Báo cáo Y học: Matrilysin (matrix metalloprotease-7) cleaves membrane-bound annexin II and enhances binding of tissue-type plasminogen activator to cancer cell surfaces docx

Results Cleavage of annexin II by matrilysin It was previously found that active matrilysin specifi-cally binds to surfaces of colon cancer cells and induces prominent cell aggregation [1

membrane-bound annexin II and enhances binding of

tissue-type plasminogen activator to cancer cell surfaces Jun Tsunezumi1,2, Kazuhiro Yamamoto1, Shouichi Higashi1,2and Kaoru Miyazaki1,2

1 Division of Cell Biology, Kihara Institute for Biological Research, Yokohama City University, Japan

2 Graduate School of Integrated Sciences, Yokohama City University, Japan

Matrix metalloproteinases (MMPs) form a group of

more than 20 zinc-dependent enzymes that are

involved in the processing of several components of

the extracellular matrix (ECM) They play roles in

many physiological processes, such as bone remodeling

and organogenesis, and have additional roles in the

reorganization of tissues during pathological

conditions such as inflammation and invasion and

metastasis of cancer cells [1,2] Many recent studies

have provided evidence that the biological activities of

various cell surface molecules are proteolytically

modulated by several MMPs, including membrane-type MMPs, gelatinase A (MMP-2), gelatinase B (MMP-9), stromelysin (MMP-3), and matrilysin (MMP-7) [3–6] These metalloproteinases are likely to regulate cellular functions by activating, inactivating or releasing membrane proteins Such regulation of cell surface proteins, as well as MMP-catalyzed degra-dation of the ECM, a natural barrier against tumor invasion, is important for tumor metastasis

Matrilysin, the smallest of the MMPs, has broad substrate specificity and has been demonstrated to

Keywords

annexin II; cancer cells; matrilysin; matrix

metalloproteinase; plasminogen activator

Correspondence

K Miyazaki, Division of Cell Biology, Kihara

Institute for Biological Research, Yokohama

City University, 641-12 Maioka-cho,

Totsuka-ku, Yokohama, Kanagawa 244-0813, Japan

Fax: +81 45 820 1901

Tel: +81 45 820 1905

E-mail: miyazaki@yokohama-cu.ac.jp

(Received 9 May 2008, revised 22 July

2008, accepted 30 July 2008)

doi:10.1111/j.1742-4658.2008.06620.x

Matrilysin (matrix metalloproteinase-7) plays important roles in tumor progression It was previously found that matrilysin binds to the surface of colon cancer cells to promote their metastatic potential In this study, we identified annexin II as a novel membrane-bound substrate of matrilysin Treatment of human colon cancer cell lines with active matrilysin released

a 35 kDa annexin II form, which lacked its N-terminal region, into the culture supernatant The release of the 35 kDa annexin II by matrilysin was significantly enhanced in the presence of serotonin or heparin Matri-lysin hydrolyzed annexin II at the Lys9–Leu10 bond, thus dividing the protein into an N-terminal nonapeptide and the C-terminal 35 kDa frag-ment Annexin II is known to serve as a cell surface receptor for tissue-type plasminogen activator (tPA) Although the matrilysin treatment liberated the 35 kDa fragment of annexin II from the cell surface, it significantly increased tPA binding to the cell membrane A synthetic N-terminal non-apeptide of annexin II bound to tPA more efficiently than intact annexin II This peptide formed a heterodimer with intact annexin II in test tubes and

on cancer cell surfaces These and other results suggested that the nonapep-tide generated by matrilysin treatment might be anchored to the cell mem-brane, possibly by binding to intact annexin II, and interact with tPA via its C-terminal lysine It is supposed that the cleavage of cell surface

annex-in II by matrilysannex-in contributes to tumor annex-invasion and metastasis by enhanc-ing tPA-mediated pericellular proteolysis by cancer cells

Abbreviations

ECM, extracellular matrix; MMP, matrix metalloproteinase; PVDF, poly(vinylidene difluoride); siRNA, small interfering RNA; TAPI-1, N-(R)-[2-(hydroxyaminocarbonyl)-methyl]-4-methylpentanoyl- L -naphthylalanyl- L -alanine-2-aminoethyl amide; tPA, tissue-type plasminogen activator.

degrade or process a variety of matrix and nonmatrix

molecules [7] Unlike most MMPs, which are expressed

by stromal cells, matrilysin is principally expressed by

epithelial cells [8] This enzyme seems to be one of the

most important MMPs in human colon cancers,

because the expression of matrilysin is highly

corre-lated with malignancy and metastatic potential of the

cancers, especially in their liver metastasis [9] It has

recently been reported that active matrilysin

specifi-cally binds to the surface of colon cancer cells and

induces notable cell aggregation due to processing of

the cell membrane protein(s) Furthermore, these

aggregated cells showed greatly enhanced metastatic

potential in the nude mouse model [10,11] Therefore,

it seems important to identify cell surface proteins that

are specifically cleaved by matrilysin, to elucidate the

mechanism of the matrilysin-induced phenotypic

changes of cancer cells, such as enhancement of

homo-typic cell adhesion and metastatic potential

Annexin II belongs to a family of calcium-dependent

phospholipid-binding proteins that are expressed in

diverse tissues and cell types [12] Annexin II was

ini-tially identified as an intracellular molecule without a

signal peptide, but later studies revealed extracellular

localization of annexin II in many kinds of tissues and

cells [13] The mechanism of secretion of cytoplasmic

annexin II is mostly unknown, but a stress-induced

protein secretion pathway has been suggested in

vascu-lar endothelial cells [14] Many studies have shown

that extracellular annexin II is involved in the

regula-tion of a variety of cellular processes, including

pericel-lular proteolysis, cell–cell or cell–ECM adhesion, and

regulation of membrane architecture [13–16] One of

the important functions of extracellular annexin II is

its involvement in the tissue-type plasminogen

activa-tor (tPA)–plasminogen system on cell surfaces [17]

The N-terminal sequence of annexin II is required for

its binding with tPA [18]

In this study, we identified annexin II as a novel

membrane-associated substrate for matrilysin, and

investigated the biological consequence of annexin II

cleavage by matrilysin Our results suggest that the

specific cleavage of annexin II by matrilysin enhances

the binding of tPA to cancer cell surfaces, leading to

activation of the tPA-mediated pericellular proteolytic

cascade on cancer cells

Results

Cleavage of annexin II by matrilysin

It was previously found that active matrilysin

specifi-cally binds to surfaces of colon cancer cells and

induces prominent cell aggregation [10,11] In the pres-ent study, we first analyzed membrane proteins that are cleaved by matrilysin A membrane fraction of WiDr human colon carcinoma cells was prepared by the phase separation method with Triton X-114 When the membrane fraction was treated with matrilysin, several proteins, including a major protein of approxi-mately 35 kDa, were released from the membrane fraction (Fig 1A) The N-terminal amino acid sequence of the 35 kDa protein was determined to be

B

MAT

116

97

200 (kDa)

A

97

66

45

31

21

338

10

9

1

9 10

STVHEILCK LSLEGD STPPSAYGSVKAYT……

Fig 1 Cleavage of membrane proteins by matrilysin (MAT) and identification of annexin II (A) Membrane fraction of WiDr cells was prepared by Triton X-114 phase separation as described in Experimental procedures The membrane fraction obtained from one confluent culture in a 90-mm dish (approximately 3 · 10 8 cells) was incubated without ( )) or with (+) 100 n M matrilysin at 37 C for 3 h The incubated membrane proteins were again subjected to phase separation with Triton X-114, and proteins released into the aqueous phase were separated by SDS/PAGE, transferred to

a PVDF membrane, and visualized by staining with Coomassie Brilliant Blue R250 Closed arrowhead, a major 35-kDa band identified as an annexin II fragment; open arrowheads, other major differential bands in the matrilysin-treated sample Ordinate, molecular sizes in kDa of marker proteins Other experimental con-ditions are described in Experimental procedures (B) Domain struc-ture of annexin II and the site where it is cleaved by matrilysin N-terminal sequence analysis of the 35 kDa protein band revealed that annexin II had been cleaved between Lys9 and Leu10.

LSLEGDHSTPPSAY by automated protein

sequenc-ing, and this sequence was identical to the amino acid

sequence from residues 10 to 23 of annexin II

(Fig 1B)

To determine whether annexin II is directly cleaved

by matrilysin, we used both a recombinant human

annexin II and a natural annexin II purified from

CaR-1 human colon carcinoma cells Matrilysin

effec-tively cleaved the 36 kDa recombinant annexin II and

converted it to the 35 kDa form (Fig 2) This cleavage

was inhibited by an MMP inhibitor,

N-(R)-[2-(hydrox-

yaminocarbonyl)-methyl]-4-methylpentanoyl-l-naphthyl-alanyl-l-alanine-2-aminoethyl amide (TAPI-1), but not

by a mixture of inhibitors for serine, aspartic and

cysteine proteinases The N-terminal amino acid

sequence of the 35 kDa, cleaved annexin II was

iden-tical to that of the membrane-derived annexin II

fragment (LSLEGDHSTPPSAY) These results

indi-cate that matrilysin cleaves the peptidyl bond between

Lys9 and Leu10 of annexin II (Fig 1B)

When the annexin II purified from CaR-1 cells was

analyzed by immunoblotting, it showed two distinct

bands at approximately 36 and 72 kDa under

non-reducing conditions, but a single 36 kDa band under

reducing conditions (Fig 3A) The 72 kDa protein was

thought to be a homodimer of annexin II cross-linked

with a disulfide bond Next, the natural annexin II was

incubated with matrilysin and four other MMPs, and

then analyzed by immunoblotting under nonreducing

conditions (Fig 3B) Matrilysin and MMP-2 almost

completely converted the 36 kDa annexin II to the

35 kDa cleaved form In addition, these MMPs also

extinguished the 72 kDa band, suggesting that the

72 kDa annexin II dimer had been cross-linked by a

disulfide bond between the cysteine residues of two monomer molecules at amino acid position 8 (Fig 1B) MMP-9 appeared to cleave annexin II weakly

Matrilysin-catalyzed cleavage of annexin II on the cell surface

Although annexin II does not have a signal sequence,

it is found on cell surfaces of many types of cultured cells [19–21] Indeed, flow cytometric analysis revealed the existence of annexin II on cell surfaces of WiDr cells (data not shown) To examine whether matrilysin cleaves annexin II on cell surfaces, three kinds of human colon cancer cell lines (CaR-1, WiDr and DLD-1) and a human breast cancer cell line (MDA-MB) were individually incubated with purified matrilysin in culture conditions After the treatment, annexin II fragments released into the culture media

−

+ +

(MAT)

(Inh.)

Native Cleaved (36)

(35)

Fig 2 Cleavage of purified annexin II by matrilysin (MAT)

Recom-binant annexin II (1 lgÆmL)1) was incubated at 37 C for 3 h with

(+) or without ( )) 50 n M matrilysin in the presence or absence of

the MMP inhibitor TAPI (4 l M ) or a proteinase inhibitor mixture

[Mix.; 0.2 m M 4-(2-aminoethyl)benzenesulfonyl fluoride, 0.16 l M

ap-rotinin, 0.025 m M bestatin, 7.5 l M E-64, 0.01 m M leupeptin, and

5 l M pepstatin] The digests were analyzed by immunoblotting with

an antibody against annexin II under reducing conditions Other

experimental conditions are described in Experimental procedures.

Arrowheads indicate native and cleaved annexin II bands at 36 and

35 kDa, respectively.

A

B

Fig 3 Immunoblotting of purified natural annexin II and its cleav-age by five kinds of MMP (A) Immunoblotting of natural annexin II purified from CaR-1 cells under nonreducing ( )) and reducing (+) conditions 2ME, 2-mercaptoethanol The bands at 72 and 36 kDa correspond to dimeric and monomeric forms of annexin II, respec-tively (B) The natural annexin II (2 lgÆmL)1protein) was incubated

in 50 lL of a reaction mixture without (None) or with 5 n M each of MMP-1, MMP-2, MMP-3, MMP-9, or matrilysin (MAT) (MMP-7) at

37 C for 16 h The sample was subjected to nonreducing SDS/ PAGE followed by immunoblotting analysis with annexin II anti-body Other experimental conditions are described in Experimental procedures Arrowheads, annexin II bands.

were analyzed by immunoblotting Although the

super-natants from the control cultures did not contain

annexin II at detectable levels, all supernatants from

the matrilysin-treated cultures contained the 35 kDa

annexin II fragment (Fig 4A) The amount of released

annexin II was highest in CaR-1 cells The whole

lysate of WiDr cells contained only the 36 kDa native

annexin II These results, as well as the result shown in

Fig 1A, demonstrate that matrilysin cleaves annexin II

on the cell surface and releases the 35 kDa, C-terminal

fragment of annexin II into the culture medium

The loss of cell surface annexin II after matrilysin treatment was confirmed using CaR-1 cells by two dif-ferent methods Cell ELISA indicated that the immu-noreactivity for cell surface annexin II was decreased

by about 40% after matrilysin treatment (Fig 4B) Immunofluorescence staining for annexin II also indi-cated partial loss of the immunosignals for annexin II

on the cell surface after matrilysin treatment (Fig 4C)

As MMP-2 cleaved purified annexin II in a test tube, as shown in Fig 3B, we also tested whether this MMP cleaved annexin II on the cell surface and released its soluble form When CaR-1 cells were trea-ted with active MMP-2, however, no annexin II frag-ment was detectable in the culture supernatant, suggesting that the cleavage of annexin II on the cell surface is specific for matrilysin (data not shown)

It is known that annexin II binds to glycosaminogly-cans such as heparan sulfate proteoglyglycosaminogly-cans and sialo-glycoproteins and phospholipids on the cell surface, and many of the interactions are mediated by calcium ions [22,23] Serotonin (5-hydroxytryptamine) interacts with N-acetylneuraminic acid, which is often contained

in glycolipids and glycoproteins on the cell surface [24] We examined the synergistic effects of matrilysin with serotonin, heparin and EDTA on the release of annexin II from the cell surface WiDr cells and CaR-1 cells were treated with serotonin, heparin or EDTA in the presence or absence of matrilysin, and the released annexin II was analyzed by immunoblotting (Fig 5) When WiDr cells were treated with each of these reagents, the intact (or full-length) annexin II was released into the culture supernatant at a higher level than the 35 kDa annexin II released by the matrilysin treatment alone When the cells were treated with sero-tonin or heparin in the presence of matrilysin, release

WiDr lysate

36 35

120

100

80

C ll ELISA ell ELISA with CaR-1 ith C R 1

xin II (%) 60

40

20

None MAT

0

A

B

C

Fig 4 Matrilysin-catalyzed cleavage of annexin II on the cell

sur-face (A) Three human colon carcinoma cell lines (CaR-1, WiDr, and

DLD-1) and human breast carcinoma cell line MDA-MB in

mono-layer cultures were incubated in 2 mL of serum-free medium with

(+) or without ( )) 50 n M matrilysin (MAT) at 37 C for 3 h Proteins

released into the culture medium were concentrated by

trichloro-acetic acid precipitation and analyzed by immunoblotting under

reducing conditions with the antibody against annexin II As a

con-trol, whole lysate of WiDr cells was run on the same gel (B) CaR-1

cells were treated with (MAT) or without (None) matrilysin as

above, and the amount of annexin II remaining on the cell surface

was measured by cell ELISA Each value represents the

mean ± SD of three independent results (C) CaR-1 cells were

treated with (MAT) or without (None) matrilysin as above, and

ann-exin II remaining on the cell surface was visualized by

immunofluo-rescence staining Detailed experimental conditions are described

in Experimental procedures.

Fig 5 Release of membrane-bound annexin II by matrilysin and three reagents CaR-1 cells and WiDr cells in monolayer cultures were incubated in the serum-free medium without ( )) or with (+)

50 n M matrilysin (MAT) in the presence or absence (None) of 0.2 m M serotonin or 1 mgÆmL)1 heparin at 37 C for 3 h as described in Fig 4 Alternatively, the same cultures were incubated with 5 m M EDTA at 25 C for 5 min Proteins released into the culture medium were analyzed by immunoblotting with the antibody against annexin II as described in Fig 4 Other experi-mental conditions are described in Experiexperi-mental procedures.

of cleaved annexin II was significantly increased as

compared with the level after the single treatment with

matrilysin In contrast to the case of WiDr cells, intact

annexin II was slightly or never released when CaR-1

cells were treated with serotonin, heparin or EDTA

alone However, when they were treated with both

matrilysin and serotonin or heparin, the amount of

cleaved annexin II released was significantly increased

These results suggest that annexin II is bound to sialic

acid-containing molecules and heparan sulfate

proteo-glycans on the cell surface, and the strength of

inter-action differs between the two cell types It seems

likely that serotonin and heparin promote matrilysin-catalyzed annexin II cleavage by weakening the inter-action of annexin II with the cell surface receptors

Binding of tPA to surfaces of matrilysin-treated cells

It has been shown that tPA binds to annexin II on the surfaces of human endothelial cells [18,25] In this study, we investigated whether annexin II cleavage by matrilysin affects tPA binding to the cell surface (Fig 6) First, we examined the effects of matrilysin

Fig 6 Effects of matrilysin and annexin II siRNA on binding of tPA to cancer cells (A) CaR-1 cells were treated with (+) or without ( ))

50 n M matrilysin (MAT) and/or 0.2 m M serotonin (Sero.) as shown in Fig 5 After the treatment, the cells were washed and then further incubated with 5 n M tPA and 5 l M TAPI-1 at 37 C for 1 h The incubated cells were collected, washed and dissolved in the SDS/PAGE sam-ple buffer and subjected to immunoblotting for tPA (top panel) and enolase as an internal loading control (center panel) Annexin II released into the culture supernatant by the matrilysin/serotonin treatment is shown in the lower panel (B) Detection of tPA bound to the cell surface

by cell ELISA CaR-1 cells were pretreated without (None) or with 50 n M matrilysin (MAT) on 96-well plates for 3 h, and incubated with tPA and TAPI-1 as above To quantify tPA bound to the cell surface, the cultures were subjected to cell ELISA according to the method described in Experimental procedures Each value represents the mean ± SD of triplicate assays (C) Enzymatic activity of tPA bound to the cell surface CaR-1 cells were treated with the indicated concentrations of matrilysin and then with tPA as shown above The catalytic activ-ity of tPA bound to the cell surface was assayed using the fluorogenic peptide 3145v as a substrate Each value represents the mean ± SD

of triplicate assays Annexin II released into the culture supernatant by the matrilysin treatment is shown in the lower panel (D) Effects of annexin II siRNA on tPA binding to matrilysin-treated cells CaR-1 cells were inoculated onto 24-well culture plates and treated with

annex-in II siRNA or a control RNA Two days later, these cells were treated with matrilysannex-in and then tPA as described above The cells were washed, lysed in the SDS/PAGE sample buffer and subjected to immunoblotting for tPA, annexin II (ANX-II) and enolase-1 as an internal loading control The cleaved annexin II (Sol ANX-II) released into the culture medium is shown in the upper panel Other experimental condi-tions are described in Experimental procedures.

and serotonin on tPA binding to CaR-1 cells

(Fig 6A) Unexpectedly, the single treatment with

matrilysin significantly increased the binding of

exoge-nous tPA to CaR-1 cells, while releasing the cleaved

annexin II into the culture supernatant Furthermore,

when CaR-1 cells were treated with both matrilysin

and serotonin, tPA binding to the cell surface was

greatly enhanced by the presence of serotonin The

enhancement of tPA binding to the cell surface was

coincident with the release of cleaved annexin II The

enhancement of tPA binding to the cell surface by

matrilysin was confirmed when the amount of tPA on

the cell surface was assayed by cell-based ELISA

(Fig 6B) Moreover, the assay of tPA activity on the

cell surface clearly showed that matrilysin treatment

increased tPA activity in a dose-dependent manner

(Fig 6C)

To show the direct link between tPA binding and

annexin II cleavage by matrilysin, we examined the

effect of suppression of annexin II expression on tPA

binding (Fig 6D) Treatment of CaR-1 cells with a

small interfering RNA (siRNA) for annexin II

signifi-cantly and specifically suppressed expression of

ann-exin II protein Suppression of annann-exin II expression

also suppressed binding of exogenous tPA to the

matrilysin-treated cells, as well as the release of cleaved

annexin II All of these results strongly suggest that

matrilysin enhances binding of tPA to cancer cells by

cleaving annexin II on the cell surface The bound tPA

maintained its enzymatic activity on the cancer cell

surface

Mechanism for matrilysin-induced tPA binding to

the cell surface

Matrilysin cleaves annexin II on the cell surface into

the N-terminal peptide Ac-STVHEILCK and the

C-terminal large peptide of 35 kDa, releasing the latter

peptide from the cell surface It was assumed that

Ac-STVHEILCK remained on the cell surface and

bound tPA To test this possibility, we used a synthetic

Ac-STVHEILCK peptide

First, the direct interaction between Ac-STVHEILCK

and tPA was examined using ELISA plates precoated

with the peptide At an optimal dose, the peptide

coat-ing increased the amount of tPA bound to the plates to

about twice that bound to the control plates (Fig 7A)

Next, tPA binding was examined in the presence or

absence of Ac-STVHEILCK on the plates precoated

with the purified, native annexin II or with the

annexin II cleaved by matrilysin The tPA binding was

slightly but significantly more efficient on the cleaved

annexin II than on the native one, and on either plate

the addition of the soluble N-terminal peptide signifi-cantly suppressed tPA binding to the plate (Fig 7B) These results suggested that tPA was able to bind Ac-STVHEILCK The competitive effect of the syn-thetic peptide was also examined for tPA binding to CaR-1 cells tPA, with or without the peptide, was applied to the cells pretreated with or without matrilysin (Fig 7C) Although Ac-STVHEILCK at 0.2 mm had

no effect on the nontreated cells, it significantly inhibited tPA binding to the CaR-1 cells pretreated with matrilysin These results strongly suggested that the matrilysin-enhanced tPA binding to cell membranes depended, at least in part, on the N-terminal peptide fragment Ac-STVHEILCK, which was generated by the matrilysin-catalyzed cleavage of annexin II

To obtain further evidence that the N-terminal peptide binds tPA on the cell surface, CaR-1 cells, without matrilysin treatment, were incubated with the N-terminal peptide and then with tPA Treat-ment of CaR-1 cells with the peptide increased the amount of tPA bound to the cell surface (Fig 8A) This implies that the N-terminal peptide binds both

an unidentified cell surface molecule and tPA on the cell surface Annexin II is known to form a hetero-tetramer complex, which consists of two annexin II molecules and two p11 (or S100A10) molecules [26] Indeed, the CaR-1 cell-derived annexin II formed a homodimer cross-linked with a disulfide bond, as shown in Fig 3 Therefore, it seemed possible that the N-terminal peptide bound an annexin II mono-mer on the cell surface To test this possibility, CaR-1 cells were first incubated with the N-terminal peptide, and then with heparin to release it from the cell surface, and the released annexin II was analyzed

by immunoblotting (Fig 8B) In the absence of hep-arin, annexin II was scarcely detected in the culture supernatant of CaR-1 cells When heparin was added

to the cells without the peptide treatment, a single band of the 36 kDa annexin II was detected in the culture supernatant However, when heparin was applied to the peptide-treated cells, the immunoblot-ting of the culture supernatant showed two close bands at 36 and 37 kDa under nonreducing condi-tions, but a single band of 36 kDa under reducing conditions The 37 kDa band appeared to be the heterodimer protein of the 36 kDa annexin II monomer with Ac-STVHEILCK cross-linked with a disulfide bond Indeed, the 37 kDa annexin II hetero-dimer was observed more clearly when purified ann-exin II was incubated with Ac-STVHEILCK in a test tube (Fig 8C, lane 4) The production of the

37 kDa annexin II heterodimer appeared to be enhanced when the treatment was done in the

presence of heparin (Fig 8C, lane 5) In addition,

the 37 kDa annexin II heterodimer was faintly

detected even when the purified annexin II was

digested by matrilysin (Fig 8C, lane 2), although it

did not increase in amount when the cleaved annexin II was incubated with the uncleaved form (lane 3) These results suggested that the 37 kDa nonapeptide–intact annexin II complex might be produced on the matrilysin-treated cancer cells However, we failed to recover the 37 kDa annexin II complex from the matrilysin-treated cells (data not shown) This is probably due to the low concentration of the peptide in the matrilysin-treated cells

It has been reported that tPA binds to a lysine residue via its kringle-2 domain [27] This suggests that tPA binds to the C-terminal lysine residue of Ac-STVHEILCK, which is produced from annexin II

by matrilysin treatment To test this possibility, we performed a competition assay using e-aminocaproic acid as a C-terminal lysine analog e-Aminocaproic acid at 10 mm strongly inhibited tPA binding to both the matrilysin-treated cells and the untreated cells (Fig 8D) This competitive effect was much more evident than that obtained with 0.2 mm Ac-STVHEILCK (Fig 7C) However, 1 mm e-amino-caproic acid scarcely inhibited tPA binding (data not shown) These results support the hypothesis that the N-terminal Ac-STVHEILCK peptide may be linked

to an annexin II monomer on the cell surface, and tPA efficiently binds to the C-terminal lysine of this peptide The relatively low blocking activity of exogenous Ac-STVHEILCK suggests that the membrane-bound peptide may have higher affinity than the free peptide

A

B

C

Fig 7 Interaction of tPA with Ac-STVHEILCK (A) Direct interaction between Ac-STVHEILCK and tPA Ninety-six-well microtiter plates were coated with the indicated concentrations of Ac-STVHEILCK (N-peptide) in NaCl/Pi at 4 C overnight These wells were fixed with 10% formaldehyde for 10 min and washed three times with NaCl/Picontaining 0.1% Tween-20 After blocking with 1.2% BSA

in NaCl/Pi, each well was incubated with 5 n M tPA at 37 C for 2.5 h After fixing, the relative amount of tPA bound to each well was assayed by ELISA as described in Experimental procedures (B) Binding of tPA to native and matrilysin-cleaved annexin II (ANX-II) proteins Purified natural annexin II (1 l M ) was digested with

50 n M matrilysin at 37 C for 3 h The untreated (Native) and digested (Cleaved) annexin II were individually coated on 96-well microtiter plates overnight Using these annexin II-coated wells, the tPA binding assay in the presence (+) or absence ( )) of 200 l M

Ac-STVHEILCK was carried out as described above (C) Competitive inhibitory effect of Ac-STVHEILCK on tPA binding to CaR-1 cells CaR-1 cells were incubated with 50 n M matrilysin on 96-well plates

at 37 C for 3 h The tPA binding to the CaR-1 cells in the presence (+) or absence ( )) of 200 l M Ac-STVHEILCK was analyzed as shown in Fig 5B In (A), (B) and (C), the data represent the mean ± SD of triplicate assays.

The present study identified annexin II as a novel

mem-brane-bound substrate for matrilysin Matrilysin

cleaved annexin II on the surfaces of human colon

cancer cells, releasing a major C-terminal sequence of

annexin II from the cell membrane The matrilysin

treatment of cancer cells facilitated the binding of tPA

to the cell surface We previously showed that active matrilysin efficiently binds to cholesterol sulfate on the cell membranes of colon cancer cells, retaining its enzy-matic activity [11] This MMP, together with cholesterol sulfate, was localized in the lipid microdomain so-called raft of cell membrane [11] Annexin II is also localized

in the membrane domain raft [28] Thus, it is highly likely that the membrane-bound active matrilysin

A

C

D

B

Fig 8 Interaction of tPA with Ac-STVHEILCK on the cell surface and its inhibition by e-aminocaproic acid (A) Enhancement of tPA binding

to cells by Ac-STVHEILCK CaR-1 cell were incubated with (+) or without ( )) 400 l M Ac-STVHEILCK (N-peptide) on 24-well plates containing serum-free medium at 37 C for 6 h and then with 5 n M tPA for 1 h The tPA bound to the cells, as well as enolase as an internal loading control, was analyzed by immunoblotting as shown in Fig 6A (B) Formation of 37 kDa heterodimer on cells treated with Ac-STVHEILCK CaR-1 cells were incubated with (+) or without ( )) 400 l M Ac-STVHEILCK and further incubated in the serum-free medium supplemented with (+) or without ( )) 1 mgÆmL )1heparin for 2 h Annexin II released into the culture medium was analyzed by immunoblotting under

non-reducing ( )2ME) and reducing (+2ME) conditions The 37 kDa band indicated by the open arrowhead seems to be a complex of annexin II with Ac-STVHEILCK Closed arrowheads indicate annexin II monomer (C) Formation of 37 kDa heterodimer in test tubes Purified annexin II (1 l M ) was incubated at 37 C for 2 h without (lane 1) or with (lanes 4 and 5) 100 l M Ac-STVHEILCK in 50 m M Tris/HCl (pH 7.5) containing

150 m M NaCl, 5 m M CaCl2and 0.01% Brji 35 in the absence (lanes 1 and 4) or presence (lane 5) of 100 lgÆmL)1heparin In other tubes, the purified annexin II (1 l M ) was incubated with 5 n M matrilysin at 37 C for 2 h, and the reaction was terminated by adding 10 l M TAPI-1 (lane 2) The cleaved annexin II (1 l M ) was incubated with the uncleaved annexin II (1 l M ) at 37 C for 2 h (lane 3) These samples were ana-lyzed by immunoblotting under nonreducing conditions (D) Inhibition of tPA binding to cancer cells by e-aminocaproic acid CaR-1 cells were treated with (+) or without ( )) matrilysin (MAT) on 24-well or 96-well culture plates and then incubated with 5 n M tPA plus 5 l M TAPI-1 in the presence (+) or absence ( )) of 10 m M e-aminocaproic acid (eACA) The amounts of tPA on the cell surface were analyzed by immuno-blotting (left panel) and cell ELISA (right panel) Each bar represents the mean ± SD of triplicate assays Other experimental conditions are described in Fig 6 and Experimental procedures.

efficiently cleaves annexin II on cancer cell surfaces.

MMP-2 and MMP-9 cleaved purified annexin II, but

they appeared not to cleave annexin II on the cell

sur-face, indicating that the cleavage of the

membrane-bound annexin II is specific for matrilysin The specific

cleavage of cell surface annexin II by matrilysin may

result from the specific binding of matrilysin to the

cancer cells In our previous study, among three MMPs

tested (matrilysin, MMP-2 and MMP-3), only

matrily-sin was able to bind to the cancer cells [10] and

choles-terol sulfate [11]

Annexin II is expressed in epithelial cells of various

tissues, including the epidermis, pancreas and breast

[19–21], and vascular endothelial cells [25] In these

kinds of cells, some annexin II molecules are found on

the cell surface Annexin II is known to interact with

membrane phospholipids and glycosaminoglycans such

as heparin, heparan sulfate [22,23] and fucoidan as a

sulfated fucopolysaccharide, in a calcium-dependent or

calcium-independent manner [29,30] Serotonin is

known to interact with glycolipids and glycoproteins

containing N-acetylneuraminic acid [24] In this study,

heparin, serotonin and EDTA released different

amounts of intact or matrilysin-cleaved annexin II from

cell membranes, suggesting that annexin II binds to cell

membranes via multiple receptors Our data also suggest

that the manner of annexinin II binding to cell

mem-branes varies considerably from one cell type to another

Membrane-bound annexin II is thought to play

impor-tant roles in various biological processes, such as

fibrinolysis [31], cell–ECM adhesion [13], protease

binding to cell membranes [15], ligand-mediated cell

signaling [32], and virus infection [33] One of the best

characterized functions of extracellular annexin II is its

action as a membrane-bound receptor for tPA on

vascu-lar endothelial cells [25,34] Some previous studies have

demonstrated that Cys8 in the N-terminal region is

essential for tPA binding to the cell surface [18]

Unexpectedly, the present study showed that

matri-lysin treatment of colon cancer cells led to marked

enhancement of tPA binding to the cancer cell surface,

although the tPA receptor annexin II was cleaved and

released from the cell membrane Our experiments with

the synthetic peptide Ac-STVHEILCK, which

corre-sponds to the N-terminal nine amino acid peptide of

annexin II generated by the matrilysin-catalyzed

cleav-age of annexin II, gave rise to the possibility that the

N-terminal peptide is responsible for the enhanced

binding of tPA to the cell surface First,

matrilysin-induced tPA binding to cells correlated well with the

extent of annexin II cleavage by matrilysin, and the

suppression of annexin II expression by siRNA

decreased tPA binding (Fig 6) Second, tPA bound to

the synthetic peptide coated on plastic plates in a dose-dependent manner, and the tPA binding was more effi-cient than that to the entire annexin II molecule (Fig 7A,B) Third, pretreatment of colon cancer cells with the synthetic peptide significantly increased tPA binding to the cells, whereas the peptide competitively suppressed tPA binding to the matrilysin-treated cells (Figs 7C and 8A) All these results support the hypoth-esis that the N-terminal annexin II peptide produced

by matrilysin remains bound to cell membranes and functions as a receptor for tPA tPA is known to have high affinity for lysine tPA binds to lysyl–Sepharose through its kringle-2 domain, and this interaction is blocked by l-lysine or e-aminocaproic acid as a C-ter-minal lysine analog [27] The kringle-2 domain of tPA directly interacts with e-aminocaproic acid [35] Krin-gle-2-mediated tPA binding to the C-terminal lysines plays an important role in the degradation of fibrin clots [36,37] Partial degradation of fibrin by plasmin generates C-terminal lysines, which function as new binding sites for tPA, resulting in further activation of plasminogen on the fibrin clot [38] On the basis of these facts, it seems very likely that tPA binds to the C-terminal lysine of the N-terminal annexin II frag-ment Ac-STVHEILCK remaining on cell membranes Although we cannot exclude the possibility that tPA binds to the C-terminal lysines of other protein frag-ments that are produced by the matrilysin activity, the result of the siRNA experiment with annexin II shown

in Fig 6D suggests that the annexin II fragment may play a major role in matrilysin-induced tPA binding

to the tumor cell surface On the basis of the amount

of annexin II fragment released by the treatment with matrilysin plus heparin, we estimated that at least 0.56–1.4 fmol of annexin II per 106cells (3.4– 8.4· 105molecules per cell) exists on the surface

of CaR-1 cells (Fig 5) We have determined that matrilysin binds to the tumor cell surface with a Kd

of 7 nm (K Yamamoto, J Tsunezumi, S Higashi and

K Miyazaki, unpublished data) The binding capacity

of tumor cells for matrilysin was estimated to be 0.25–1.0 fmol per 106cells (1.5–6.0· 105 molecules per cell) From the data shown in Fig 4B, it is also assumed that the matrilysin treatment produces at least 1.4–3.4· 105molecules per cell of the N-terminal ann-exin II peptide, most likely on the tumor cell surface The N-terminal sequence of annexin II has previ-ously been reported to interact with p11 (or S100A10), which is often regarded as the annexin II light chain,

to form a heterotetramer complex [26] In this study,

we detected the possible annexin II dimer of approxi-mately 72 kDa, but we failed to detect p11 in the annexin II complex with a specific antibody (data not

shown) Our data indicated that exogenous N-terminal

annexin II peptide bound to the cancer cell surface, and

the bound peptide was recovered as a 37-kDa

heterodi-mer complex with the intact annexin II molecule from

the cell surface when the cells were treated with heparin

This heterodimer complex was linked by a disulfide

bond, and was produced efficiently when the peptide

was incubated with the intact annexin II in test tubes

These results strongly suggest that the N-terminal

ann-exin II peptide remains as the 37 kDa complex with the

intact annexin II on the surfaces of matrilysin-treated

cells However, this possibility was not confirmed,

because we failed to detect this nonapeptide–annexin II

complex in the matrilysin-treated cancer cells (data not

shown) Thus, it is also possible that the N-terminal

annexin II peptide binds to cell membranes through p11

or other membrane molecules

The plasminogen activator–plasmin system is well

known to play important roles not only in fibrinolysis

but also in ECM degradation during tissue remodeling

[39,40] Like urokinase-type plasminogen activator,

tPA binds to some membrane proteins, including

ann-exin II [41] The binding of tPA or urokinase-type

plasminogen activator to the membrane receptors

greatly increases plasminogen activator activity on the

membranes, i.e the conversion of plasminogen to

plas-min [34,42] In addition, it has become evident that the

receptor binding of the two plasminogen activators

induces cell growth signaling without the need for their

proteolytic activities [32,43] In the present study, the

treatment of colon cancer cells with matrilysin resulted

in efficient cleavage of annexin II and enhanced

bind-ing of tPA to cell membranes We confirmed that the

tPA bound to the matrilysin-treated cancer cells

showed plasminogen activator activity, but we could

not detect significant activation of MAP kinase

signal-ing in the tPA-treated cells (data not shown) Many

types of MMP, including matrilysin, MMP-3 and

MMP-2/9, are efficiently activated by trypsin-like

serine proteinases [44–46] Plasmin is thought to be the

most important activator for these MMPs It is

supposed that the tPA on the cell membrane activates

the proforms of these MMPs by producing plasmin

[47] Thus, the cleavage of annexin II by matrilysin

may trigger the proteinase amplification cascade or

cycling in the pericellular space of cancer cells These

proteolytic activities are likely to promote the invasive

growth of tumor cells and subsequent metastasis

Past studies have suggested that matrilysin is

involved in the malignancy and metastasis of human

cancers [9] We have previously reported that active

matrilysin binds to the surfaces of colon cancer cells

and induces notable cell aggregation, probably due to

cleavage of membrane protein(s) [13,14] However, the cleavage of annexin II by matrilysin appeared not to induce cell–cell adhesion, because the suppression of annexin II synthesis by RNA interference did not inhi-bit cell aggregation (data not shown) Therefore, it is likely that membrane proteins other than annexin II are also degraded or processed by matrilysin These actions of matrilysin may also contribute to the malig-nant growth of cancer cells Understanding the patho-logical significance of the cleavage of annexin II and other membrane substrates by matrilysin seems to be important in designing new targets for cancer therapies

Experimental procedures Antibodies and other reagents

The sources of reagents used were as follows: human recombinant annexin II was from AmProx (Carlsbad, CA, USA); human Glu-plasminogen and Lys-plasminogen were from Hematologic Technologies (Essex Junction, VT, USA); tPA and Protease Inhibitor Cocktail Set III were from Calbiochem (San Diego, CA, USA); human recom-binant MMP-9, human recomrecom-binant interstitial

(Temecula, CA, USA); human recombinant matrilysin and 6-aminohexanoic acid (e-aminocaproic acid) were from Wako Pure Chemical Industries (Osaka, Japan); the MMP substrates 3145v (Pyr-Gly-Arg-MCA) and 3105v (Boc-Glu-Lys-Lys-MCA) and the synthetic MMP inhibitor TAPI-1 were from Peptides Institute (Osaka, Japan); and serotonin (5-hydroxytryptamine hydrochloride) was from Sigma Aldrich (St Louis, MO, USA) Commercial antibodies against human antigens used were: mouse monoclonal anti-body against tPA from Abcam (Cambridge, MA, USA); rabbit polyclonal antibody against enolase, mouse mono-clonal antibody against annexin II and rabbit polymono-clonal antibody against annexin II from Santa Cruz (Santa Cruz,

CA, USA); and goat polyclonal antibody against p11 from R&D systems (Minneapolis, MN, USA) The mouse mono-clonal antibody 11B4G against human matrilysin was a kind gift from T Tanaka (Nagahama Institute of Oriental Yeast Co., Shiga, Japan) Human MMP-2 was prepared in our laboratory as previously reported [10] The N-terminal annexin II peptide (Ac-STVHEILCK) was synthesized by Hayashi Kasei (Osaka, Japan) All other chemicals used for the experiments were of analytical grade or the highest quality commercially available

Cell lines and culture conditions

Four human colon cancer cell lines (Colo 201, DLD-1, WiDr, and CaR-1) and the human breast cancer cell line MDA-MB were obtained from Japanese Cancer Resources