Pathophysiology 11 (2004) 153–158

Mechanical stretch induces matrix metalloproteinase 1production in human hepatic stellate cells

Takashi Goto∗, Ken-ichiro Mikami, Kouich Miura, Shigetoshi Ohshima, Kazuo Yoneyama,Kunio Nakane, Daisuke Watanabe, Michiro Otaka, Sumio Watanabe

Department of Gastroenterology, Akita University School of Medicine, 1-1-1 Hondo, Akita 010-8543, Japan

Received 20 May 2004; received in revised form 14 June 2004; accepted 1 July 2004

Abstract

In increasing portal blood flow, hepatic stellate cells (HSC) may be lengthened in response to mechanical stretch stimulation and theirfunction may be changed. However, little is known about the influence of mechanical stretch on hepatic stellate cells. We examined productionof matrix metalloproteinases (MMP), tissue inhibitors of metalloproteinases (TIMP), and extracellular matrix by hepatic stellate cells toi ed cyclicallyu agen IIIp measuredb relative tou pressions ells increasep©

K

1

cdp[rci

tTpa

nsey be

eenury,xtra-P)m-inin,

torsro-

e toen 1tra-CNA

on

0d

nvestigate the relationship between mechanical stretch and hepatic fibrosis. LI90 cells, human hepatic stellate cells, were stretchsing the Flexer cell strain unit. Concentrations of MMP1, MMP2, TIMP1, TIMP2, type I collagen C-telopeptide (1CTP), procollropeptide (PIIIP), and hyaluronic acid in culture supernatants were determined. MMP1, MMP2, and TIMP1 mRNA expression wasy reverse transcription-polymerase chain reaction (RT-PCR). In stretched LI90 cells, concentration of MMP1 showed an increasenstretched cells, but concentrations of MMP2, TIMPl, and TIMP2 showed a decrease. MMP1/TIMP1 ratio and MMP1 mRNA exhowed an increase in stretched cells. Our finding suggested that in the early phase of portal hypertension, hepatic stellate croduction of MMPl and decrease production of TIMP1 and TIMP2, activated by mechanical stretch.2004 Elsevier Ireland Ltd. All rights reserved.

eywords:Portal hypertension; Hepatic fibrosis; Matrix metalloproteinases; Tissue inhibitor of metalloproteinases; LI90 cells

. Introduction

Hepatic stellate cells (HSC), which correspond to peri-ytes surrounding capillaries, are located within the suben-othelial space of Disse, are characterized by long, branchingrocesses and take part in the pathogenesis of liver disease

1,2]. Hepatic stellate cells store retinyl palmitate, deliveretinol to extrahepatic tissues, regulate sinusoidal microcir-ulation, synthesize extracellular matrix proteins, and partic-pate in fibrotic scarring in chronic hepatic diseases.

In chronic hepatic diseases, especially liver cirrhosis, por-al hypertension is one of the most important complications.he pressure gradient within the portal venous system is aroduct of vascular resistance (backward flow theory)[3,4]nd portal blood flow (forward flow theory)[5,6]. Increas-

∗ Corresponding author. Tel.: +81 188 84 6104; fax: +81 188 36 2611.E-mail address:takashi@doc.med.akita-u.ac.jp (T. Goto).

ing portal blood flow, HSC may be lengthened in respoto mechanical stretch stimulation and their function machanged.

Hepatic fibrosis results from a relative imbalance betwsynthesis and degradation of matrix proteins. In liver injboth the amount and relative composition of the hepatic ecellular matrix are altered. Matrix metalloproteinases (MMhave a function in the turnover of extracellular matrix coponents such as collagens, proteoglycans, elastin, lamfibronectin, and other glycoproteins, while tissue inhibiof metalloproteinases (TIMP) are low-molecular weight pteins that inhibit MMP in a 1:1 molar ratio[7,8]. Iredaleet al. reported that TIMP1 expression increased relativMMP1 and TIMP1 expression and preceded procollagexpression in liver after bile duct ligation and carbon techloride administration[9]. Herbst et al. reported that HSexpressed increased encoding of TIMP1 and TIMP2 mRin liver fibrosis and in vitro, TGF-�1 enhanced expressi

928-4680/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.pathophys.2004.07.003

154 T. Goto et al. / Pathophysiology 11 (2004) 153–158

of TIMP1 and MMP2 mRNA[10]. HSC, then, produceMMP and TIMP, shifting their balance in many kinds of liverinjury.

MMP1, also termed interstitial collagenase, degrades col-lagens I, II, III, VII, and X as well as gelatins, entactin,aggrecan, and link protein. MMP2, or gelatinase A, de-grades gelatins, collagens I, IV, V, VII, X, and XI, fi-bronectin, laminin, aggrecan, elastin, large tenascin C, andvitronectin. MMP2 also degrades�-amyloid protein precur-sor, thus showing�-secretase-like activity[8]. Type I andtype III collagen accounts for about 85% of all extracellularmatrix protein in the liver. Okazaki et al. reported that col-lagenase activity increased and MMP1 played an importantrole in experimental hepatic fibrosis[11].

Previous studies have revealed a relationship betweenmechanical stretch and matrix metalloproteinases and tis-sue inhibitors of metalloproteinases in many kinds of cul-tured cells. However, little is known about the influenceof mechanical stretch stimulation on production of MMPand TIMP in HSC, which play an important role in hep-atic fibrosis. Therefore, we examined production of MMP,TIMP, and extracellular matrix by HSC to investigatethe relationship between mechanical stretch and hepaticfibrosis.

2

2

LI90( hisc ioidh har-a rac-th ea con-n V, V,a edi ka,J ex-p ateds ort,P cul-t weres te of3 lex-c tes,a d vac-u retch( ottomc mu-l ain-t erec

2.2. Measurement of MMP, TIMP, and extracellularmatrix levels in culture supematants of LI90

Concentration of MMP1, MMP2, TIMP1, and TIMP2 inculture supematants was determined with enzyme immunoas-say (EIA) kits (Fuji Chemical Industries Ltd., Toyama,Japan). Concentration of type I collagen C-telopeptide(1CTP) (by radioimmunoassay), procollagen III propeptide(PIIIP) (by radioimmunoassay), and hyaluronic acid (bysandwich binding protein assay) in culture supematants wasdetermined.

2.3. MMP1, MMP2, and TIMP1 mRNA expression

MMP1, MMP2, and TIMP1 mRNA expression in LI90cells was measured by reverse transcription-polymerasechain reaction (RT-PCR). RNA was extracted using anIsogen kit (Nippongene, Tokyo, Japan); then, comple-mentary cDNA was synthesized from the RNA sam-ples using Moloney murine leukemia virus reverse tran-scriptase (MMLV-RT; GIBCO BRL, Gaithersburg, MD)at 37◦C for 1 h with random primers. Sequences cho-sen as primers were as follows[13,14]: MMP1, sense5′CAA AAT CCT GTC CAG CCC ATCG3′ and anti-sense 5′CGG CAA CTT CGT AAG CAG CTTC3′;M ′ ′as e5 e5 e5 -f lec-t ainedw ght.T ensit-o NIH,B g thef 1m

2

da madeb airedM toi

3

3

chedf ed afi

. Methods

.1. Cell culture

We used a cultured human hepatic stellate cell line,provided by Dr. Tomokazu Matsuura), in this study. Tell line, established by Murakami et al. from an epithelemangioendothelioma arising in human liver, is well ccterized as consisting of HSC-like cells that lack cha

eristics of endothelial cells or macrophages[12]. LI90 cellsad a polygonal shape, had well developed�-smooth musclctin filaments in their cytoplasm, and produced variousective tissue components, such as collagen types I, III, Ind VI, laminin, and fibronectin. LI90 cells were maintain

n a growth factor-enriched medium, GIT (WAKO, Osaapan) containing 10% fetal bovine serum (FBS). Foreriments, the cells were cultured on type I collagen-coilastic membranes in 6-well plates (Flexcell, McKeespA). After attachment of cells, about 70% confluent,

ure medium was changed to serum-free GIT, and cellstretched cyclically to attain from 10% elongation at a ra0 cycles/min for 24 h using the Flexercell strain unit (Fell). This system consists of flexible-bottom culture plavacuum base plate, hoses, and a computer-controlle

um pump. Computer settings determine the extent of stexpressed as a percentage) imposed on the flexible-bulture wells as well as the duration of mechanical stiation. The vacuum manifold that held the plates was mained at 37◦C in a humidified incubator with an atmosphontaining 5% CO2.

MP2, sense 5ACGGACAAAGAGTTGGCAGT3 andnti-sense 5′GTCACATCGCTCCAGACTTG3′; TIMP1,ense 5TTGTTGCTGGGCTGATAGC3′ and anti-sens′CAGGATTCAGGCTATCTGGG3′, and �-actin, sens′CAG AGC AAG AGA GGC ATC CT3′ and anti-sens′TAG CAC AGC ATG GAT AGC AA 3′. PCR was perormed for 25 cycles. Amplicons were analyzed by erophoresis on 3% agarose gels, after which they were stith ethidium bromide and observed under ultraviolet lihe density of these bands was analyzed by scanning dmeter using an NIH image program (Wayne Rasband,ethesda), and the relative density was calculated usin

ormula: relative density (%) = MMP1, MMP2, and TIMPRNA expression/�-actin mRNA expression.

.4. Statistics

Results are presented as the mean± S.D. for stretchend unstretched cells. Statistical comparisons wereetween stretched and unstretched cultures by unpann–WhitneyU tests. Ap value < 0.05 was considered

ndicate significance.

. Results

.1. Morphological change of stretched LI90 cells

In phase-contrast microscopic studies, LI90 cells stretor 24 h were narrower than unstretched cells and showner network of processes (Fig. 1).

T. Goto et al. / Pathophysiology 11 (2004) 153–158 155

Fig. 1. Phase-contrast microscopic views of LI90 cells. (A) Unstretched LI90 cells and (B) LI90 cells stretched for 24 h. Stretched LI90 cells were narrowerthan unstretched cells and showed a finer network of processes.

3.2. Concentration of MMP1, MMP2, TIMP1, andTIMP2 in culture supernatants of LI90 cells

The concentration of MMP1 in culture supernatantsshowed an increase in stretched as opposed to unstretchedLI90 cells, while in contrast the concentrations of MMP2,TIMP1, and TIMP2 showed a decrease in stretched cellscompared to unstretched cells: MMP1 in unstretched cells,197.8± 23.0 ng/ml, and stretched cells, 224.8± 13.9 ng/ml(p< 0.05); MMP2 in unstretched cells, 91.8± 6.6 ng/ml,and stretched cells, 68.8± 1.7 ng/ml (p< 0.05); TIMP1in unstretched cells, 62.8± 5.9 ng/ml, and stretched cells,47.5± 2.1 ng/ml (p< 0.05); and TIMP2 in unstretchedcells, 46.0± 2.9 ng/ml, and stretched cells, 35.8± 1.5 ng/ml(p< 0.05) (Fig. 2).

3.3. MMP/TIMP ratios in culture supernatants of LI90cells

The MMP1/TIMP1 ratio was increased in stretchedcells relative to unstretched cells: MMP1/TIMP1 ratio inunstretched cells was 3.2± 0.3 and 4.7± 0.2 (p< 0.05)in stretched cells. And the MMP1/TIMP2 ratio was in-creased in stretched cells relative to unstretched cells, too:MMP1/TIMP2 ratio in unstretched cells was 4.3± 0.8 com-p heM ig-

nificant difference between stretched and unstretched cells:MMP2/TIMP2 ratio in unstretched cells was 2.0± 0.1 com-pared to 1.9± 0.1 in stretched cells; MMP2/TIMP1 ratioin unstretched cells was 1.5± 0.2 compared to 1.5± 0.1 instretched cells (Fig. 3).

3.4. Concentrations of 1CTP, PIIIP, and hyaluronic acidin culture supernatants of LI90 cells

Concentration of type I collagen C-telopeptide, pro-collagen III propeptide, and hyaluronic acid in cul-ture supernatants showed no significant differencebetween stretched and unstretched cells (1CTP: un-stretched cells, 3.7± 0.3 ng/ml, and stretched cells,3.6± 0.5 ng/ml; PIIIP: unstretched cells, 0.32± 0.05 U/ml,and stretched cells, 0.35± 0.09 U/ml; hyaluronic acid:unstretched cells, 893.6± 148.8 ng/ml, and stretched cells,828.0± 244.1 ng/ml) (Fig. 4).

3.5. MMP1, MMP2, and TIMP1 mRNA expression ofLI90 cells

The MMP1 mRNA expression/�-actin mRNA expres-sion ratio was increased in stretched cells relative to un-stretched cells, (unstretched cells, 1.20± 0.10, and stretchedc lm be-

ared to 6.3± 0.6 (p< 0.05) in stretched cells. But tMP2/TIMP2 ratio and MMP2/TIMP1 ratio showed no s

ells, 1.48 ± 0.10; p< 0.05). But the MMP2 and TIMPRNA expression showed no significant difference

156 T. Goto et al. / Pathophysiology 11 (2004) 153–158

Fig. 2. Measurement of MMP1, MMP2, TIMP1, and TIMP2 in culture supernatants of LI90 cells. The concentration of MMP1 in culture supernatants showedan increase in stretched as opposed to unstretched LI90 cells (p< 0.05), while in contrast the concentrations of MMP2, TIMP1, and TIMP2 showed a decreasein stretched cells compared to unstretched cells (p< 0.05).

Fig. 3. Measurement of MMP/TIMP ratios in culture supernatants of LI90 cells. MMP1/TIMP1 ratio and MMP1/TIMP2 ratio showed increase in stretched cellsrelative to unstretched cells (p< 0.05). But MMP2/TIMP2 ratio and MMP2/TIMP1 ratio showed no significant difference between stretched and unstretchedcells.

T. Goto et al. / Pathophysiology 11 (2004) 153–158 157

Fig. 4. Measurement of 1CTP, PIIIP, and hyaluronic acid in culture supernatants of LI90 cells. Concentration of 1CTP, PIIIP, and hyaluronic acid in culturesupernatants showed no significant difference between stretched and unstretched cells.

Fig. 5. MMP1, MMP2, and TIMP1 mRNA expression of LI90 cells. The MMP1 mRNA expression/�-actin mRNA expression ratio was increased in stretchedcells relative to unstretched cells (p< 0.05). But MMP2, and TIMP1 mRNA expression showed no significant difference between stretched and unstretchedcells.

tween stretched and unstretched cells (MMP2 mRNA/�-actinmRNA; unstretched cells, 1.05± 0.21, and stretched cells,0.91± 0.31; TIMP1 mRNA/�-actin mRNA: unstretchedcells, 1.15± 0.27, and stretched cells, 0.85± 0.29) (Fig. 5).

4. Discussion

Portal hypertension, which results from increased intra-hepatic resistance to blood flow after liver injury, is a ma-jor complication of scarring in the liver. Von Leeuwen etal. reported Disse’s space was significantly increased in pa-tients with chronic active hepatitis (CAH), CAH in transi-tion to cirrhosis, and cirrhosis compared with near normalsubjects[15]. In the hyperdynamic portal circulation and in-crease Disse’s space, HSC may be exposed to mechanicalstress caused by congested blood flow distending sinusoids.As a hyperdynamic circulation acts upon intrahepatic ves-sels, this hemodynamics may change the function of HSC

located within the subendothelial space of Disse along si-nusoids. Sakata et al. reported that mechanical stretch in-duces TGF-beta synthesis in LI90, human hepatic stellate celllines[16]. It has been shown that in the early stages of hep-atic fibrosis there is a remodeling of basement membranesand increased expression of MMPs[17]. This mechanicalstretch may induce various extracellular matrix constituents,MMP, and TIMP, too. Previous studies have revealed a rela-tionship between mechanical stretch and matrix metallopro-teinases and tissue inhibitors of metalloproteinases in manykinds of cultured cells. For example, Grote et al. reportedmechanical stretch enhanced mRNA expression of MMP2[18], Honda et al. reported mechanical stretch increased themRNA level of MMP1, MMP3, and MMP9 in rabbit chon-drocytes[19], O’Callaghan et al. reported mechanical stretchincreased MMP2 activity in human vascular smooth musclecells [20], and Park et al. reported mechanical stretch in-creased the mRNA level of MMP1 in rat bladder smoothmuscle cells[21]. In these cultured cells, mechanical stretch

158 T. Goto et al. / Pathophysiology 11 (2004) 153–158

induced MMP and these results suggest mechanical stretchmay promote degradation of extracellular matrix.

In this study, we investigated the relationship betweenmechanical stretch and production of MMP and TIMP inHSC. Mechanical stretch increased MMP1 production anddecreased TIMP1 and TIMP2 production, resulting in anincreased MMP1/TIMP1 ratio and MMP1/TIMP2 ratio.MMP1 mRNA expression was also increased in stretchedcells. Because MMP1 attacks collagen type I, II, III, VII,and X and MMP1 also degrades casein and cartilage proteo-glycan, mechanical stretch may induce potential of degrad-ing extracellular matrix in liver disease. But concentration of1CTP, PIIIP, and hyaluronic acid, which were used as markerof liver fibrosis, showed no significant difference relative tounstretched cells in this study. Under culture conditions withmechanical stretch, the Flexercell apparatus induced cyclicmechanical elongation by forces transmitted within the planeof the silastic membrane and the attached cells. For 24 h du-ration of mechanical stretch like that in this study, HSC maychange production of MMP and TIMP but not yet increaseproduction of the extracellular matrix.

Our findings suggested that in the early phase of portal hy-pertension, HSC increase production of MMP1 and decreaseproduction of TIMP1 and TIMP2, and these changes mayreflect HSC activated by mechanical stretch.

R

titialinu-Rev.

cellacute

ep-

od.uid685–

[6] J. Vorobioff, J.E. Bredtfeldt, R.J. Groszmann, Hyperdynamic circu-lation in portal hypertensive rat model-a primary maintenance ofchronic portal hypertension, Am. J. Physiol. 244 (1983) 652–657.

[7] H. Nagase, A.J. Barrett, J.F. Woessner Jr., Nomenclature and glossaryof the matrix metalloproteinases, Matrix 1 (1992) 421–424.

[8] H. Nagase, Activation mechanisms of matrix metalloproteinases.,Biol. Chem. 378 (1997) 151–160.

[9] J.P. Iredale, R.C. Benyon, M.J.P. Arthur, et al., Tissue inhibitor ofmetalloproteinase-1 messenger RNA expression is enhanced relativeto interstitial collagenase messenger RNA in experimental liver in-jury and fibrosis, Hepatology 24 (1996) 176–184.

[10] H. Herbst, T. Wege, S. Milani, et al., Tissue inhibitor ofmetalloproteinase-1 and -2 RNA expression in rat and human liverfibrosis, Am. J. Pathol. 150 (1997) 1647–1659.

[11] I. Okazaki, K. Maruyama, Collagenase activity in experimental hep-atic fibrosis, Nature 252 (1974) 49–50.

[12] K. Murakami, T. Abe, M. Miyazawa, et al., Establishment of a newhumancell line, LI90, exhibiting characteristics of hepatic Ito (fat-storing) cells, Lab. Invest. 72 (1995) 731–739.

[13] R. Lichtinghagen, K. Breitenstein, B. Arndt, T. Kuhbacher, KH.W.Boker, Comparison of matrix metalloproteinase expression in normaland cirrhotic human liver, Virchows Arch. 432 (1998) 153–158.

[14] H. Nakano, Y. Okamoto, Effect of acetaldehyde on the gene expres-sion of type IV collagenases and those inhibitors in cultured humanIto-cells, Acta hepatologica japonica 39 (1998) 240–245.

[15] D.J. Von Leeuwen, S.C. Howe, P.J. Scheuer, S. Sherlock, Por-tal hypertension in chronic hepatitis: relationship to morphologicalchanges, Gut 31 (1990) 339–343.

[16] R. Sakata, T. Ueno, T. Nakamura, H. Ueno, M. Sata, Mechanicalstretch induces TGF-beta synthesis in hepatic stellate cells, Eur. J.

[ leallo-t. 24

[ nicalmatrixtive

[ gni-red

[ cel-role

[ etch-Am.

eferences

[1] K. Wake, Perisinusoidal stellate cells (Fat-storing cells, interscells, lipocyte), their related structure in and around the liver ssoids, and vitamin A-storing cells in extrahepatic organs, Int.Cytol 66 (1980) 303–353.

[2] K. Hirata, I. Ogata, Y. Ohta, K. Fujiwara, Hepatic sinusoidaldestruction in the development of intravascular coagulation inliver failure of rats, J. Pathol. 158 (1989) 157–165.

[3] A.O. Whipple, The problem of hypertension in relation to the hatosplenopathies, Ann. Surg. 122 (1945) 449–475.

[4] A.H. Moreno, A.R. Burchell, L.M. Rousselet, et al., Portal bloflow in cirrhosis of the liver, J. Clin. Invest. 46 (1967) 436–445

[5] C.L. Witte, M.H. Witte, Splanchnic circulatory and tissue fldynamics in portal hypertension, Fed. Proc. 42 (1983) 11689.

Clin. Invest. 34 (2004) 129–136.17] Y.L. Li, M. Sato, N. Kojima, M. Miura, H. Senoo, Regulatory ro

of extracellular matrix components in expression of matrix metproteinases in cultured hepatic stellate cells, Cell Struct. Func(1999) 255–261.

18] Karsten Grote, Inna Flach, Maren Luchtefeld, et al., Mechastretch enhances mRNA expression and proenzyme release ofmetalloproteinase-2(MMP-2) via NAD(P)H oxidase-derived reacoxygen species, Circ. Res. 92 (2003) e80–e86.

19] K. Honda, S. Ohno, K. Tanimoto, et al., The effects of high matude cyclic tensile load on cartilage matrix metabolism in cultuchondrocytes, Eueur. J. Cell Biol. 79 (2000) 601–609.

20] C.J. O’callaghan, B. Williams, Mechanical strain-induced extralular matrix production by human vascular smooth muscle cell:of TGF-betal, Hypertension 36 (2000) 319–324.

21] J.M. Park, R.M. Adam, C.A. Peters, et al., AP-1 mediates strinduced expression of HB-EGF in bladder smooth muscle cells,J. Physiol. Cell Physiol. 277 (1999) C294–C301.