Functional imaging of legumain in cancer using a new quenched activity-based probe

Article abstract of DOI:10.1021/ja307083b

Legumain is a lysosomal cysteine protease whose biological function remains poorly defined. Legumain activity is up-regulated in most human cancers and inflammatory diseases most likely as the result of high expression in populations of activated macrophages. Within the tumor microenvironment, legumain activity is thought to promote tumorigenesis. To obtain a greater understanding of the role of legumain activity during cancer progression and inflammation, we developed an activity-based probe that becomes fluorescent only upon binding active legumain. This probe is highly selective for legumain, even in the context of whole cells and tissues, and is also a more effective label of legumain than previously reported probes. Here we present the synthesis and application of our probe to the analysis of legumain activity in primary macrophages and in two mouse models of cancer. We find that legumain activity is highly correlated with macrophage activation and furthermore that it is an ideal marker for primary tumor inflammation and early stage metastatic lesions.

Full text of DOI:10.1021/ja307083b

Article
pubs.acs.org/JACS
Functional Imaging of Legumain in Cancer Using a New Quenched
Activity-Based Probe
Laura E. Edgington,†,‡ Martijn Verdoes,‡ Alberto Ortega,‡ Nimali P. Withana,‡ Jiyoun Lee,‡,⊥
Salahuddin Syed,‡ Michael H. Bachmann,§ Galia Blum,‡,⊥,# and Matthew Bogyo*,†,‡,∥
‡
§
∥
†Cancer Biology Program, Departments of Pathology, Pediatrics, and Microbiology and Immunology, Stanford School of
Medicine, 300 Pasteur Drive, Stanford, California 94305-5324, United States
S
* Supporting Information
ABSTRACT: Legumain is a lysosomal cysteine protease
whose biological function remains poorly defined. Legumain
activity is up-regulated in most human cancers and
inflammatory diseases most likely as the result of high
expression in populations of activated macrophages. Within
the tumor microenvironment, legumain activity is thought to
promote tumorigenesis. To obtain a greater understanding of
the role of legumain activity during cancer progression and inflammation, we developed an activity-based probe that becomes
fluorescent only upon binding active legumain. This probe is highly selective for legumain, even in the context of whole cells and
tissues, and is also a more effective label of legumain than previously reported probes. Here we present the synthesis and
application of our probe to the analysis of legumain activity in primary macrophages and in two mouse models of cancer. We find
that legumain activity is highly correlated with macrophage activation and furthermore that it is an ideal marker for primary
tumor inflammation and early stage metastatic lesions.
Precise roles for stromal-derived legumain have not been
thoroughly defined; however, it likely contributes to the tumor-
promoting inflammation associated with most malignancies.18
One antitumor strategy involved triggering an immune
response toward legumain-expressing cells with a mini-gene
vaccine. This led to ablation of TAMs and subsequent
reduction in tumor volume, angiogenesis, and metastasis.17,19
Other strategies have focused on prodrugs that, when processed
by legumain, release cytotoxic agents such as doxorubicin
specifically in cells with high levels of active legumain.1,2,16,20,21
These pro-drugs have been shown to have a profound impact
on survival in both syngeneic and human xenograft tumor
models in mice. Efficacy of these strategies illuminates
important roles for legumain-expressing cells in tumor
progression, invasion, and metastasis.
As with any protease, analysis of expression of legumain at
the gene or protein level gives little insight into the regulation
of its enzymatic activity. Legumain is synthesized as an inactive
zymogen that becomes proteolytically active through an
autocleavage event upon arriving in the acidic environment of
the lysosome. Once active, legumain is also subject to inhibition
by endogenous proteins such as the cystatins. Recently, our
group reported a near-infrared fluorescent activity-based probe,
LP-1, to detect legumain activity.22 This probe contains an aza-
epoxide electrophile that binds specifically and irreversibly to
the catalytic thiol in the active site of legumain. Because LP-1 is
not quenched, the free, unbound probe must clear before
INTRODUCTION
■
Legumain, or asparaginyl endopeptidase, is a cysteine protease
found primarily in the acidic environment of the lysosome,
although in some cell types it may also be associated with the
cell surface.1,2 As its name indicates, it was first identified in
legumes as a vacuolar processing enzyme with unique substrate
specificity toward asparagine residues in the P1 position.
Legumain is most highly expressed in the kidney, where it
contributes to normal protein catabolism and renal homeo-
stasis.3 It also promotes extracellular matrix remodeling in
proximal tubule cells through degradation of fibronectin.4
Additionally, it, along with several members of the cysteine
cathepsin family, plays important roles in immune function,
where it can initiate invariant chain processing during Class II
MHC antigen presentation.5−9
Beyond its roles in normal physiology, legumain is associated
with a number of inflammatory diseases such as atheroscle-
rosis,4,10,11 stroke,12 and cancer. Its expression is upregulated in
the majority of solid tumors2 and correlates with decreased
survival in human patients bearing breast,13 colorectal,14 and
ovarian15 tumors. A number of in vitro assays have
demonstrated that HEK 293 cells overexpressing legumain
show increased potential for cell migration, invasion, and
angiogenesis.2 These cells also show increased metastasis from
subcutaneous tumors in vivo compared to nontransfected
control cells.2 Within the tumor microenvironment, however,
legumain is most highly expressed by tumor-associated
macrophages (TAMs) and not by the tumor cells them-
selves.16,17
Received: July 24, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 1. Schematic of a quenched probe binding to a cysteine protease. Thiol attack of the acyloxymethyl ketone results in loss of the quenching
group, leading to an increase in fluorescence emission.
Figure 2. (A) Fluorescent SDS-PAGE showing dose-dependent labeling of legumain by LE28 in RAW cell macrophage extracts. (B)
Immunoprecipitation of LE28-labeled RAW cell lysates using a legumain-specific polyclonal antibody. (C) RAW cells were labeled at the indicated
dose of LE28 for one hour followed by lysis and fluorescent SDS-PAGE. (D) RAW cells were incubated with 1 μM LE28 for the indicated time
period followed by lysis and fluorescent SDS-PAGE. (E) Microscopy images of RAW cells exposed to LE28 (red) for 0 or 5 h. Lysotracker (green)
was used to reveal the location of lysosomes. (F) Microscopy of RAW cells exposed to DMSO vehicle or 100 μM of the legumain inhibitor, LI-1
followed by incubation with LE28. (G) SDS-PAGE analysis of RAW cells exposed to DMSO vehicle or LI-1, followed by labeling with LE28.
specific signals can be detected in vivo. To circumvent this
problem, and thereby increase the specificity of the fluorescent
signal, we describe here a “smart” legumain probe that is
intrinsically quenched and becomes fluorescent only upon
specific, covalent binding to legumain. This new probe shows
comparable potency toward legumain in cell lysates and
enhanced labeling of legumain in intact cells, organs, and
tumors when compared to LP-1. We show that this reagent can
be used to image legumain activity in both normal mice and in
two murine models of cancer. We also use the probe to
B
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
monitor the induction of legumain activity in alternatively
activated bone-marrow-derived macrophages. These data
provide insight into how this protease could be up-regulated
in the inflammatory tumor microenvironment and provide a
valuable new tool for imaging of inflammation events.
of legumain at all concentrations tested (Figure 2c). We also
performed a labeling time course in intact cells. Legumain
activity could be detected as little as 30 min after probe
addition and was saturated after two hours (Figure 2d). We also
performed identical experiments in COLO205 colorectal
cancer cells. While the potency of LE28 was similar in both
cell lines (Figure S4b, Supporting Information), we found that
saturation of legumain labeling in COLO205 cells did not occur
until four hours after probe addition (Figure S4c, Supporting
Information). The more rapid uptake by macrophages supports
our hypothesis that the probe enters cells by endocytosis or
macropinocytosis.
Next, we examined the LE28-labeled RAW cells by
microscopy. We observed a punctate staining pattern consistent
with lysosomal accumulation. As anticipated, LE28 staining
colocalized with Lysotracker, confirming lysosomal distribution
(Figure 2e). To show that the fluorescent signal was specific to
legumain, we also treated RAW cells with a legumain-specific
inhibitor, LI-1,22 prior to the addition of LE28. Compared to a
control that was pretreated with a DMSO vehicle, the LI-1
treated cells showed dramatically less signal (Figure 2f). To
verify that this was due to legumain inhibition, we also analyzed
cells treated in parallel by fluorescent SDS-PAGE. Indeed,
nearly all of the LE28 labeling was blocked upon pretreatment
with LI-1 (Figure 2g).
LE28 Reactivity toward Caspases. In RAW cells, the
quenched probe LE28 exhibited specific labeling of legumain.
Since its structure was designed based on our nonquenched
caspase probe, AB50, we wondered whether it retained
reactivity toward caspases. We used an in vitro model of
apoptosis in which COLO205 cells were stimulated with anti-
DR5, a monoclonal antibody that initiates the extrinsic
apoptosis pathway.28−30 As we observed previously, AB50
labeled caspases and legumain at similar levels in intact cells
(Figure S5a, Supporting Information). LE28, however, showed
virtually no detectable caspase labeling and dramatically
increased legumain labeling. When labeling was carried out in
apoptotic lysates at neutral pH, both probes labeled caspases;
however, LE28 was far less potent (Figure S5b, Supporting
Information). In acidic pH, both probes labeled multiple
legumain bands (Figure S5c, Supporting Information). These
bands were confirmed to be legumain by immunoprecipation
with a legumain-specific antibody (Figure S5d, Supporting
Information). Taken together, these data indicate that the bulky
quenching group on LE28 effectively reduces the probe’s
affinity for caspases and sequesters it within the endocytic
pathway, where it binds exclusively to legumain.
In Vitro Comparison of LE28 and LP-1. Next we wanted
to compare the efficacy of LE28 with that of the aza-asparagine
epoxide probe, LP-1, which is the current gold standard for
detecting legumain activity in vitro and in vivo.22 Labeling was
comparable with both probes in RAW cell extracts (pH 4.5)
where LP-1 showed slightly increased potency compared to the
LE28 probe (Figure S6a, Supporting Information). In intact
cells, on the other hand, LE28 labeled more legumain at 1 μM
than LP-1 (Figure S6b, Supporting Information), perhaps due
to increased uptake of LE28 into the lysosomes compared to
the less bulky LP-1.
RESULTS AND DISCUSSION
■
Development of a Quenched Activity-Based Probe
for Legumain. To generate a quenched legumain probe, we
needed to utilize an electrophile that would produce a leaving
group upon covalent binding with the active site cysteine. The
aza-epoxide warhead in LP-1 is not suitable for such a purpose
as nucleophilic attack results in opening of the epoxide ring. We
turned instead to the acyloxymethyl ketone (AOMK) electro-
phile that we have successfully used to make quenched probes
for cathepsins.23,24 Upon binding of the enzyme to the AOMK
reactive group, the O-acyl group is released, making it an ideal
place to introduce a quenching moiety (Figure 1).
Legumain exhibits a strong preference for cleavage after
asparagine residues; however, it has also been shown to bind
AOMK probes containing a P1 aspartic acid.25−27 At low pH in
the lysosome, the acid side chain is protonated, thereby
permitting binding in the S1 pocket of legumain. Because P1
Asn-AOMK probes are not stable over time,27 we opted to
move forward with AOMKs that contain a P1 Asp. Our
previously reported probe, AB50, containing the sequence Glu-
Pro-Asp-AOMK, primarily targeted caspases. However, it also
demonstrated significant cross-reactivity with legumain.28 We
hypothesized that adding a bulky quenching group to AB50
would encourage uptake into the cell by macropinocytosis/
endocytosis rather than direct entry into the cytoplasm.
Sequestration of the probe in the endosomal pathway would
increase access to legumain and limit interaction with caspases,
thereby producing a legumain-selective quenched probe.
We first compared fluorescence emission of increasing
concentrations of both AB50 and LE28 and found LE28 to
be highly quenched (Figure S2, Supporting Information). To
test the capacity for LE28 to bind legumain, we first used cell
extracts prepared from mouse macrophages (RAW 264.7 cells)
lysed in citrate buffer (pH 4.5). (To verify equal loading in this
and subsequent experiments, we have included data from our
BCA protein quantification assays. Protein concentration was
measured for each experiment, and uniform amounts of protein
were used for each sample within the same gel (Figure S3,
Supporting Information).) We incubated these extracts with
increasing concentrations of LE28, and then proteins were
resolved by SDS-PAGE followed by scanning for Cy5
fluorescence. We observed labeling of a single legumain species
at 36 kDa (Figure 2a). This band was detectable at probe
concentrations as low as 100 pM with an EC50 ∼10 nM and
saturation above 50 nM. To confirm that the labeled species
was indeed legumain, we performed an immunoprecipitation
assay on probe-labeled lysates with a legumain-specific
polyclonal antibody (Figure 2b). To verify that LE28 was
activity-dependent, we preincubated RAW cell lysates with
recombinant cystatin C, a naturally expressed legumain
inhibitor, followed by addition of LE28. We observed dose-
dependent competition of legumain labeling by LE28 (Figure
S4a, Supporting Information). Cystatin C N39A, a mutant
lacking the residue critical for legumain binding, however, was
unable to block labeling.
LE28 contains a P1 aspartic acid, which requires protonation
to initiate legumain binding, while LP-1 has a P1 asparagine. To
examine the pH dependence of both probes, we lysed RAW
cells in citrate buffer at varying pH, followed by labeling with
either LE28 or LP-1 (Figure S6c, Supporting Information). We
Next, we examined the ability of LE28 to label legumain in
intact RAW cells. As in lysates, we observed exclusive labeling
C
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
found that LP-1 labeled legumain equally well up to pH 6 and
then dropped off sharply at neutral pH, consistent with the
reported pH dependence of legumain.31 As expected, LE28
labeling decreased markedly above pH 5 due to the need for
protonation of the aspartate side chain, and no labeling was
observed at neutral pH.
Regulation of Legumain Activity. Little is known about
how legumain expression and activity is regulated in the tumor
microenvironment. One study reported that legumain
expression is upregulated in RAW cells exposed to the
cytokines IL-4, IL-10, or IL-13, which differentiate macrophages
to the M2 phenotype.17 We were unable to recapitulate these
results in our laboratory and found that RAW cells basally
express high legumain levels. To further investigate legumain
activity in macrophages, we turned to freshly differentiated
macrophages isolated from murine bone marrow. Primary
macrophages were stimulated for 96 h followed by labeling with
LE28.
activity in response to cytokine treatment. In lysates, however,
the 24 kDa protein was not observed, suggesting that it may be
perturbed by the lysis conditions.
We also examined the kinetics of legumain up-regulation in
response to IL-4/-10 stimulation. A substantial increase was
detected as early as 24 h after exposure, and activity/expression
continued to rise throughout the time course (Figure 3c,d).
Appearance of the 24 kDa legumain band occurred only after
72 h, suggesting regulated processing as macrophages become
more active. When lysates were labeled, we observed the same
time-dependent increase in legumain activity as found in intact
cells (Figure S7c, Supporting Information).
Next we compared the legumain activity of the 4T1 breast
tumor cell with that of primary macrophages (Figure 3e).
When we cocultured an equal number of tumor cells and
macrophages, we saw a sharp increase in legumain activity. This
suggests that factors such as IL-4 and IL-10 (and most certainly
others) released by tumor cells are able to significantly induce
the expression and activation of legumain in macrophages,
though additional experiments are needed to confirm this.
Lysates prepared from xenografted tumors also showed
elevated levels of legumain activity compared to the parental
4T1 line, suggesting macrophage infiltration and legumain
regulation within the tumor microenvironment. We also
performed similar experiments with two other cancer cell
lines. Legumain activation in COLO205 colon cancer cells
followed the same trend as 4T1 cells (Figure S7d, Supporting
Information). However, HCT116 colon cancer cells showed
very little up-regulation both in coculture and in vivo (Figure
S6e, Supporting Information).
IL-4 treatment resulted in an increase of both expression and
activity of legumain compared to nonstimulated controls
(Figure 3a,b). Conversely, IL-10 treatment did not affect
In Vivo Imaging with LE28. While LE28 and LP-1 were
both effective activity-based probes for legumain in vitro, we
anticipated that the real advantage of LE28 would be for optical
imaging. Because LE28 is quenched, it should have dramatically
improved in vivo properties since there is no need for clearance
of the unbound probe to obtain image contrast. We therefore
examined probe clearance by optical imaging in healthy nude
mice. LE28 was injected by the tail vein following imaging of
Cy5 fluorescence over time using the IVIS 100 system (Figure
4a). LE28 produced low fluorescence at early time points. A
specific signal in the kidney became detectable after about one
hour and continued to increase over time. Injection of LP-1, on
the other hand, resulted in whole body fluorescence at early
time points, and after one hour, most of the fluorescent signal
had disappeared (Figure S8a, Supporting Information).
After eight hours, we collected the major organs from mice
injected with each probe and performed a biochemical analysis.
For both probes, the highest labeling was observed in the
kidney and to a lesser extent the liver (Figure 4b and Figure
S8b, Supporting Information). In fact, LE28 labeling in the
kidneys was so high that it was possible to visualize the probe
signal using the noninvasive fluorescence imaging methods,
which are limited by poor penetration depth. Low levels of
signal were also detected in the spleen, intestine, pancreas, and
heart, but not in the lungs or brain. The distribution of labeling
in the organs was largely consistent with the total legumain
levels as detected by Western blot (Figure 4c). However, the
Western blot shows higher levels of legumain in the spleen than
detected by LE28 or LP-1. This may reflect inhibition of
legumain in the spleen or poor probe delivery to this organ.
While the labeling patterns for both probes were very similar,
the intensity of the LE28 signal was more than five times
brighter than LP-1. We attribute this difference to the increased
Figure 3. Regulation of legumain activity. (A) Legumain activity in
primary macrophages stimulated with the indicated cytokine, followed
by labeling of intact cells by LE28 and fluorescent SDS-PAGE. (B)
Antilegumain Western blot of samples in (A). (C) Time course of
legumain activation in primary macrophages exposed to IL-4/-10.
Intact cells were labeled with LE28 followed by SDS-PAGE. (D)
Antilegumain Western blot of samples in (C). (E) Legumain activity in
macrophages alone, in 4T1 tumor cells alone, or in coculture. The
rightmost column shows activity in tumor cells that were implanted
subcutaneously in mice for one week. For each group, lysates were
prepared and labeled with LE28.
legumain levels. This was surprising given that in dendritic cells
IL-10 has been shown to down-regulate expression of cystatin
C, an endogenous legumain inhibitor.32 When we treated
macrophages with both IL-4 and IL-10 simultaneously, we
observed considerable up-regulation of three species, which
were confirmed to be processed forms of legumain by immune
precipitation (Figure S7a, Supporting Information). We
considered the possibility that the increase in legumain activity
that we observed may have been due to increased uptake of
LE28 by activated macrophages that have an increased potential
for endocytosis of extracellular material. Therefore, we also
performed a parallel experiment in which macrophages were
lysed post cytokine treatment and then labeled with LE28
(Figure S7b, Supporting Information). The same activation
trend was observed, confirming an up-regulation of legumain
D
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. In vivo imaging with LE28. (A) Clearance of the LE28 signal in normal mice imaged on an IVIS 100 machine. (B) Fluorescent SDS-PAGE
analysis of tissues removed from mice after the indicated time and corresponding antilegumain Western blot. KD = kidney, LV = liver, SP = spleen,
IN = intestine, PN = pancreas, HT = heart, LG = lung, and BR = brain. (C) Western blot analysis of total legumain expression in samples from (B)
using a legumain-specific polyclonal antibody. (D) Mice bearing HCT-116 xenograft tumors on their backs were injected with LE28 and imaged over
time using an IVIS 100. (E) Tumors from three mice were removed and imaged ex vivo, followed by fluorescent SDS-PAGE analysis.
half-life of LE28 in combination with the enhanced cellular
uptake that we observed in vitro. Furthermore, analysis of
tissues after 28 h probe circulation revealed nearly identical
levels of probe labeling, indicating that LE28-bound legumain is
sustained over time (Figure S8c, Supporting Information). The
most mature form of legumain was previously reported to be 36
kDa. In the kidney and liver, labeled proteins with molecular
weights lower than 36 were confirmed to be processed forms of
legumain, as shown by immunoprecipitation with a legumain-
specific antibody (Figure S8d,e, Supporting Information).
We also wanted to assess the ability of LE28 to detect
legumain in tumors by noninvasive imaging. To this end, we
used a simple xenograft model using HCT-116 human
colorectal carcinoma cells. We injected tumor-bearing mice
with LE28 and monitored whole body fluorescence over the
course of 28 h. The LE28 signal could be detected around the
periphery of the tumors in as early as 30 min with gradual
penetration throughout the tumor over time (Figure 4d).
Maximal contrast between tumor and normal tissue was
achieved after seven hours, and the signal remained constant
up to 28 h. Importantly, ex vivo fluorescence correlated with
levels of legumain labeling by SDS-PAGE (Figure 4e). In the
same experiment, we also performed a direct comparison of
LE28 to the nonquenched probe, LP-1 (Figure S9, Supporting
Information). LE28 showed much brighter tumor signal, which
persisted for longer than that of LP-1. LE28 also produced
better contrast of the tumor over the surrounding normal tissue
(Figure S9a−c, Supporting Information). Futhermore, the ex
vivo fluorescence produced by LE28 was strikingly brighter
than LP-1, corresponding to dramatically increased labeling of
legumain as assessed by SDS-PAGE (Figure S9d, Supporting
Information).
Imaging Legumain Activity in a Syngeneic Tumor
Model. Legumain is highly expressed in macrophages and
plays roles in immune function. Infiltration of immune cells
into the tumor has been shown to be a hallmark of
tumorigenesis and contributes to the smoldering inflammation
characteristic of the tumor microenvironment. The nude mice
used in the previous study lack a complete immune system;
therefore, we wanted to test LE28 in a more physiologically
relevant model. We turned to a syngeneic model using
immunocompetent Balb/c mice. 4T1 cells derived from a
spontaneous mouse mammary tumor were injected by tail vein
to simulate metastasis. Since the cells were engineered to
express luciferase, tumor progression could be monitored by
bioluminescence imaging. Once tumors were established, we
injected LE28. Six hours later, we injected luciferin and then
performed bioluminescence and fluorescence imaging and
biochemical analysis of excised tissues (Figure 5a). The
majority of mice developed nodules in the lung. We found
that even the smallest nodules that we could detect by
luciferase were positive for Cy5 fluorescence and thus legumain
̈
activity (Figure 5b). No signal was detected in naıve lungs. As
the tumor burden increased, we saw increased levels of
fluorescence, which largely colocalized with luciferase activity
but also spread to the surrounding normal tissue. We believe
this is due to the fact that LE28 provides a broad measure of
inflammation of the lung, while the luciferase signal remains
confined to tumor tissues. Solid tumors also appeared
throughout the peritoneum, and these tissues were also positive
for the LE28 probe signal.
When we analyzed whole lung lysates by SDS-PAGE, we
found that legumain levels were increased in tissues with high
tumor burden compared to normal lungs and those with few
tumors. We were surprised to see labeling of additional protein
E
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 5. Imaging legumain activity in a syngeneic tumor model. (A) Schematic of experimental setup. (B) Lungs bearing increasing burden of 4T1
experimental metastases were imaged for luciferase activity (bioluminescence) and legumain activity (LE28 fluorescence). (C) Whole lung lysates
from (B) were analyzed by fluorescent SDS-PAGE and Western blot using a legumain-specific antobody. (D) Individual lung tumors from (B) were
dissected as well as solid tumors arising in other regions of the peritoneum and analyzed by fluorescent SDS-PAGE and Western blot using a
legumain-specific antobody.
bands in the lungs with high tumor burden (Figure 5c). These
labeled proteins were consistent with the sizes of the lysosomal
cysteine cathepsins. Their identity was confirmed by
immunoprecipitation with cathepsin-specific antibodies (not
shown). When we dissected out individual lung tumors,
however, labeled cathepsins were not observed (Figure 5d).
This indicates that cathepsin labeling may result from high
levels of these proteases due to inflammation in the
surrounding tissues. Solid tumors isolated from other parts of
the body produced considerably higher levels of legumain
activity than the lung tumors, indicating that the metastatic
niche may be important for legumain regulation.
brighter legumain signal in vivo compared to nonquenched
probes such as LP-1.
We used LE28 to image legumain activity in normal tissues
̈
in naıve mice and in inflammation associated with xenografted
tumors and metastases in a syngeneic model. In addition to
kidney and liver, where legumain is constitutively active, we
observed an accumulation of fluorescent signal in tumors.
Contrary to some reports that legumain is only expressed in
macrophages within the tumor microenvironment,16 in vitro,
we clearly detect legumain in primary macrophages, and its
activity increases upon exposure to inflammatory cytokines or
tumor cells. We also detect varying levels of activity in several
tumor cell lines in vitro.
Studies using legumain-activated prodrugs highlight the
importance of legumain-expressing cells during tumor pro-
gression.1,2,16,20,21 However, precise roles for legumain activity
within both tumor and stroma have yet to be carefully
delineated. It is also not clear whether blocking the activity of
legumain could have a therapeutic effect in a number of
conditions involving inflammation. LE28 will likely have great
value in future efforts to assess the potential of this protease as a
drug target in cancer and diseases involving increased
inflammation.
CONCLUSIONS
■
In this study, we present a newly designed activity-based probe
for the lysosomal protease legumain. We found that the main
advantage of LE28 over previously reported legumain probes is
that it is fluorescently quenched and therefore acts as a smart
probe that can be used for real-time imaging of legumain
activity. Because no fluorescent signal is produced until the
probe binds active legumain, the background signal is
dramatically reduced, and the need for clearance of the free
probe is eliminated. Furthermore, enhanced signal-to-noise
ratios makes LE28 ideal for microscopy experiments and in vivo
imaging. An added benefit of the large, bulky quenching group
is that it appears to enhance uptake of the probe into the
lysosomal compartments of cells. As a result, LE28 provides
MATERIALS AND METHODS
■
Probe Synthesis. The synthesis of LE28 is outlined in Scheme 1
and described in detail in the Supporting Information (Figure S1).
Lysate Labeling. Cells were seeded in six-well plates (typically 1.5
× 106/well) 24 h prior to harvest. Cells were then washed once with
F
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 1. Synthesis of LE28
PBS, scraped, and pelleted. Cells were then lysed in acidic buffer (50
mM sodium citrate [pH 4.5], 0.1% CHAPS, 1% NP-40, and 4 mM
DTT), snap-frozen in liquid N2, and centrifuged at 4 °C for ten
minutes. Supernatants were collected, and protein concentration was
determined using a BCA kit. An amount of 40 μg of total protein was
then aliquotted, followed by addition of the indicated probe from a
100× DMSO stock solution, yielding a final DMSO concentration of
1%. Samples were incubated for 30 min at 37 °C and then solubilized
with 4× sample buffer. Proteins were resolved by 15% SDS-PAGE and
scanned using a Typhoon flatbed laser scanner (excitation 633 nm/
emission 670 nm). For the pH dependence assay, citrate buffer was
adjusted to the indicated pH just prior to lysis.
Intact Cell Labeling. Cells were seeded as above 24 h prior to
labeling. Media were replaced, and a probe was added from a 1000×
DMSO stock solution, yielding a final DMSO concentration of 0.1%.
Cells were incubated for 1 h at 37 °C unless otherwise noted, washed
once with PBS, scraped, and pelleted. Cells were then lysed in
hypotonic lysis buffer (50 mM PIPES [pH 7.4], 10 mM KCl, 5 mM
MgCl2, 2 mM EDTA, 4 mM DTT, and 1% NP-40), snap-frozen, and
centrifuged at 4 °C. Supernatants were then solubilized with 4×
sample buffer, and 40 μg of total protein was analyzed by SDS-PAGE
as above.
Assessing Caspase Cross-Reactivity. COLO205 human colon
cancer cells were plated in six-well dishes 24 h prior to treatment.
Medium was refreshed, and anti-DR5 (20 μg/mL) was added for 4 h
followed by hypotonic lysis as above. For intact labeling experiments,
probes were added at the time of anti-DR5 treatment. For lysate
labeling, probes were added to 50 μg of apoptotic lysate for one hour.
Samples were analyzed by SDS-PAGE as above.
Microscopy. RAW cells were seeded at a density of 100 000 cells
in an eight-well coverslip chamber (Lab-Tek). Twenty-four hours later,
cells were treated with either a DMSO or 1 μM probe for five hours.
For the last 30 min, Lysotracker-green (100 nM) was added to the
cells. Cells were then washed three times with PBS, and phenol red-
free complete medium was added. Cells were imaged live at 40× using
a Zeiss Axiovert 200 M confocal microscope in both Cy5 and FITC
channels. For the competition experiment, cells were pretreated with
LI-1 or DMSO for one hour, followed by addition of 1 μM LE28 for 5
h, followed by epifluorescent imaging with the same microscope.
Bone-Marrow-Derived Macrophages. Tibias and femurs were
flushed with DMEM, and red blood cells were eliminated with
PharmLyse (BD). The remaining cells were plated in DMEM
containing 10 ng/mL of M-CSF for five days to promote macrophage
differentiation. On the fifth day, cells were plated in six-well dishes at a
density of 1 × 106 cells/well. IL-4, IL-10, or both (10 ng/mL) were
added to the cells for the indicated number of days. Alternatively,
macrophages were cocultured at a one-to-one ratio for four days. At
Tissue Labeling. Tissues were sonicated in muscle lysis buffer (1%
Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 4 mM DTT, PBS
[pH 7.4]) and analyzed as above.
G
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the end of the experiment, cells were either labeled with 1 μM LE28 or
lysed in citrate buffer, pH 4.5, and then labeled, followed by SDS-
PAGE analysis as above.
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Western bBotting. Western blots were performed according to
standard procedure. Sheep antimouse legumain affinity purified
polyclonal antibody was purchased from R&D Systems (AF2058)
and used at 1:1000 in 5% milk/PBS-T overnight at 4 °C. Donkey
antisheep-HRP secondary was used at 1:3000 for one hour at RT.
Immunoprecipitation. An amount of 100 μg of probe-labeled
protein was diluted into 500 μL of RIPA buffer (PBS [pH 7.4], 1 mM
EDTA, 0.5% NP-40) along with 5 μL of sheep antimouse legumain
antibody. Samples were incubated on ice for 10 min followed by
addition of 40 μL of slurry of prewashed Protein A/G agarose beads
(Santa Cruz). Samples were agitated overnight at 4 °C. The
supernatant was removed and precipitated with acetone followed by
freezing for two hours, centrifugation at high speed, and resuspension
in 1× sample buffer. The beads were washed four times with RIPA
buffer and once with 0.9% NaCl and then boiled in 2× sample buffer.
Input, pulldown, and supernatant samples were analyzed by SDS-
PAGE as above.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by a grant from the National
Insittutes of Health (R01 EB005011 to MB). The authors
would like to thank M. Abrahamson for the kind gift of
recombinant Cystatin C mutants. We also thank T. Doyle at
the Stanford Small Animal Facility for assistance with the
optical imaging studies. Lastly, we thank S. Lynch at the
Stanford NMR Facility and A. Chien and T. McLaughlin at the
Stanford Mass Spec Facility for their assistance with detailed
compound characterization.
Assessing Legumain Activity in Healthy Mice. All animal
experiments were performed according to specific guidelines approved
by the Stanford Administrative Panel on Laboratory Animal Care.
Eight-week-old nude mice (Charles Rivers) were injected with LE28
(20 nmol in 20% DMSO/PBS, ∼2 mg/kg) by tail vein. Mice were
anesthetized with isofluorane and then imaged at the indicated time
points using an IVIS 100 system. Tissues were removed and analyzed
by SDS-PAGE as above.
Xenograft Tumor Model. Six-week-old, female nude mice were
purchased from Charles Rivers. HCT-116 human colorectal carcinoma
cells (3 × 106 in 30 μL of 0.5% BSA/PBS, ATTC) were injected
subcutaneously on their backs and permitted to grow for eight days.
The indicated probe was injected at a dose of 20 nmol in 20% DMSO/
PBS and imaged using the IVIS 100 system at the indicated time
points. Tumors were then removed, imaged ex vivo using an FMT
system, and analyzed by SDS-PAGE as described above.
Syngeneic Experimental Lung Metastasis Model. Six-week-
old Balb/c mice were purchased from Jackson Laboratories. One
hundred thousand 4T1 mouse mammary cancer cells expressing both
GFP and luciferase were injected by tail vein. Tumor growth was
monitored using the IVIS 100 system by bioluminescence imaging 10
min after intraperitoneal injection of D-luciferin (150 mg/kg in PBS).
Once tumors were established, mice were injected with probes at a
dose of 20 nmol in 20% DMSO/PBS. Six hours later, mice were
injected with D-luciferin as above, and tissues were imaged for both
bioluminescence and fluorescence using the IVIS 100 system. Lysates
were prepared from whole lungs for SDS-PAGE analysis as described
above.
REFERENCES
■
(1) Liu, C.; Sun, C.; Huang, H.; Janda, K.; Edgington, T. Cancer Res.
2003, 63, 2957.
(2) Liu, Y.; Bajjuri, K. M.; Liu, C.; Sinha, S. C. Mol. Pharm. 2012, 9,
168.
(3) Miller, G.; Matthews, S. P.; Reinheckel, T.; Fleming, S.; Watts, C.
FASEB J 2011, 25, 1606.
(4) Morita, Y.; Araki, H.; Sugimoto, T.; Takeuchi, K.; Yamane, T.;
Maeda, T.; Yamamoto, Y.; Nishi, K.; Asano, M.; Shirahama-Noda, K.;
Nishimura, M.; Uzu, T.; Hara-Nishimura, I.; Koya, D.; Kashiwagi, A.;
Ohkubo, I. FEBS Lett. 2007, 581, 1417.
(5) Maehr, R.; Hang, H. C.; Mintern, J. D.; Kim, Y. M.; Cuvillier, A.;
Nishimura, M.; Yamada, K.; Shirahama-Noda, K.; Hara-Nishimura, I.;
Ploegh, H. L. J. Immunol. 2005, 174, 7066.
(6) Shirahama-Noda, K.; Yamamoto, A.; Sugihara, K.; Hashimoto,
N.; Asano, M.; Nishimura, M.; Hara-Nishimura, I. J. Biol. Chem. 2003,
278, 33194.
(7) Matthews, S. P.; Werber, I.; Deussing, J.; Peters, C.; Reinheckel,
T.; Watts, C. J. Immunol. 2010, 184, 2423.
(8) Watts, C. Nat. Immunol. 2004, 5, 685.
(9) Manoury, B.; Mazzeo, D.; Li, D. N.; Billson, J.; Loak, K.;
Benaroch, P.; Watts, C. Immunity 2003, 18, 489.
(10) Mattock, K. L.; Gough, P. J.; Humphries, J.; Burnand, K.; Patel,
L.; Suckling, K. E.; Cuello, F.; Watts, C.; Gautel, M.; Avkiran, M.;
Smith, A. Atherosclerosis 2010, 208, 83.
(11) Clerin, V.; Shih, H. H.; Deng, N.; Hebert, G.; Resmini, C.;
Shields, K. M.; Feldman, J. L.; Winkler, A.; Albert, L.; Maganti, V.;
Wong, A.; Paulsen, J. E.; Keith, J. C., Jr.; Vlasuk, G. P.; Pittman, D. D.
Atherosclerosis 2008, 201, 53.
(12) Ishizaki, T.; Erickson, A.; Kuric, E.; Shamloo, M.; Hara-
Nishimura, I.; Inacio, A. R.; Wieloch, T.; Ruscher, K. J. Cereb. Blood
Flow Metab. 2010, 30, 1756.
(13) Gawenda, J.; Traub, F.; Luck, H. J.; Kreipe, H.; von Wasielewski,
R. Breast Cancer Res. Treat. 2007, 102, 1.
ASSOCIATED CONTENT
* Supporting Information
■
S
Figure S1: LE28 Mass Spec and NMR. Figure S2: Quenching
efficiency. Figure S3: In vitro LE28 labeling, related to Figure 2.
Figure S4: Caspase reactivity of LE28. Figure S5: In vitro
comparison of LE28 and LP-1. Figure S6: Regulation of
legumain activity, related to Figure 3. Figure S7: Imaging with
LP-1, related to Figure 4. This material is available free of
charge via the Internet at http://pubs.acs.org.
(14) Murthy, R. V.; Arbman, G.; Gao, J.; Roodman, G. D.; Sun, X. F.
Clin. Cancer Res. 2005, 11, 2293.
(15) Wang, L.; Chen, S.; Zhang, M.; Li, N.; Chen, Y.; Su, W.; Liu, Y.;
Lu, D.; Li, S.; Yang, Y.; Li, Z.; Stupack, D.; Qu, P.; Hu, H.; Xiang, R. J.
Cell. Biochem. 2012, 113, 2679.
(16) Wu, W.; Luo, Y.; Sun, C.; Liu, Y.; Kuo, P.; Varga, J.; Xiang, R.;
Reisfeld, R.; Janda, K. D.; Edgington, T. S.; Liu, C. Cancer Res. 2006,
66, 970.
(17) Luo, Y.; Zhou, H.; Krueger, J.; Kaplan, C.; Lee, S. H.; Dolman,
C.; Markowitz, D.; Wu, W.; Liu, C.; Reisfeld, R. A.; Xiang, R. J. Clin.
Invest. 2006, 116, 2132.
(18) Hanahan, D.; Weinberg, R. A. Cell 2011, 144, 646.
(19) Lewen, S.; Zhou, H.; Hu, H. D.; Cheng, T.; Markowitz, D.;
Reisfeld, R. A.; Xiang, R.; Luo, Y. Cancer Immunol. Immunother. 2008,
57, 507.
AUTHOR INFORMATION
Corresponding Author
mbogyo@stanford.edu
■
Present Addresses
⊥Department of Global Medical Science, Sungshin Women’s
University, Seoul 142−732, Korea.
#Institute of Drug Research, The Hebrew University, Jerusalem,
Israel 91120.
H
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(20) Stern, L.; Perry, R.; Ofek, P.; Many, A.; Shabat, D.; Satchi-
Fainaro, R. Bioconjugate Chem. 2009, 20, 500.
(21) Bajjuri, K. M.; Liu, Y.; Liu, C.; Sinha, S. C. ChemMedChem 2011,
6, 54.
(22) Lee, J.; Bogyo, M. ACS Chem. Biol. 2010, 5, 233.
(23) Blum, G.; von Degenfeld, G.; Merchant, M. J.; Blau, H. M.;
Bogyo, M. Nat. Chem. Biol. 2007, 3, 668.
(24) Blum, G.; Mullins, S. R.; Keren, K.; Fonovic, M.; Jedeszko, C.;
Rice, M. J.; Sloane, B. F.; Bogyo, M. Nat. Chem. Biol. 2005, 1, 203.
(25) Kato, D.; Boatright, K. M.; Berger, A. B.; Nazif, T.; Blum, G.;
Ryan, C.; Chehade, K. A.; Salvesen, G. S.; Bogyo, M. Nat. Chem. Biol.
2005, 1, 33.
(26) Sexton, K. B.; Witte, M. D.; Blum, G.; Bogyo, M. Bioorg. Med.
Chem. Lett. 2007, 17, 649.
(27) Loak, K.; Li, D. N.; Manoury, B.; Billson, J.; Morton, F.; Hewitt,
E.; Watts, C. Biol. Chem. 2003, 384, 1239.
(28) Edgington, L. E.; Berger, A. B.; Blum, G.; Albrow, V. E.; Paulick,
M. G.; Lineberry, N.; Bogyo, M. Nat. Med. 2009, 15, 967.
(29) Adams, C.; Totpal, K.; Lawrence, D.; Marsters, S.; Pitti, R.; Yee,
S.; Ross, S.; Deforge, L.; Koeppen, H.; Sagolla, M.; Compaan, D.;
Lowman, H.; Hymowitz, S.; Ashkenazi, A. Cell Death Differ. 2008, 15,
751.
(30) Edgington, L. E.; van Raam, B. J.; Verdoes, M.; Wierschem, C.;
Salvesen, G. S.; Bogyo, M. Chem. Biol. 2012, 19, 340.
(31) Li, D. N.; Matthews, S. P.; Antoniou, A. N.; Mazzeo, D.; Watts,
C. J. Biol. Chem. 2003, 278, 38980.
(32) Xu, Y.; Schnorrer, P.; Proietto, A.; Kowalski, G.; Febbraio, M.
A.; Acha-Orbea, H.; Dickins, R. A.; Villadangos, J. A. J. Immunol. 2011,
186, 3666.
I
dx.doi.org/10.1021/ja307083b | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX