell, C. A.; Dame, J. B.;
James, M. N. G. Structural insights into the activation of P. vivax
plasmepsin. J. Mol. Biol. 2003, 329, 505–524. (b) Borelli, C.; Ruge,
E.; Schaller, M.; Monod, M.; Korting, H. C.; Huber, R.; Maskos, K. The
crystal structure of the secreted aspartic proteinase 3 from Candida
albicans and its complex with pepstatin A. Proteins: Struct., Funct.,
Bioinf. 2007, 68, 738–748. (c) Fitzgerald, P. M. D.; McKeever, B. M.;
VanMiddlesworth, J. F.; Springer, J. P.; Heimbach, J. C.; Leu, C.-T.;
Herber, W. K.; Dixon, R. A. F.; Darke, P. L. Crystallographic analysis of
a complex between human immunodeficiency virus type 1 protease and
33) Sielecki, A. R.; Fedorov, A. A.; Boodhoo, A.; Andreeva, N. S.; James,
M. N. G. Molecular and crystal structure of monoclinic porcine
˚
pepsin refined at 1.8 A resolution. J. Mol. Biol. 1990, 214, 143–170.
34) Rao-Naik, C.; Guruprasad, K.; Batley, B.; Rapundalo, S.; Hill, J.;
Blundell, T.; Kay, J.; Dunn, B. M. Exploring the binding prefer-
ences/specificity in the active site of human cathepsin E. Proteins:
Struct., Funct., Bioinf. 1995, 22, 168–181.
˚
(
35) Ghoneim, O. M.; Legere, J. A.; Golbraikh, A.; Tropsha, A.; Booth,
acetyl-pepstatin at 2.0 A resolution. J. Biol. Chem. 1990, 265, 14209–
R. G. Novel ligands for the human histamine H
1
receptor: synth-
14219. (d) Fujimoto, Z.; Fujii, Y.; Kaneko, S.; Kobayashi, H.; Mizuno,
H. Crystal structure of aspartic proteinase from Irpex lacteus in
complex with inhibitor pepstatin. J. Mol. Biol. 2004, 341, 1227–
1235. (e) Fujinaga, M.; Chernaia, M. M.; Tarasova, N. I.; Mosimann,
S. C.; James, M. N. G. Crystal structure of human pepsin and its
complex with pepstatin. Protein Sci. 1995, 4, 960–972. (f) Kamitori, S.;
Ohtaki, A.; Ino, H.; Takeuchi, M. Crystal structures of Aspergillus
oryzae aspartic proteinase and its complex with an inhibitor pepstatin at
esis, pharmacology, and comparative molecular field analysis
studies of 2-dimethylamino-5-(6)-phenyl-1,2,3,4-tetrahydro-
naphthalenes. Bioorg. Med. Chem. 2006, 14, 6640–6658.
36) For example, conformations where the N-C direction was re-
versed, where the ligand folded back on itself, or where a unit other
than statine resided at the catalytic center.
37) Rich, D. H. Pepstatin-derived inhibitors of aspartic proteinases. A
close look at an apparent transition-state analogue inhibitor.
J. Med. Chem. 1985, 28, 263–273.
(
(
(
(
˚
1.9 A resolution. J. Mol. Biol. 2003, 326, 1503–1511. (g) Yang, J.;
Quail, J. W. Structure of the Rhizomucor miehei aspartic proteinase
˚
38) Scarborough, P. E.; Dunn, B. M. Redesign of the substrate
complexed with the inhibitor pepstatin A at 2.7 A resolution. Acta
specificity of human cathepsin D: the dominant role of position
287 in the S
39) Arnold, D.; Keilholz, W.; Schild, H.; Dumrese, T.; Stevanovi ꢀc , S.;
Rammensee, H.-G. Substrate specificity of cathepsins D and E
determined by N-terminal and C-terminal sequencing of peptide
pools. Eur. J. Biochem. 1997, 249, 171–179.
Crystallogr. D 1999, D55, 625–630.
2
subsite. Protein Eng. 1994, 7, 495–502.
(55) (a) Asojo, O. A.; Gulnik, S. V.; Afonina, E.; Yu, B.; Ellman, J. A.;
Haque, T. S.; Silva, A. M. Novel uncomplexed and complexed
structures of plasmepsin II, an aspartic protease from Plasmodium
falciparum. J. Mol. Biol. 2003, 327, 173–181. (b) Bone, R.; Vacca,
J. P.; Anderson, P. S.; Holloway, M. K. X-ray crystal structure of the
(
40) Copeland, R. A. Slow binding inhibitors. Evaluation of Enzyme
Inhibitors in Drug Discovery: a Guide for Medicinal Chemists and
Pharmacologists, Wiley & Sons: Hoboken, NJ, 2005; pp 141-177.
41) Marcinkeviciene, J.; Luo, Y.; Graciani, N. R.; Combs, A. P.;
Copeland, R. A. Mechanism of inhibition of β-site amyloid pre-
cursor protein-cleaving enzyme (BACE) by a statine-based peptide.
J. Biol. Chem. 2001, 276, 23790–23794.
HIV protease complex with L-700,417, an inhibitor with pseudo C
2
symmetry. J. Am. Chem. Soc. 1991, 113, 9382–9384. (c) Coates, L.;
Erskine, P. T.; Wood, S. P.; Myles, D. A. A.; Cooper, J. B. A neutron
Laue diffraction study of endothiapepsin: implications for the aspartic
proteinase mechanism. Biochemistry 2001, 40, 13149–13157. (d)
Fraser, M. E.; Strynadka, N. C. J.; Bartlett, P. A.; Hanson, J. E.; James,
M. N. G. Crystallographic analysis of transition-state mimics bound to
penicillopepsin: phosphorus-containing peptide analogues. Biochem-
istry 1992, 31, 5201–5214. (e) James, M. N. G.; Sielecki, A. R.;
Hayakawa, K.; Gelb, M. H. Crystallographic analysis of transition state
mimics bound to penicillopepsin: difluorostatine- and difluorostatone-
containing peptides. Biochemistry 1992, 31, 3872–3886.
(
(
42) Bull, H. G.; Thorberry, N. A.; Cordes, M. H. J.; Patchett, A. A.;
Cordes, E. H. Inhibition of rabbit lung angiotensin-converting
enzyme by N -[(S)-1-carboxy-3-phenylpropyl]-L-alanyl-L-proline
and N -[(S)-1-carboxy-3-phenylpropyl]-L-lysyl-L-proline. J. Biol.
R
R
Chem. 1985, 260, 2952–2962.
(
43) Bernardo, M. M.; Brown, S.; Li, Z.-H.; Fridman, R.; Mobashery,
S. Design, synthesis, and characterization of potent, slow-binding
inhibitors that are selective for gelatinases. J. Biol. Chem. 2002, 277,
(56) Arnold, K.; Bordoli, L.; Kopp, J.; Schwede, T. The SWISS-
MODEL workspace: a web-based environment for protein struc-
ture homology modelling. Bioinformatics 2006, 22, 195–201.
(57) Sielecki, A. R.; Fujinaga, M.; Read, R. J.; James, M. N. G.
11201–11207.
˚
(
44) Dunn, G. P.; Ikeda, H.; Bruce, A. T.; Koebel, C.; Uppaluri, R.; Bui,
J.; Chan, R.; Diamond, M.; White, J. M.; Sheehan, K. C. F.;
Refined structure of porcine pepsinogen at 1.8 A resolution.
J. Mol. Biol. 1991, 219, 671–692.