| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Max Planck Institut für physiologische und klinische Forschung, Abt. Molekulare Zellbiologie, 61231 Bad Nauheim, Germany; and
* Johannes Gutenberg-Universität Mainz, Anatomisches Institut, Makroskopischer Bereich, D-55099 Mainz, Germany
1Correspondence: Max Planck Institut für physiologische und klinische Forschung, Abt. Molekulare Zellbiologie, Parkstr. 1, D-61231 Bad Nauheim, Germany. E-mail: hannes.drexler{at}kerckhoff.mpg.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-t-butyl-L-glutamyl-L-alanyl-L-leucinal
(PSI) -induced apoptosis at low concentrations (EC50=24
nM), whereas at least 340-fold higher concentrations of PSI were
necessary to obtain the same effect in confluent, contact-inhibited
cells. PSI-mediated cell death could be blocked by a caspase-3
inhibitor (Ac-DEVD-H), but not by a caspase-1 inhibitor (Ac-YVAD-H),
suggesting that a caspase-3-like enzyme is activated during PSI-induced
apoptosis. When applied to the embryonic chick chorioallantoic
membrane, a rapidly expanding tissue, PSI induced massive apoptosis
also in vivo. PSI treatment of the CAM led to the
formation of areas devoid of blood flow due to the induction of
apoptosis in endothelial and other cells and to the collapse of
capillaries and first order vessels. Our results demonstrate that
proteasomal inhibitors such as PSI may prove effective as novel
anti-angiogenic and anti-neoplastic substances.—Drexler, H. C. A., Risau, W., Konerding, M. A. Inhibition of proteasome
function induces programmed cell death in proliferating endothelial
cells.
Key Words: angiogenesis • proteolysis • cell cycle • p27Kip1
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2
).
From several studies it has become clear recently that the
ubiquitin-proteasome system is also linked to both apoptotic cell death
and caspase activation (3
4
5
6
7
8
9
10)
. Two types of cellular
responses have been described when cells are treated with inhibitors of
proteasomal activity. In tumor cells such as U937, MOLT-4, L5178Y, Rvc
lymphoma, and HL60 cells, interference with proteasomal function
constitutes one of the signals capable of engaging the death machinery
(5
6
7
8
, 11)
. In these rapidly proliferating transformed
cells, the proteasome-mediated hydrolysis of certain proteins
apparently has to occur in a constitutive fashion to prevent apoptotic
death in the absence of another specific death signal. In contrast to
these cells, the opposite response is observed in differentiated,
quiescent primary neurons as well as in thymocytes; inhibition of
proteasomal activity actually reduces the extent of apoptosis in these
cells (3
, 4)
.
Based on these results, we have hypothesized that the proliferative
status of a particular cell type may exert a distinct influence on
cellular viability when proteasomal activity is blocked (5
, 12)
. Thus, HL60 cells that have been terminally differentiated
to nonproliferating macrophage-like cells by phorbolester treatment
are significantly less sensitive to the induction of apoptosis by PSI
compared to the rapidly proliferating untreated HL60 precursors
(5)
. Similar observations were made in a p53-dependent
cell death model involving rat-1 fibroblasts, in which the state of
confluence determines the sensitivity toward apoptosis induction by a
proteasome inhibitor (9)
.
Using primary bovine endothelial cells as well as an human endothelial
cell line, we attempted to further delineate the link between the
proliferative status and apoptosis induction by proteasomal inhibitors
in vitro and in vivo. We studied endothelial
cells because they are central to the pathophysiology of various
disease states, characterized by extensive formation of new blood
vessels such as tumor growth, diabetic retinopathies, or rheumatoid
arthritis (for review, see ref 13
). Under these conditions, quiescent
endothelial cells within established blood vessels become activated by
the action of paracrine growth factors such as vascular endothelial
growth factor, reenter the cell cycle, and resume proliferation
(14
15
16
17)
. This situation is closely recapitulated by
endothelial cells in vitro: They remain within the cell
cycle when kept under subconfluent culture conditions but become
largely quiescent when reaching confluence, thus providing a suitable
experimental system to address the question of differential
sensitivities toward proteasomal inhibition. Any compound that
preferentially will kill proliferating endothelial cells but leave the
quiescent endothelium unharmed would be very useful in therapeutic
approaches involving the vasculature as target.
The chorioallantoic membrane (CAM) of the developing chick embryo was
selected as a complimentary in vivo assay system. The CAM is
a densely vascularized and rapidly growing extra-embryonic membrane
that has been used for many years to investigate the effect of a
variety of substances on the formation of new blood vessels
(18)
as well as for studies targeted at the mechanisms of
tumor cell invasion (19)
.
Here we present evidence to support the assertion that induction of endothelial apoptosis by proteasomal inhibitors preferentially occurs in proliferating endothelial cells and could represent a useful approach in the search for novel anti-vascular and anti-neoplastic strategies.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and maintained in MDCB131 medium supplemented with
8% fetal calf serum, penicillin (100 IU/ml), and streptomycin (100
µg/ml) at 5% CO2. Human umbilical vein
endothelial cells (HUE; ATCC CRL-1730) were cultured under the same
conditions. Madin Darby canine kidney epithelial cells (MDCK; ATCC
CCL-34) were grown in Dulbecco’s modified Eagle’s medium (DMEM)
containing 10% fetal calf serum, penicillin, and streptomycin.
DNA fragmentation assay
HUE cells that had been plated at 20000
cells/cm2 24 h before the assay were treated
with 250 µM of the caspase inhibitors acetyl-Tyr-Val-Ala-Asp-aldehyde
(Ac-YVAD-H) or acetyl-Asp-Glu-Val-Asp-aldehyde (Ac-DEVD-H) (Peptide
Institute, Osaka, Japan) or the equivalent volume of the solvent
dimethyl sulfoxide (DMSO) for 1 h prior to addition of the
proteasomal inhibitors. Incubation of the cells was continued for a
further 18 h and the extent of DNA fragmentation was analyzed as
described previously with some minor modifications (21)
.
Briefly, cells were lysed in a solution containing 0.5% Triton X-100,
10 mM EDTA, and 10 mM Tris-HCl pH 7.5 for 30 min on ice, the lysate was
centrifuged at 13,000 x g for 10 min at 4°C, and the
resulting supernatant was sequentially extracted with phenol, followed
by phenol/chloroform/iso-amylalcohol (25:24:1; v/v/v). Low molecular
weight DNA was precipitated by the addition of 0.5 vol 7M ammonium
acetate and 2.5 vol ice-cold ethanol. The precipitate was washed in
75% ethanol, incubated for 30 min at 37°C in TE buffer containing
RNase A (20 µg/ml; Boehringer Mannheim), and subjected to
electrophoresis on 1.5% agarose gels.
Determination of proteasomal activity
BAE cells (1x106) were plated onto 10 cm
plastic dishes on the day before the actual experiment and then treated
for 16 h with the protease inhibitors leupeptin, E64, calpain
inhibitor II (ALLM), carbobenzoxy-Leu-Leu-norval-aldehyde (LLnV;
MG115), or PSI (50 µM final concentration), which were diluted from
1000x DMSO stock solutions. DMSO at a final concentration of 0.1%
served as a vehicle-only control. Floating and adherent cells were
harvested, lysed by sonication in ice-cold solution of 20 mM Tris-HCl
(pH 7.5) containing 20% glycerol, centrifuged in a bench-top
centrifuge (10 min, 2000xg), and the protein concentration
of the resulting supernatant determined by the BCA assay
(22)
. A total of 15 µg of protein in a volume of 200
µl lysis buffer supplemented with 5 mM CaCl2
was assayed for chymotryptic and peptidyl-glutamyl-hydrolyzing activity
in the presence of 10 µM succinyl-Leu-Leu-Val-Tyr-aminomethylcoumarin
(Suc-LLVY-AMC; Bachem, Heidelberg, Germany) and 10 µM
carbobenzoxy-Leu-Leu-Glu-aminomethylcoumarin (Z-LLE-AMC; Peptide
Institute), respectively.
Determination of caspase activities
HUE cells (8x105) were plated onto 10 cm
plastic dishes, allowed to attach overnight, and inhibitors were added
to a final concentration of 50 µM for the times indicated in Fig. 4
.
DMSO, which served as the vehicle, was added to a final concentration
of 0.1%. Floating and adherent cells were then harvested, lysed in
buffer I (1% Triton X-100, 10 mM KCl, 20 mM HEPES pH 7.4, 1.5 mM
MgCl2, 0.2 mM EGTA, 1 mM PMSF), snap-frozen in
liquid nitrogen, and stored at -80°C until required.
|
Lysates were centrifuged at 13,000 x g for 5 min at 4°C; the supernatants were collected and protein concentrations were determined using the BCA assay. A 40 µl aliquot of each sample was assayed in the presence of 20 µM of the caspase-3 substrate acetyl-Asp-Glu-Val-Asp-aminomethylcoumarin (Ac-DEVD-AMC) or the caspase-1 substrate acetyl-Tyr-Val-Ala-Asp-aminomethylcoumarin (Ac-YVAD-AMC; Bachem) in buffer II (50 mM PIPES-KOH pH 7.0, 0.1 mM EDTA, 1 mM DTT, 10% glycerol; total volume 250 µl) for 40 min at 37°C and the fluorescence intensity was measured (excitation 350 nm/emission 430 nm). All samples were also assayed in the presence of the caspase inhibitors Ac-YVAD-H (10 µM) and Ac-DEVD-H (1 µM; Peptide Institute) to exclude the possibility of nonspecific substrate cleavage. Specific activity was expressed as pmol AMC liberated per minute and per microgram of protein in comparison to a standard curve calculated from the amount of free AMC (Sigma, Deisenhofen, Germany).
Western blotting
HUE cell lysates were prepared from cells treated with 50 µM
PSI for the times indicated in Fig. 2B
by lysing the cells
in 1% sodium dodecyl sulfate (SDS), 10 mM Tris-HCl pH 7.5 and
denaturation at 95°C for 10 min. Equal amounts of protein were
separated by 12% SDS-polyacrylamide gel electrophoresis, transferred
to nitrocellulose filters, and blocked overnight with TBST (50 mM Tris
pH 7.5, 200 mM NaCl, 0.05% Tween 20) containing 5% non-fat dry milk
powder. Membranes were incubated for 1 h with primary antibodies
obtained from DAKO (Bcl-2), Santa Cruz, Heidelberg, Germany (Bcl-xL,
Bax, c-myc), Transduction Laboratories, Dianova, Hamburg,
Germany (p27Kip1, bad, ß-catenin), Calbiochem,
Bad Homburg, Germany (PARP), and Biogenesis, Poole, United Kingdom
(glyceraldehyde phosphate dehydrogenase; GAPDH). After washing and
incubation with peroxidase-conjugated secondary antibodies (Dianova),
blots were developed with the SuperSignal chemoluminescence substrate
(Pierce, KMF, St. Augustin, Germany).
|
Viability assay
To determine the differential effect of proteasomal inhibitors
on the viability of subconfluent and confluent cells, MTT assays were
performed as described previously (23)
. Briefly, BAE cells
were plated either at 3000 cells/well (low density) or 30000 cells/well
(high density) onto 96-well plates and grown overnight. Protease
inhibitors were then added in triplicate from 100-fold concentrated
stock solutions prepared in DMSO at final concentrations between 100 pM
and 10 µM (see Fig. 6
) to the low density cultures and incubated for
an additional 48 h prior to the addition of the MTT solution. The
formazan crystals generated were subsequently dissolved in DMSO and
absorption measured at 570–650 nm. High density cultures were grown to
confluence for at least 4 days after plating and prior to the 48 h
incubation period in the presence of the protease inhibitors.
|
FACS analysis
Analysis of cell cycle distribution was performed as described
(24)
, using a FACScan flow cytometer equipped with
CellQuest and ModFit LT software (Becton Dickinson, Heidelberg,
Germany).
Chorioallantoic membrane assays
All experiments with chick embryos were performed in
ovo. After aspiration of 2 ml of egg white from the pointed side
of the egg, a window was cut into the eggshell on embryonic day 3 or 4
(E3, E4). The eggs were resealed with tape and further incubated until
E9.
To obtain an indication of the number of cycling cells in the CAM tissue during the incubation period, a total of 4 E9 chick embryos were injected with 25 mg/kg BrdU (Sigma) into an appropriate vein of the chorioallantoic membrane, returned to the incubator, and cultivated for an additional 24 h. Approximately 1 cm2 pieces of the CAM were then cut out, washed in phosphate-buffered saline, and fixed in Carnoy’s solution. After dehydration and embedding in paraffin, 8 µM sections were cut and stained with an anti-BrdU mouse monoclonal antibody (Boehringer Mannheim, Germany) according to the manufacturer’s instructions. Counterstaining of the sections was performed with Gill’s hematoxylin (Serva, Heidelberg, Germany); sections were mounted using Aquamount (Merck, Darmstadt, Germany) and photos were taken with a Zeiss Axiophot microscope.
To investigate inhibitor efficacy on the CAM tissue, 1 part of a 5 mM PSI solution in DMSO was carefully added to a solution made by mixing 3.5 parts of a sterile 1% methylcellulose solution and 2.5 parts sterile water. A series of 7 µl aliquots of this mixture, each containing a total amount of 3.1 µg or 1 µg of PSI, were pipetted onto an bacteriological grade petri dish, air dried, and placed onto E10 CAMs for 2 days. Control pellets were prepared with an equal amount of leupeptin or with DMSO as vehicle. To aid in the visualization of the vascular system, embryos and CAMs were then injected with 10% Luconyl blue (BASF, Ludwigshafen) in phosphate-buffered saline using pulled glass capillaries and photographs were taken using a stereo microscope with camera adapter.
Microvascular corrosion casting and electron microscopy
The microvascular architecture of the CAM located below the PSI
and control pellets was highlighted by means of microvascular corrosion
casting. The windows of the egg shells of eight eggs were enlarged
under a stereo microscope. After a peripheral incision in the CAM was
made, the main artery was incised with a pair of ocular scissors and a
glass capillary (World Precision Instruments, Inc. Germany), pulled to
the adequate diameter, was inserted and sealed with a drop of
cyanoacrylate (Loctite). Efflux was enabled by making another incision
into a main draining vessel. The vascular system was flushed with up to
5 ml of prewarmed (37–39°C) physiological saline (0.9% NaCl) under
mechanically controlled pressure. Fixation was performed with up to 5
ml of 2.5% phosphate-buffered glutaraldehyde (pH 7.40, 300 mosmol)
over a period of 2–5 min. A methacrylate based acrylic resin solution,
prepared by mixing 4 ml of Mercox CL-2B, 0.2 g of catalyst MA
(Vilene Med. Co., Tokyo, Japan) and 1 ml of methylmethacrylate monomers
(Merck), was used as casting medium. After complete polymerization of
the casting solution, the membranes were dissected and macerated in a
solution of 5% potassium hydroxide (Fluka, Neu-Ulm, Germany) at 40°C
for 1–2 days. The resulting corrosion casts of all specimens were
freeze-dried, mounted, and sputter-coated as described
(25)
. Examination of the microvascular corrosion casts
were carried out with a Stereoscan Mk-250 scanning electron microscope
(Cambridge, England). Alternatively, in situ fixed CAM
segments were cut out, processed for semi-thin sectioning, and stained
with methylene blue and Azur II (26)
. Selected specimens
were trimmed for ultra-thin sectioning and counterstained with uranyl
acetate and lead citrate.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Cell death induction experiments with proteasome
inhibitors were carried out with both endothelial cell types in order
to confirm that the observed effects were not specific for one cell
type only. PSI-induced endothelial cell death was also linked to
cleavage of chromatin into oligonucleosomal fragments, a hallmark of
apoptotic cell death (Fig. 2A
, lane 1), as well as to the processing of
poly(ADP-ribose)polymerase (PARP) and to that of ß-catenin (Fig. 2B
), substrates for caspase-3 and caspase-6, respectively
(27
, 28)
. The Western blotting experiments were carried
out with HUE cell lysates only, as several antibodies available for
this study did not react with the corresponding bovine antigens.
|
Two other inhibitors of proteasomal function, LLnV and LLL (MG132), had
the same effect as PSI (not shown). In contrast, the addition of ALLM,
a potent calpain inhibitor (Ki=9 nM) but only a weak
inhibitor of proteasomal function (29)
, to cultured BAE
cells did not induce any changes in cellular morphology or viability
(not shown).
To confirm that proteasomal activity is actually inhibited in
cells treated with proteasomal inhibitors, the cleavage of Suc-LLVY-AMC
and Z-LLE-AMC was monitored in total cell lysates that were prepared
15 h after incubation of BAE cells with various inhibitors.
Cleavage of both substrates, which are processed by the chymotryptic
(Fig. 3A
) and peptidyl-glutamyl hydrolyzing activity of the
proteasome, respectively (Fig. 3B
), was clearly reduced with
PSI and LLnV but not with leupeptin, E64, or the calpain inhibitor ALLM
(Fig. 3A, B
). Further evidence that proteasomal function is
blocked by the action of PSI is derived from the observation that the
relative protein levels of the cyclin-dependent kinase inhibitor (cki)
p27Kip1 start to accumulate between 6 and 9 h after addition of the proteasome inhibitors (Fig. 2B
).
|
The anti-p27Kip1 antibody we used also detected a
band of ~22 kDa in the cell lysates collected 15 and 24 h after
addition of the inhibitor. Cleavage of the ckis p21 and p27 has been
linked to caspase activation and the induction of apoptosis in human
umbilical vein endothelial cells subjected to growth factor deprivation
(30)
, further supporting the observation that endothelial
cell death follows an apoptotic pathway. In contrast to
p27Kip1, no change in the relative amounts of
either GAPDH, the anti-apoptotic proteins bcl2 and bcl-xl, or the
proapoptotic protein bad were observed in the Western blotting
experiments with lysates of PSI-treated cells (Fig. 2B
).
PSI-induced apoptosis in endothelial cells can be blocked by
Ac-DEVD-H but not by Ac-YVAD-H
The induction of apoptosis in the monocytic cell line HL60 by
blocking proteasomal function is dependent on the activation of a
caspase-3-like enzyme, which is accompanied by cleavage of PARP. This
apoptotic response can be prevented by preincubation of the cells with
Ac-DEVD-H, but not with Ac-YVAD-H. In an analogous manner, a
caspase-3-like enzyme also becomes activated during PSI-mediated
endothelial cell apoptosis. Cleavage of the florigenic caspase-3
substrate Ac-DEVD-AMC was detected in HUE cell lysates between 6 and
9 h after addition of the proteasome inhibitor PSI and reached a
maximum at 15 h (Fig. 4A
), which is consistent with the kinetics observed for PARP
and ß-catenin cleavage in HUE cells (Fig. 2B
). In
contrast, caspase-1-like activity could not be detected throughout the
course of these experiments (Fig. 4B
). The rise in
caspase-3-like activity (Fig. 4A
) was not due to nonspecific
substrate hydrolysis by the total cell lysate, since inclusion of the
caspase-3 inhibitor Ac-DEVD-H in the assay mixture completely abrogated
the measured caspase-3-like activity to background levels (not shown).
In addition, only the caspase-3 inhibitor Ac-DEVD-H blocked the
morphological changes and DNA fragmentation that occurred as a result
of the PSI treatment (Fig. 2A
, lane 2), whereas the
caspase-1 inhibitor Ac-YVAD-H was unable to inhibit the action of PSI
(Fig. 2A
, lane 3). Neither of the two caspase inhibitors
alone or the solvent DMSO displayed any effect on the pattern of DNA
fragmentation (Fig. 2B
, lanes 4–6). We conclude from these
results that in endothelial cells, proteasomal inhibitors induce cell
death through an apoptotic pathway and that apoptosis occurs
independently of caspase-1-like family members (caspases 1, 4, and 5),
but instead is mediated by a caspase-3-like activity (caspases 3, 6,
and 7). It remains to be investigated, however, whether activator
caspases, such as caspase-8 or caspase-9, are also involved during this
type of apoptosis.
Proliferating and confluent endothelial cells display differential
sensitivity toward inhibition of proteasome function
To determine whether the endothelial cell response toward
proteasomal inhibition would be different in actively proliferating and
in contact-inhibited, quiescent cells, cellular viability was assessed
by measuring the capacity of the cells to reduce the tetrazolium salt
MTT (Fig. 5
). A dramatic loss of viability was observed in BAE cells plated at low
cell density, which had been incubated for 2 days with PSI
(EC50=24 nM). Cellular viability was also reduced
under these circumstances with LlnV, albeit to a lesser degree
(EC50=402 nM). Similar results were obtained at
low cell density with HUE cells (EC50,PSI=7 nM;
EC50,LLnV=310 nM) and with MDCK epithelial cells
(EC50,PSI=16 nM;
EC50,LLnV=196 nM), indicating that cell death is
effectively induced in proliferating cells at rather low concentrations
of PSI and LLnV. The observed cytotoxic effect apparently is not
specific for endothelial cells but affects other epithelioid cell types
as well, such as MDCK cells, which show a similar growth behavior.
|
BAE cells, in contrast, that had been plated at high density and grown
to confluence for 4 days displayed a greatly reduced sensitivity toward
treatment with PSI or LLnV (EC50,PSI=8.15 µM;
EC50,LLnV=2.72 µM): a 340-fold higher
concentration of PSI and a 6.8-fold higher concentration of LLnV were
necessary to achieve a 50% reduction of viability. Similar results
were obtained with HUE cells grown to confluence (1119-fold for PSI;
8.1-fold for LLnV) and with confluent MDCK cells (194-fold for PSI,
15.1-fold for LLnV) when assayed under the same conditions. Table 1
summarizes the results of the viability assays.
|
FACS analysis confirmed that the percentage of BAE cells within the
S-phase compartment of the cell cycle was reduced by nearly 80% in
confluent cultures compared to BAE cells growing at low density
(Fig. 6A, B
). Similar results were obtained for HUE cells (not
shown).
From these observations we conclude that inhibition of proteasomal activity has a proapoptotic effect primarily in endothelial cells that are in a proliferative state and that significantly higher doses of proteasome inhibitors are necessary to achieve a similar degree of apoptosis in contact-inhibited, quiescent endothelial cells.
Proteasome inhibitors induce cell death and vascular regression
in vivo
To examine whether the inhibition of proteasomal activity would
exert a proapoptotic effect in a proliferating tissue in
vivo, a series of CAM assays were carried out. Since the results
of our in vitro experiments indicated that inhibition of
proteasome function preferentially affects proliferating endothelial
cells, we first wanted to determine which and how many cells of the CAM
would be entering the S-phase of the cell cycle during an extended
incubation period, similar to the one used for the treatment of the CAM
with inhibitor-loaded methylcellulose discs (see below). Chick embryos
were therefore labeled continuously for 24 h with BrdU, and
paraffin sections of the CAMs of these embryos were analyzed for the
locations of BrdU incorporation. Numerous endothelial cell nuclei were
labeled after this period apart from cells in the chorionic and
allantoic epithelium of the CAM and apart from mesenchymal cells
(Fig. 7A, B
). Even individual nuclei of endothelial cells within the
wall of large caliber vessels were labeled, indicating that although
these vessels have a more mature vessel wall composition than
capillaries and show an increased state of quiescence, endothelial
proliferation has not yet come to a complete halt. From these results
it became obvious that any compound that is applied to the CAM for a
period of more than 24 h would encounter numerous cycling
endothelial as well as epithelial and mesenchymal cells.
|
Next, PSI-, leupeptin-, or DMSO-containing methylcellulose discs were
placed onto the embryonic CAM for 48 h and then analyzed by a
variety of techniques. Paraffin sections of PSI- and DMSO-treated CAMs
were stained with the TUNEL technique and the nuclear dye Hoechst
33342. Intense labeling of cell nuclei with the TUNEL technique was
observed in cells that had been in close contact with the
PSI-containing methylcellulose disc (Fig. 7C
), with the
proportion of TUNEL-labeled cells diminishing in cell layers located
more distally to the disc. Cells that gave a positive signal with the
TUNEL technique were also found to contain condensed or fragmented
nuclei when sections were double-stained with the nuclear dye Hoechst
33342 (Fig. 7D
), indicating that cell death in fact occurred
by an apoptotic process. In contrast, in sections of DMSO-treated
control CAMs, apoptotic cells could be detected only with low frequency
(not shown).
Morphological examination as well as injection of Luconyl dye into the
embryonic vascular system revealed that entry of the dye into the
vessels of the area beneath the PSI-containing disc was prevented and
that smaller vessels in particular were no longer efficiently perfused
(Fig. 8A
). In several CAMs, small patent capillaries with a regular
vessel wall architecture could no longer be detected by morphological
analysis. Large diameter vessels within these CAMs were not affected to
the same extent as capillaries, because access of the Luconyl dye to
the lumen of these vessels was not occluded in most cases. In some
instances the gross morphological appearance of large-caliber vessels
appeared slightly irregular (compare Fig. 8C, D
).
|
DMSO that was used as solvent control and the same amount of leupeptin,
a tripeptide protease inhibitor containing a carboxy-terminal aldehyde
group similar to PSI or LLnV, did not elicit any significant changes in
the vascular patterning or perfusion efficacy in the CAM (Fig. 8B
). Corrosion casts show that in the area of the CAM
located directly under the PSI-containing discs, the densities of the
capillary plexus as well as those of the first order vessels are
markedly reduced (Fig. 8C
), in contrast to DMSO-containing
control discs (Fig. 8D
). Semi-thin sections of CAMs exposed
to PSI (Fig. 8E
), but not those exposed to DMSO (Fig. 8F
), revealed numerous pyknotic nuclei or nuclear fragments
and further support the notion that PSI induced massive apoptosis in
the CAM tissue. Endothelial cells with a cuboidal cell shape were
detected in individual vessels (Fig. 8E
). These endothelial
cells have lost their contact to neighboring endothelial cells and may
be in the process of detachment from the basement membrane. We have
found endothelial cells with a similar morphology in the Tunica
vasulosa lentis, a dense capillary plexus associated with the
posterior half of the developing lens, after the onset of vascular
regression in this system (31)
.
The induction of apoptosis by PSI was not restricted to endothelial
cells but was observed throughout all layers of the CAM. In addition to
the effects on the vasculature in terms of occlusion and vessel wall
destruction, the semi-thin and ultra-thin sections revealed an increase
of the total CAM thickness in some instances (Fig. 8E, G
).
In sections taken from control CAMs, however, only occasional
disruptions of the amniotic epithelium were seen in response to the
discs (Fig. 8F, H
). Table 2
summarizes the results of the CAM experiments.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
PSI as well as LLnV have been used successfully as inhibitors of
proteasomal function in various experimental systems (29
, 32)
. In this study, we first addressed the question of whether
proteasome inhibitors would be able to selectively induce apoptosis in
proliferating endothelial cells and spare endothelial cells that are
rendered quiescent by contact inhibition. Treatment of subconfluent
endothelial cells with proteasomal inhibitors led to a marked reduction
of endothelial cell viability. Investigation of various parameters such
as cellular morphology, nuclear condensation and fragmentation, and DNA
laddering clearly indicated that the mode of endothelial cell death
inflicted by proteasomal inhibitors followed an apoptotic and not an
necrotic pathway. In addition, a caspase-3-like enzyme becomes
activated during endothelial cell apoptosis, which can be blocked by a
caspase-3 inhibitor. The cleavage of PARP and ß-catenin is also
indicative of the activation of such a caspase (27
, 28)
.
Thus, proteasomal inhibitors are capable of eliciting apoptotic cell
death not only in immortalized or tumor cell lines, as shown before,
but can also induce apoptosis in proliferating primary cells.
Our results also confirmed that PSI is able to effectively block
proteasome-mediated degradation of protein in endothelial cells; this
is in the first line shown by the early and strong increase in the
relative amounts of the cki p27Kip1 that starts
between 3 and 6 h after addition of PSI.
p27Kip1 is a well-known substrate of the
ubiquitin-dependent proteasomal degradation pathway and is regulated in
a cell cycle-dependent fashion under physiological conditions
(33)
. The reduction of proteasome-mediated proteolysis
observed after treatment with PSI and LlnV is corroborated by the
reduced levels of total cellular chymotryptic as well of the postacidic
residue cleaving activity in endothelial cell extracts. Other protease
inhibitors such as E64, leupeptin, or the calpain inhibitor ALLM did
not produce the same effect when added to the cells. Both ALLM and LLnV
have been shown to potently inhibit cleavage of cathepsin B and calpain
substrates with similar Ki values
(Ki=9–12 nM) (29)
, but the observation
that ALLM could not induce endothelial apoptosis strongly argues
against a role for calpains or cathepsin B in this apoptosis model.
Thus, the reduction of two enzymatic activities related to proteasome
function along with the increase in the relative amounts of
p27Kip1 and the lack of any measurable
ALLM-mediated effect suggest that PSI as well as LLnV primarily affect
proteasomal activity in endothelial cells.
Cycling cells in general seem to be particularly prone to the
inhibition of proteasomal function (5
6
7
8
, 11)
. This aspect
of proteasome functionality, at least in proliferating cells,
apparently resides in the constitutive and ordered removal of
protein(s), which, if accumulating within the cell, will promote the
activation of caspases. Various proteins have been described that are
subject to regulation by proteasomal degradation and at the same time
play a functional role in apoptosis. These include the tumor suppressor
gene p53 (reviewed in 34
), the proto-oncogene
c-myc (35)
, the transcription factors E2F
(36)
and c-fos (37)
, and bax, one of the
proapoptotic members of the bcl-2 family (38)
.
The exact mechanism and the effector molecules responsible for
apoptosis induction by proteasomal inhibitors are unknown. It is not
clear, for instance, whether cell death induction by inhibition of
proteasome function is dependent on the presence of p53
(9)
, as it can also occur in cells that lack a functional
p53 gene (5
, 45)
. Also, NF
B activation, which has been
demonstrated to promote survival (46
47
48)
and to control
the expression of certain anti-apoptotic genes (49
50
51
52)
,
is suppressed by proteasomal inhibitors and could contribute to
endothelial apoptosis. However, overexpression of a dominant negative
IB protein, at least, is not toxic to endothelial cells by itself
and always requires the action of an additional apoptotic stimulus such
as tumor necrosis factor 
(53
54
55)
.
The increase observed in p27Kip1 apparently
precedes the cleavage of PARP and ß-catenin by 3 h, indicating
that caspase activation occurs either concomitantly or shortly after
the p27Kip1 accumulation becomes overt. It may be
speculated that such an aberrant accumulation of the cki,
p27Kip1, which usually blocks transition into the
S-phase of the cell cycle, is in opposition to a cellular machinery
that is primed to carry out further cell divisions. Under such
conditions the cell may not have the appropriate mechanisms to handle
this signaling conflict other than launching the cell death program
(39)
. Such a scenario of an internal signaling conflict as
the mediator of apoptosis is supported by experiments in which
adenovirus-mediated overexpression of p27Kip1 in
multiple cancer cell lines is sufficient for the induction of apoptosis
(40
, 41)
. Reduced p27Kip1 levels, on
the other hand, have been associated with enhanced tumor growth and a
decreased survival prognosis for cancer patients (42
43
44)
.
Enhanced cdk activity due to caspase-mediated cleavage of the cki
proteins p21Waf1/Cip1 and p27Kip1 has also been
suggested to be ultimately responsible for the cell death of human
umbilical vein endothelial cells (30)
. In our experiments,
though, p27Kip1 accumulation and the onset of
apoptosis always preceded the appearance of a p22 cleavage fragment by
~9 h. Even at 24 h, when the p22 fragment is detectable,
considerably more uncleaved and intact p27 appears to be present in the
lysates of the inhibitor-treated cells than in the control cells at
0 h. Furthermore, a 22–23 kDa cleavage fragment of p27 Kip1 has
been described recently that is generated through a caspase-like
activity and acts as a potent growth inhibitor unrelated to apoptosis
induction (56)
. Thus, it seems unlikely that in our model
system enhanced cdk activity plays a causative role in the induction of
apoptosis.
In cells treated with the inhibitors lactacystin or MG132 (LLL), the
proapoptotic bcl-2 family member bax accumulates in a polyubiquitinated
form (38)
. Bax is able to form heterodimers with other
members of the bcl-2 family, including bcl-2 itself and bcl-xL, thereby
antagonizing the anti-apoptotic properties of these molecules. The bax
protein, however, could not be detected in Western blotting experiments
when using HUE cell lysates (under conditions that otherwise give a
clear signal with lysates prepared from other cell types), whereas the
relative levels of bcl-2 and bcl-xL remained unchanged. Therefore, we
exclude the possibility that accumulation of bax protein to abnormally
high levels contributes to apoptosis in endothelial cells elicited by
proteasomal inhibitors.
The pronounced proapoptotic effect of PSI on proliferating endothelial
cells in vitro prompted us to investigate whether PSI would
be able to induce apoptosis also in vivo in a proliferating
tissue, and we adopted the CAM model system for that purpose. In an
earlier study it was shown that the proteasome and cathepsin B
inhibitor lactacystin (57
, 58)
could reduce plasminogen
activator production in endothelial cells in vitro and would
thereby contribute to the inhibition of neovascularization in
vivo (59)
. However, as an histological examination of
the lactacystin-treated tissue has not been performed in this study,
the effects that lactacystin exerts in vivo currently are
not clear. We therefore carefully investigated the histological changes
associated with the treatment of the CAM tissue with the proteasomal
inhibitor PSI. Massive induction of apoptosis was observed in the CAM
of developing chick embryos when PSI was applied at 3.1 µg/disc and
(albeit less pronounced) at 1 µg/disc. The morphological changes
revealed by light and electron microscopy as well as by the detection
of cells containing fragmented nuclei and the identification of
TUNEL-positive cells further substantiated this finding.
The detection of apoptotic cells external to the vasculature is also in concurrence with the observations that PSI has an apoptosis-promoting effect not only on subconfluent endothelial cells, but also on other proliferating cells, such as MDCK kidney epithelial cells, and that cycling cells can be detected throughout the CAM tissue and not only within the vascular beds.
Prolonged PSI application to the CAM tissue resulted in the collapse of
vessels located beneath the polymer discs containing the inhibitor.
Injection of dye into the embryonic vascular system indicated that
access to the vessel lumen in the CAM tissue beneath the polymer disc
is obstructed. A likely explanation for this effect is that apoptosis
of endothelial cells comprising the wall of plexus vessels leads to the
collapse of these small vessels and to failure of perfusion. This can
be clearly seen in the semi-thin section shown in Fig. 8F
,
where the dense subepithelial capillary plexus present in the control
CAM (Fig. 8E
) is completely missing. The larger veins and
arteries deeper in the mesenchymal layer of the CAM do not appear to
respond as susceptible to the PSI-mediated proapoptotic effect and
remain largely unaffected, although they display slight irregularities
of the vessel wall surface. These large-caliber vessels may be more
mature than capillaries and hence may not contain the same percentage
of cycling endothelial cells, rendering them less susceptible to the
effects of proteasome inhibitors such as PSI. In addition, being a
lipophilic compound with a likely restricted diffusion range, PSI may
not be able to accumulate around these large-caliber vessels that are
located more distal from the polymer discs to sufficiently high enough
concentrations to induce apoptosis in these vessels.
Based on the observed differential sensitivity of proliferating and
quiescent endothelial cells to proteasomal inhibition in
vitro and on the anti-vascular effect that proteasomal inhibitors
exercise in vivo, we propose that such actively dividing
endothelial cells represent an attractive target for cell killing by
proteasomal inhibitors and that it is this impact on the vasculature
that contributes to the growth inhibitory effect that proteasomal
inhibitors have on certain tumors (60
, 61)
. Provided that
an optimized dosage regimen can be developed, destruction of the
vasculature of a tumor by proteasomal inhibitors, in combination with a
potent direct proapoptotic effect on the tumor cells, may constitute an
effective way to enhance conventional tumor therapy.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication May 4, 1999. Revised for publication September 1, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
ter, M., Becherer, J. D., Irmler, M., Tschopp, J., Martinou, J.-C. (1996) Involvement of the proteasome in the programmed cell death of NGF-deprived sympathetic neurons. EMBO J 15,3845-3852[Medline]
This article has been cited by other articles:
|
|
 |
|
  F. Zhang, A. J. Paterson, P. Huang, K. Wang, and J. E. Kudlow Metabolic Control of Proteasome Function Physiology, December 1, 2007; 22(6): 373 - 379. [Abstract] |
