- Email: [email protected]
Evaluation of matrix metalloproteinases and their endogenous tissue inhibitors in biliary atresia–associated liver fibrosis
Journal of Pediatric Surgery (2005) 40, 1568 – 1573
www.elsevier.com/locate/jpedsurg
Evaluation of matrix metalloproteinases and their endogenous tissue inhibitors in biliary atresia–associated liver fibrosis Chih-Sung Hsieha, Jiin-Haur Chuanga,d,*, Chao-Cheng Huangb, Ming-Huei Choud, Chia-Lin Wud, Shin-Yi Leea, Chao-Long Chenc a
Department of Pediatric Surgery, Chang Gung Memorial Hospital, Kaohsiung Hsien, 833, Taiwan Department of Pathology, Chang Gung Memorial Hospital, Kaohsiung Hsien, 833, Taiwan c Department of Surgery, Chang Gung Memorial Hospital, Kaohsiung Hsien, 833, Taiwan d The Graduate Institute of Clinical Medicine, Chang Gung University, Kaohsiung Hsien, 833, Taiwan b
Index words: Biliary atresia; Liver fibrosis DNA microarray; Real time QRT-PCR; Matrix metalloproteinase; Tissue inhibitor of metalloproteinase
Abstract Background/Purpose: Matrix metalloproteinases (MMPs) and their endogenous tissue inhibitors (TIMPs) are major proteases responsible for remodeling the liver tissue, but their roles in biliary atresia (BA)–associated liver fibrosis are not clear. Methods: A DNA microarray containing complementary DNA clones of 10 MMPs and 4 TIMPs was used to compare the expression profiles of the liver cytokines among 3 patients with BA at the time of Kasai procedure (KP) with 3 at the time of liver transplantation (LT). Liver samples from 2 children without liver fibrosis were used as normal controls. Those genes that were differentially expressed by more than 2-fold between groups were further quantified with real time quantitative reverse transcription–polymerase chain reaction (QRT-PCR) and validated with gel electrophoresis. Results: In normal human liver, messenger RNAs (mRNAs) of TIMP-1, -2, and -3, but not of TIMP-4 and none of the 10 MMPs studied, were expressed in DNA microarray. With progression of liver fibrosis, only mRNA of MMP-7, but not other MMPs, was induced to express at a significantly higher level in the array. Despite its low level of expression, MMP-9 mRNA was significantly upregulated in KP but downregulated in LT, whereas MMP-2, which was not showed in the array, was significantly upregulated in LT than in KP and control in real time QRT-PCR. There was a more than 2-fold increase in TIMP-1 and TIMP-2 mRNA expression in LT over control in the array, which was confirmed in subsequent real time QRT-PCR. The expression of TIMP-3 mRNA was significantly downregulated in KP than in control. Conclusions: This study verified differential expression of MMPs and TIMPs in different stages of BA, with emphasis on the role of TIMP-1, -2, and -3 as well as MMP-2, -7, and -9 transcripts in remodeling of liver tissue during the progress of BA-associated liver fibrosis. D 2005 Published by Elsevier Inc.
T Corresponding author. Tel.: +886 7 731723#8715 or 8892; fax: +886 7 7311696. E-mail address: [email protected] (J.-H. Chuang). 0022-3468/$ – see front matter D 2005 Published by Elsevier Inc. doi:10.1016/j.jpedsurg.2005.06.028
Evaluation of MMPs and their endogenous tissue inhibitors in BA-associated liver fibrosis Biliary atresia (BA) is the most common disease that leads to liver cirrhosis in children. Biliary atresia is still the leading cause of liver transplantation (LT) for children [1-5]. The fibrogenic mechanism in BA may include activation of hepatic stellate cells and increase of extracellular matrix (ECM)—a complex process involving many cytokines of which a few have been identified [6-13]. The ECM holds cells together and maintains 3-dimensional structure of the organ, including the liver. Normal liver contains a relatively small amount of ECM component. Liver fibrosis is the result of extensive tissue remodeling with a net increase of ECM. The key enzymes responsible for degradation and deposition of all the protein components of ECM and basement membrane are matrix metalloproteinases (MMPs) and their endogenous inhibitors. DNA microarray is a powerful tool that enables gene expression analysis in the human liver tissues on a genomewide scale. Two recent DNA microarray studies of BA arrived at interesting but different conclusions on the pathogenesis of BA [14,15]. Matrix metalloproteinase 7 was the only MMP upregulated in fibrotic liver in both studies and MMP-2 was downregulated in BA in one of the studies. Neither of them mentioned the status of tissue inhibitors of metalloproteinases (TIMPs) in their study. The roles of MMPs and TIMPs in BA-associated liver fibrosis are still not well elucidated. In this study, we used a human cytokine gene expression array consisting of 10 members of MMPs and 4 TIMPs to explore the problems.
1. Materials and methods 1.1. Patients and samples For DNA microarray, biopsy specimens were obtained from the liver of 3 patients during Kasai procedure (KP) for BA. There were 2 females and 1 male, ranging in age from 1.3 to 1.7 months. Another 3 liver samples were obtained from patients with BA at the time of LT, with 1 female and 2 males, ranging in age from 10 months to 15 years. Two control liver samples were obtained from 1 female and 1 male patient without liver fibrosis at the age of 2 and 5 months, respectively. One was obtained from a patient who underwent repair of the hiatal hernia; a piece of liver attached to the hernia ring was excised to facilitate closure. Another patient with neonatal hepatitis presented with jaundice; the liver tissue was obtained for equivocal diagnosis of BA. The liver tissues in both control patients were devoid of liver fibrosis microscopically, despite abnormal liver function in the latter case. The liver samples were immediately immersed in liquid nitrogen after snatch and were stored at 808C. Histologic grading of liver fibrosis was performed on all the 8 patients with Masson trichrome staining of the liver specimens, as described by Ferrell [16]. This study was approved by the Medical Ethics and the Human Clinical Trial Committee at Chang Gung Memorial Hospital.
1.2. RNA isolation and generation of complementary DNA
1569 33
P-labeled
Total RNA was extracted from frozen liver tissue using a commercially available reagent REZOL C&T (Protech technology, Taipei, Taiwan), which was purified by phenol-chloroform extraction and quantified by spectrophotometry with absorbance at 260 nm. Fifteen to 20 lg of total RNA was annealed with 1 lg of oligo (dT) [15] primers (Promega, Madison, Wis) and then incubated with 100 lCi of a-33P-dCTP, 800 U of M-MLV reverse transcriptase (Promega), 80 U of RNasin, and dNTP mixture (0.5 mmol/L each of dATP, dGTP, and dTTP and 2.5 lmol/L of dCTP) at 428C for 3 hours to generate 33 P-labeled complementary DNA (cDNA) [17]. The unincorporated nucleotides were removed by using a Sephadex G-25 gel-filtration spin column (Amersham Pharmacia Biotech, Piscataway, NJ).
1.3. Hybridization and array analysis After prehybridization with 3 mL of the hybridization solution (Sigma-Genosys, Woodlands, Tex) at 658C for 1 hour in a hybridization bag, the denatured 33P-labeled cDNA probes were incubated at 658C overnight with the Human Cytokine Expression Array GA001 (R & D Systems, Minneapolis, Minn). The array contained 375 cDNA sequences, including 10 MMPs from MMP-1, -3, -7, -8, -9, -10, -12, -13, -14 to -15 and 4 TIMPs from TIMP-1, -2, -3 to -4, with duplicate spots for each gene on a nylon membrane. The membrane was washed for 20 minutes 3 times with solution I (0.5 SSPE with 1% wt/vol SDS) and then with solution II (0.1 SSPE with 1% wt/vol SDS) at 658C. The membrane was air dried, autoradiographed onto a Kodak Low Storage Phosphor Screen, and analyzed with a Phosphoimager (Storm 840, Molecular Dynamics, Sunnyvale, CA). We used the ImageQuant software (Molecular Dynamics) to analyze the intensity of hybridization on each spot of the array. The intensity of each gene was an average intensity of a pair of duplicate spots. The value was further normalized by the mean value of 9 different housekeeping genes. The normalized signals were compared between groups.
1.4. Real time quantitative reverse transcription– polymerase chain reaction Quantitative reverse transcription–polymerase chain reaction (QRT-PCR) was performed with ABI 7700 Sequence Detection System (TaqMan, Perkin-Elmer Applied Biosystems) to confirm the findings from the array. Polymerase chain reaction was performed in 50 lL SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) containing 10 lmol/L forward primers and reverse primers and approximately 30 ng cDNA. The sequence of the primers of MMP-2, -7, -9, and -13 and TIMP-1, -2, -3, and -4 was designed using the Primer Express Software (Perkin-Elmer Life Science) (Table 1).
1570 Table 1
C.-S. Hsieh et al. Primer sequence of MMPs and TIMPs
Cytokines MMPs MMP-2 MMP-7 MMP-9 MMP-13 TIMPs TIMP-1 TIMP-2 TIMP-3 TIMP-4 Human actin
Forward primers
Reverse primers
5V-CAACTACGATGATGACCGCAA-3V 5V-TACAGTGGGAACAGGCTCAGG-3V 5V-GACCTGGGCAGATTCCAAAC-3V 5V-CTTCACGATGGCATTGCTGA-3V
5V-GTGTAAATGGGTGCCATCAGG-3V 5V-GGCACTCCACATCTGGGCT-3V 5V-CACGCGCAGTGAAGGTGAGC-3V 5V-AACTCATGCGCAGCAACAAG-3V
5V-CCTTCTGCAATTCCGACCTC-3V 5V-TCATTGCAGGAAAGGCCG-3V 5V-TCGGCACGCTGGTCTACAC-3V 5V-CACTCGGCACTTGTGATTCG-3V aˆ-5V-TCACCCACACTGTGCCCATCTACG-3V
5V-ATGTGCAAGAGTCCATCCTGC-3V 5V-GATGCAGGCGAAGAACTTGG-3V 5V-AGCTGGTCCCACCTCTCCA-3V 5V-GCTTCTGGCTGTTGGCTTCT-3V 5V-CAGCGGAACCGCTCATTGCCAATGG-3V
Amplification and detection were performed with the following profile [17]: 1 cycle of 958C for 10 minutes and 40 cycles of 958C for 15 seconds, 608C for 30 seconds, and 728C for 15 seconds. After amplification was completed, a final melting curve was performed with 2 minutes of denaturation at 958C, cooling to 608C, and then heating slowly until 958C (20 minutes) according to the dissociation protocol of the ABI 7700 instrument. Real time fluorescence measurement was read and a threshold cycle (CT) value for each sample was calculated by determining the point at which the fluorescence exceeded a threshold limit, that is, 10 times the standard deviation of the baseline. To confirm the specificity of the PCR reaction, PCR products were electrophoresed on a 1.5% agarose gel.
1.5. Statistical analysis For statistical analysis of the expression data, the relative value of each gene in a patient from the array was calculated and fold induction or reduction in expression of gene-specific messenger RNAs (mRNAs) between
Table 2 Cytokines MMPs MMP-2 MMP-7 MMP-9 TIMPs TIMP-1 TIMP-2 TIMP-3
samples was obtained. An analysis of variance with post hoc Bonferroni was then used to obtain the genes that showed significant difference between groups. Data in real time QRT-PCR were expressed as geometric means after back-transforming the logarithmically transformed variables and thus normally distributed. The relative expression of various genes, 2-DDCT, as measured using real time QRTPCR between groups were compared with unpaired t tests on log-transformed values. A P value of less than .05 was regarded as statistically significant.
2. Results Histologic sections of the liver studied with Masson trichrome stain revealed no fibrosis in the 2 control patients, grade II or III liver fibrosis in the 3 KP patients, and grade IV fibrosis in all 3 LT patients. The hybridization signal intensities of 10 MMPs and 4 TIMPs in cDNA microarray were shown in Fig. 1. The expression of MMPs was generally weak in the control.
Fold difference in microarray and in real time QRT-PCR (QRT-PCR ratio) Fold difference in microarray
QRT-PCR ratio
KP/CO
LT/CO
KP/CO
LT/CO
NA 3.5 F 2.0 1.6 F 0.8
NA 4.2 F 2.5 2.1 F 0.5
2.0 F 1.2 39.8 F 18.8 1.9 F 1.4
5.3 F 1.3T 91.9 F 32.0T 1.4 F 0.9
1.0 F 0.3 1.1 F 0.5 0.7 F 0.2
3.5 F 2.0 5.2 F 3.2 2.1 F 1.1
1.9 F 0.6 2.1 F 1.4 0.3 F 0.1TT
2.4 F 0.7 5.3 F 1.7T 0.7 F 0.2T
For statistical analysis of the expression data, the relative value of each gene in a patient obtained from the array is calculated and fold induction or reduction in expression of gene-specific mRNAs between samples is obtained. The QRT-PCR ratio is the relative level of transcripts measured by real time QRT-PCR in one group over that in the other. CO indicates control. T P b .05 LT vs CO. TT P b .05 KP vs CO.
Evaluation of MMPs and their endogenous tissue inhibitors in BA-associated liver fibrosis
1571
Fig. 1 Part of the scanning picture is shown after hybridization of liver tissue cDNAs with the Human Cytokine Expression Array GA001 (R & D Systems). Genomic DNA (GEN) control spots are located at the right upper and lower corner of the array, whereas 9 housekeeping genes (HG) are starting from the second cell on the same column as GEN, followed by 6 negative controls (NC). There are 10 MMPs starting from MMP-1 in the first column from the left-hand side to MMP-15 in the top row of the second column. The strong intensity of TIMP-1, -2, and -3 is located with arrow in the second column, whereas TIMP-4 that locates immediately below TIMP-3 is too weak to be visualized. The most prominent finding is faint MMP-7 expression in the control (arrowhead), which is followed by significant increase in signal intensity in KP. The signal of MMP-7 is most prominent in LT. All the other 9 MMPs continued to be weak despite increasing liver fibrosis from control to LT.
With increasing liver fibrosis, only the intensity of MMP-7 became stronger; its expression in LT was 2- and 5-fold higher than that in KP and control, respectively. The expression of the other 9 MMPs was weak despite progression of liver fibrosis. Nonetheless, there was an increase in expression of MMP-9 and MMP-13 in LT over that in control and KP in the array. The signal intensity of TIMP-1 was the strongest of the 4 members of TIMPs in the array, which was followed by TIMP-3 and TIMP-2 (Fig. 1). The signal intensity of TIMP-4 was too weak to be detected in the array and the difference in signal intensity was not significant between groups. The change of signal intensity of both TIMP-1 and TIMP-2 paralleled the progression of liver fibrosis. There were 2.5- and 3.5-fold increases in signal intensity in TIMP-1 in LT compared with KP and control, respectively. There were 4.7- and 5.2-fold increases in signal intensity in TIMP-2 in LT compared with KP and control, respectively. The signal intensity of TIMP-3 was strongest in control and weakest in KP, resulting in inverse relationship of the expression of TIMP-3 with the active fibrogenesis in BA.
By using real time QRT-PCR, MMP-7 was the only gene among the 10 members of MMPs that demonstrated sturdy and significant increase of transcripts with the progression of liver fibrosis. The QRT-PCR ratio, that is, the relative level of transcripts measured by real time QRT-PCR in one group over that in the other, was 1.7 in LT over KP, up to 39.8 in KP over control, and up to 91.9 in LT over control, respectively. The difference was highly significant ( P b.001). The MMP-9 mRNA expression was higher in KP than in LT and control in real time QRT-PCR. The expression of MMP-13 mRNA in the liver of the study groups was too low to be detected using real time QRT-PCR. Although MMP-2 was not included in this cDNA microarray, we did real time QRT-PCR to quantify MMP-2 mRNA. The results showed a significant upregulation of MMP-2 in LT than in KP and control (Table 2). There was a 2.4-fold increase in TIMP-1 in LT compared with the control. There was a 5.3-fold increase in TIMP-2 in LT compared with the control. The differences were significant in both. The expression of TIMP-3 was highest in control, followed by LT and was lowest in KP, resulting in significantly higher QRT-PCR ratio in LT over KP than in KP over control (Table 2).
1572
Fig. 2 Gel electrophoresis of PCR products after 30 cycles of amplification confirms the findings of QRT-PCR. Two samples from each group are randomly chosen without selection. The intensity of the TIMP-1, TIMP-2, MMP-2, and MMP-7 bands of representative patients in LT group (lane 5-6) is higher than in KP (lane 3-4), and the latter is again higher than in control (lane 1-2). The intensity of MMP-9 bands is highest in KP, followed by LT and control. The intensity of TIMP-3 bands is, on the contrary, lowest in KP, followed by LT and control. The expression of housekeeping gene aˆ-actin is constant between groups.
Gel electrophoresis confirmed the findings of real time QRT-PCR. For genes included in this study, the intensity of the bands correlated well with the QRT-PCR ratio (Fig. 2).
3. Discussion Our findings revealed that TIMP-4 and all 10 MMPs studied were not constitutively expressed in the normal liver. Conversely, there was constitutive expression of TIMP-1, -2, and -3. These findings are consistent with previous reports [18,19]. With increasing liver fibrosis, DNA microarray study demonstrated an increasing signal intensity in MMP-7 only, but not in the 9 other MMPs. Upregulation of MMP-7 was also demonstrated in the other 2 studies of BA using gene chips or cDNA microarray, denoting some role of this unique protease in the progress of liver fibrosis/cirrhosis associated with BA [14,15]. Matrix metalloproteinase 2 was found to be downregulated in BA over normal liver in the study of Chen et al [15], but the difference was less than 2-fold. Although MMP-2 was not included in this cDNA microarray study of BA, we still
C.-S. Hsieh et al. performed real time QRT-PCR to quantify this frequently cited protease. Our findings were consistent with upregulation of MMP-2 mRNA in one study of fibrotic human liver in BA, but contrary to the study of Chen et al [15,20]. We suspect that MMP-2, like other members of MMPs, is expressed in very low level in the liver despite increasing liver fibrosis, which may result in false downregulation when using cDNA microarray study of gene expression without further confirmation with either Northern blotting or QRTPCR. A similar problem happens to MMP-13. Matrix metalloproteinase 13 was found upregulated more than 2-fold in the array, but subsequent real time QRT-PCR failed to detect the MMP-13 transcript because of low copy number. It is known that genes expressed at low or near background levels may result in exceptionally high false-positive rate on the ratios and could be ignored in expression analysis [21]. Significant upregulation of hepatic MMP-9 mRNA was found in the early stage of BA in which the patient received KP in this study. The expression decreases in the late stage of BA, but the level was still higher than that in the control. Similar findings are found in 2 animal studies, which show rapid upregulation of MMP-9 mRNA or the proteolytic activity after experimentally induced liver fibrosis in mice or biliary fibrosis in rats [22,23]. However, a study of fibroproliferative process in the liver during chronic hepatitis C fails to show significant change of MMP-9 expression between patients with or without liver fibrosis [24]. The findings of significant upregulation of TIMP-1 and TIMP-2 in fibrotic liver in BA are consistent with the study of Benyon et al [20]. A recent report from Korea using differential expressed sequence tag screening identifies TIMP-1 overexpressed in biliary cirrhosis [25]. Our study using human cytokine cDNA array demonstrated a greater increase of TIMP-2 than TIMP-1 with increasing fibrosis in BA. The findings were further confirmed with QRT-PCR. For both TIMP-1 and TIMP-2, the fold difference in DNA microarray was correlated with the QRT-PCR ratio. The findings are consistent with in situ hybridization study of TIMP-1 and TIMP-2 expression in rat and human liver fibrosis [26], but not with the report of Kobayashi et al [27]. In the latter study, Kobayashi et al [27] used 1-step sandwich enzyme immunoassays to measure serum MMP-1 and -2 as well as TIMP-1 and -2 in 36 postoperative patients with BA and found no correlation of TIMP levels with the progress of fibrosis between groups. The difference may be explained as that MMPs and TIMPs in the blood samples are contaminated by other sources or mRNA of these cytokines in the liver fails to reflect the protein contents in situ or in the remote site such as blood. In this study, the finding of TIMP-3 expression in the liver is contrary to previous notion that this protease is not present in the liver [28]. We also find that the mRNA levels of TIMP-3 are highest in control and lowest in the early stage of BA. The findings imply significant downregulation of TIMP-3 during the stage of active fibrogenesis in BA. No similar findings have been found in the literature. Because
Evaluation of MMPs and their endogenous tissue inhibitors in BA-associated liver fibrosis the protein of TIMP-3 displays low sequence similarity to human TIMP-1 and TIMP-2 [28], the unique finding of downregulation of TIMP-3 mRNA expression in the early stage of BA implies a different role for human TIMP-3 in human liver tissue remodeling. Although altered balance between MMPs and TIMPs is thought to be important in liver fibrosis, controversy exists as to their role in fibrogenesis. These discordant findings denote a complex change in the local MMP/TIMP balance that underlies liver fibrosis caused by different liver diseases. There has been a growing focus on the role of MMPs in liver disease [29]. However, the complex findings in this study—some consistent with and some contrary to those present in the literature—suggest that the study of this family of zinc- and calcium-dependent proteases and their inhibitors is still far from completion.
Acknowledgments This study was supported by the National Science Council, Taiwan grant NSC91-2314-B-182-065 and the Chang Gung Memorial grants CMRPG8005 and CMRPG8067. We thank the Liver Transplant Team, Chang Gung Memorial Hospital–Kaohsiung, for providing the samples used in this study.
References [1] Karrer FM, Price MR, Bensard DD, et al. Long-term results with the Kasai operation for biliary atresia. Arch Surg 1996;131:493 - 6. [2] Nio M, Ohi R, Shimaoka S, et al. The outcome of surgery for biliary atresia and the current status of long-term survivors. Tohoku J Exp Med 1997;181:235 - 44. [3] Chen CL, Chen YS, de Villa VH, et al. Minimal blood loss living donor hepatectomy. Transplantation 2000;69:2580 - 6. [4] Jain A, Reyes J, Kashyap R, et al. Long-term survival after liver transplantation in 4,000 consecutive patients at a single center. Ann Surg 2000;232:490 - 500. [5] Miller CM, Gondolesi GE, Florman S, et al. One hundred nine living donor liver transplants in adults and children: a single-center experience. Ann Surg 2001;234:301 - 11. [6] Friedman SL, Maher JJ, Bissell DM. Mechanisms and therapy of hepatic fibrosis: report of the AASLD single topic basic research conference. Hepatology 2000;32:1403 - 8. [7] Benyon RC, Arthur MJP. Extracellular matrix degradation and the role of hepatic stellate cells. Semin Liver Dis 2001;21:373 - 84. [8] Faiz Kabir Uddin Ahmed A, Ohtani H, Nio M, et al. In situ expression of fibrogenic growth factors and their receptors in biliary atresia: comparison between early and late stages. J Pathol 2000;192:73 - 80. [9] Ahmed AF, Nio M, Ohtani H, et al. Clinical significance of c-kit expression in biliary atresia. Pediatr Surg Int 2001;17:601 - 3. [10] Liu C, Chiu JH, Chin TW, et al. Expression of aminopeptidase N in bile canaliculi: a predictor of clinical outcome in biliary atresia. J Surg Res 2001;10:76 - 83.
1573
[11] Davenport M, Gonde C, Redkar R, et al. Immunohistochemistry of the liver and biliary tree in extrahepatic biliary atresia. J Pediatr Surg 2001;36:1017 - 25. [12] Rosensweig JN, Omori M, Page K, et al. Transforming growth factor– beta 1 in plasma and liver of children with liver disease. Pediatr Res 1998;44:402 - 9. [13] Lee SY, Chuang JH, Huang CC, et al. Identification of transforming growth factors actively transcribed during the progress of liver fibrosis in biliary atresia. J Pediatr Surg 2004;39:702 - 8. [14] Bezerra JA, Tiao G, Ryckman FC, et al. Genetic induction of proinflammatory immunity in children with biliary atresia. Lancet 2002;360:1653 - 9. [15] Chen L, Goryachev A, Sun J, et al. Altered expression of genes involved in hepatic morphogenesis and fibrogenesis are identified by cDNA microarray analysis in biliary atresia. Hepatology 2003;38: 567 - 76. [16] Ferrell L. Liver pathology: cirrhosis, hepatitis, and primary liver tumors. Update and diagnostic problems. Mod Pathol 2000;13: 679 - 704. [17] Huang CC, Chuang JH, Huang LLH, et al. The human Delta-like 1 homologue is implicated in the progression of liver fibrosis in biliary atresia. J Pathol 2004;202:172 - 9. [18] Cursio R, Mari B, Louis K, et al. Rat liver injury after normothermic ischemia is prevented by a phosphinic matrix metalloproteinase inhibitor. FASEB J 2002;16:93 - 5. [19] Boker KH, Pehle B, Steinmetz C, et al. Tissue inhibitors of metalloproteinases in liver and serum/plasma in chronic active hepatitis C and HCV-induced cirrhosis. Hepatogastroenterology 2000;47:812 - 9. [20] Benyon RC, Iredale JP, Goddard S, et al. Expression of tissue inhibitor of metalloproteinases 1 and 2 is increased in fibrotic human liver. Gastroenterology 1996;110:821 - 31. [21] Baldi P, Long AD. A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes. Bioinformatics 2001;17:509 - 19. [22] Vaillant B, Chiaramonte MG, Cheever AW, et al. Regulation of hepatic fibrosis and extracellular matrix genes by the th response: new insight into the role of tissue inhibitors of matrix metalloproteinases. J Immunol 2001;167:7017 - 26. [23] Kossakowska AE, Edwards DR, Lee SS, et al. Altered balance between matrix metalloproteinases and their inhibitors in experimental biliary fibrosis. Am J Pathol 1998;153:1895 - 902. [24] Lichtinghagen R, Michels D, Haberkorn CI, et al. Matrix metalloproteinase (MMP)-2, MMP-7, and tissue inhibitor of metalloproteinase-1 are closely related to the fibroproliferative process in the liver during chronic hepatitis C. J Hepatol 2001;34:239 - 47. [25] Choe BH, Kim KM, Kwon S, et al. The pattern of differentially expressed genes in biliary atresia. J Korean Med Sci 2003;18: 392 - 6. [26] Herbst H, Wege T, Milani S, et al. Tissue inhibitor of metalloproteinase-1 and -2 RNA expression in rat and human liver fibrosis. Am J Pathol 1997;150:1647 - 59. [27] Kobayashi H, Li ZX, Yamataka A, et al. Clinical evaluation of serum levels of matrix metalloproteinase and tissue inhibitors of metalloproteinase as predictors of progressive fibrosis in postoperative biliary atresia patients. J Pediatr Surg 2002;37:1030 - 3. [28] Uria JA, Ferrando AA, Velasco G, et al. Structure and expression in breast tumors of human TIMP-3, a new member of the metalloproteinase inhibitor family. Cancer Res 1994;54:2091 - 4. [29] Bansal MB, Friedman SL. The enlarging role of matrix metalloproteinases in liver injury—beyond scar degradation. J Hepatol 2002; 37:293 - 4.
www.elsevier.com/locate/jpedsurg
Evaluation of matrix metalloproteinases and their endogenous tissue inhibitors in biliary atresia–associated liver fibrosis Chih-Sung Hsieha, Jiin-Haur Chuanga,d,*, Chao-Cheng Huangb, Ming-Huei Choud, Chia-Lin Wud, Shin-Yi Leea, Chao-Long Chenc a
Department of Pediatric Surgery, Chang Gung Memorial Hospital, Kaohsiung Hsien, 833, Taiwan Department of Pathology, Chang Gung Memorial Hospital, Kaohsiung Hsien, 833, Taiwan c Department of Surgery, Chang Gung Memorial Hospital, Kaohsiung Hsien, 833, Taiwan d The Graduate Institute of Clinical Medicine, Chang Gung University, Kaohsiung Hsien, 833, Taiwan b
Index words: Biliary atresia; Liver fibrosis DNA microarray; Real time QRT-PCR; Matrix metalloproteinase; Tissue inhibitor of metalloproteinase
Abstract Background/Purpose: Matrix metalloproteinases (MMPs) and their endogenous tissue inhibitors (TIMPs) are major proteases responsible for remodeling the liver tissue, but their roles in biliary atresia (BA)–associated liver fibrosis are not clear. Methods: A DNA microarray containing complementary DNA clones of 10 MMPs and 4 TIMPs was used to compare the expression profiles of the liver cytokines among 3 patients with BA at the time of Kasai procedure (KP) with 3 at the time of liver transplantation (LT). Liver samples from 2 children without liver fibrosis were used as normal controls. Those genes that were differentially expressed by more than 2-fold between groups were further quantified with real time quantitative reverse transcription–polymerase chain reaction (QRT-PCR) and validated with gel electrophoresis. Results: In normal human liver, messenger RNAs (mRNAs) of TIMP-1, -2, and -3, but not of TIMP-4 and none of the 10 MMPs studied, were expressed in DNA microarray. With progression of liver fibrosis, only mRNA of MMP-7, but not other MMPs, was induced to express at a significantly higher level in the array. Despite its low level of expression, MMP-9 mRNA was significantly upregulated in KP but downregulated in LT, whereas MMP-2, which was not showed in the array, was significantly upregulated in LT than in KP and control in real time QRT-PCR. There was a more than 2-fold increase in TIMP-1 and TIMP-2 mRNA expression in LT over control in the array, which was confirmed in subsequent real time QRT-PCR. The expression of TIMP-3 mRNA was significantly downregulated in KP than in control. Conclusions: This study verified differential expression of MMPs and TIMPs in different stages of BA, with emphasis on the role of TIMP-1, -2, and -3 as well as MMP-2, -7, and -9 transcripts in remodeling of liver tissue during the progress of BA-associated liver fibrosis. D 2005 Published by Elsevier Inc.
T Corresponding author. Tel.: +886 7 731723#8715 or 8892; fax: +886 7 7311696. E-mail address: [email protected] (J.-H. Chuang). 0022-3468/$ – see front matter D 2005 Published by Elsevier Inc. doi:10.1016/j.jpedsurg.2005.06.028
Evaluation of MMPs and their endogenous tissue inhibitors in BA-associated liver fibrosis Biliary atresia (BA) is the most common disease that leads to liver cirrhosis in children. Biliary atresia is still the leading cause of liver transplantation (LT) for children [1-5]. The fibrogenic mechanism in BA may include activation of hepatic stellate cells and increase of extracellular matrix (ECM)—a complex process involving many cytokines of which a few have been identified [6-13]. The ECM holds cells together and maintains 3-dimensional structure of the organ, including the liver. Normal liver contains a relatively small amount of ECM component. Liver fibrosis is the result of extensive tissue remodeling with a net increase of ECM. The key enzymes responsible for degradation and deposition of all the protein components of ECM and basement membrane are matrix metalloproteinases (MMPs) and their endogenous inhibitors. DNA microarray is a powerful tool that enables gene expression analysis in the human liver tissues on a genomewide scale. Two recent DNA microarray studies of BA arrived at interesting but different conclusions on the pathogenesis of BA [14,15]. Matrix metalloproteinase 7 was the only MMP upregulated in fibrotic liver in both studies and MMP-2 was downregulated in BA in one of the studies. Neither of them mentioned the status of tissue inhibitors of metalloproteinases (TIMPs) in their study. The roles of MMPs and TIMPs in BA-associated liver fibrosis are still not well elucidated. In this study, we used a human cytokine gene expression array consisting of 10 members of MMPs and 4 TIMPs to explore the problems.
1. Materials and methods 1.1. Patients and samples For DNA microarray, biopsy specimens were obtained from the liver of 3 patients during Kasai procedure (KP) for BA. There were 2 females and 1 male, ranging in age from 1.3 to 1.7 months. Another 3 liver samples were obtained from patients with BA at the time of LT, with 1 female and 2 males, ranging in age from 10 months to 15 years. Two control liver samples were obtained from 1 female and 1 male patient without liver fibrosis at the age of 2 and 5 months, respectively. One was obtained from a patient who underwent repair of the hiatal hernia; a piece of liver attached to the hernia ring was excised to facilitate closure. Another patient with neonatal hepatitis presented with jaundice; the liver tissue was obtained for equivocal diagnosis of BA. The liver tissues in both control patients were devoid of liver fibrosis microscopically, despite abnormal liver function in the latter case. The liver samples were immediately immersed in liquid nitrogen after snatch and were stored at 808C. Histologic grading of liver fibrosis was performed on all the 8 patients with Masson trichrome staining of the liver specimens, as described by Ferrell [16]. This study was approved by the Medical Ethics and the Human Clinical Trial Committee at Chang Gung Memorial Hospital.
1.2. RNA isolation and generation of complementary DNA
1569 33
P-labeled
Total RNA was extracted from frozen liver tissue using a commercially available reagent REZOL C&T (Protech technology, Taipei, Taiwan), which was purified by phenol-chloroform extraction and quantified by spectrophotometry with absorbance at 260 nm. Fifteen to 20 lg of total RNA was annealed with 1 lg of oligo (dT) [15] primers (Promega, Madison, Wis) and then incubated with 100 lCi of a-33P-dCTP, 800 U of M-MLV reverse transcriptase (Promega), 80 U of RNasin, and dNTP mixture (0.5 mmol/L each of dATP, dGTP, and dTTP and 2.5 lmol/L of dCTP) at 428C for 3 hours to generate 33 P-labeled complementary DNA (cDNA) [17]. The unincorporated nucleotides were removed by using a Sephadex G-25 gel-filtration spin column (Amersham Pharmacia Biotech, Piscataway, NJ).
1.3. Hybridization and array analysis After prehybridization with 3 mL of the hybridization solution (Sigma-Genosys, Woodlands, Tex) at 658C for 1 hour in a hybridization bag, the denatured 33P-labeled cDNA probes were incubated at 658C overnight with the Human Cytokine Expression Array GA001 (R & D Systems, Minneapolis, Minn). The array contained 375 cDNA sequences, including 10 MMPs from MMP-1, -3, -7, -8, -9, -10, -12, -13, -14 to -15 and 4 TIMPs from TIMP-1, -2, -3 to -4, with duplicate spots for each gene on a nylon membrane. The membrane was washed for 20 minutes 3 times with solution I (0.5 SSPE with 1% wt/vol SDS) and then with solution II (0.1 SSPE with 1% wt/vol SDS) at 658C. The membrane was air dried, autoradiographed onto a Kodak Low Storage Phosphor Screen, and analyzed with a Phosphoimager (Storm 840, Molecular Dynamics, Sunnyvale, CA). We used the ImageQuant software (Molecular Dynamics) to analyze the intensity of hybridization on each spot of the array. The intensity of each gene was an average intensity of a pair of duplicate spots. The value was further normalized by the mean value of 9 different housekeeping genes. The normalized signals were compared between groups.
1.4. Real time quantitative reverse transcription– polymerase chain reaction Quantitative reverse transcription–polymerase chain reaction (QRT-PCR) was performed with ABI 7700 Sequence Detection System (TaqMan, Perkin-Elmer Applied Biosystems) to confirm the findings from the array. Polymerase chain reaction was performed in 50 lL SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) containing 10 lmol/L forward primers and reverse primers and approximately 30 ng cDNA. The sequence of the primers of MMP-2, -7, -9, and -13 and TIMP-1, -2, -3, and -4 was designed using the Primer Express Software (Perkin-Elmer Life Science) (Table 1).
1570 Table 1
C.-S. Hsieh et al. Primer sequence of MMPs and TIMPs
Cytokines MMPs MMP-2 MMP-7 MMP-9 MMP-13 TIMPs TIMP-1 TIMP-2 TIMP-3 TIMP-4 Human actin
Forward primers
Reverse primers
5V-CAACTACGATGATGACCGCAA-3V 5V-TACAGTGGGAACAGGCTCAGG-3V 5V-GACCTGGGCAGATTCCAAAC-3V 5V-CTTCACGATGGCATTGCTGA-3V
5V-GTGTAAATGGGTGCCATCAGG-3V 5V-GGCACTCCACATCTGGGCT-3V 5V-CACGCGCAGTGAAGGTGAGC-3V 5V-AACTCATGCGCAGCAACAAG-3V
5V-CCTTCTGCAATTCCGACCTC-3V 5V-TCATTGCAGGAAAGGCCG-3V 5V-TCGGCACGCTGGTCTACAC-3V 5V-CACTCGGCACTTGTGATTCG-3V aˆ-5V-TCACCCACACTGTGCCCATCTACG-3V
5V-ATGTGCAAGAGTCCATCCTGC-3V 5V-GATGCAGGCGAAGAACTTGG-3V 5V-AGCTGGTCCCACCTCTCCA-3V 5V-GCTTCTGGCTGTTGGCTTCT-3V 5V-CAGCGGAACCGCTCATTGCCAATGG-3V
Amplification and detection were performed with the following profile [17]: 1 cycle of 958C for 10 minutes and 40 cycles of 958C for 15 seconds, 608C for 30 seconds, and 728C for 15 seconds. After amplification was completed, a final melting curve was performed with 2 minutes of denaturation at 958C, cooling to 608C, and then heating slowly until 958C (20 minutes) according to the dissociation protocol of the ABI 7700 instrument. Real time fluorescence measurement was read and a threshold cycle (CT) value for each sample was calculated by determining the point at which the fluorescence exceeded a threshold limit, that is, 10 times the standard deviation of the baseline. To confirm the specificity of the PCR reaction, PCR products were electrophoresed on a 1.5% agarose gel.
1.5. Statistical analysis For statistical analysis of the expression data, the relative value of each gene in a patient from the array was calculated and fold induction or reduction in expression of gene-specific messenger RNAs (mRNAs) between
Table 2 Cytokines MMPs MMP-2 MMP-7 MMP-9 TIMPs TIMP-1 TIMP-2 TIMP-3
samples was obtained. An analysis of variance with post hoc Bonferroni was then used to obtain the genes that showed significant difference between groups. Data in real time QRT-PCR were expressed as geometric means after back-transforming the logarithmically transformed variables and thus normally distributed. The relative expression of various genes, 2-DDCT, as measured using real time QRTPCR between groups were compared with unpaired t tests on log-transformed values. A P value of less than .05 was regarded as statistically significant.
2. Results Histologic sections of the liver studied with Masson trichrome stain revealed no fibrosis in the 2 control patients, grade II or III liver fibrosis in the 3 KP patients, and grade IV fibrosis in all 3 LT patients. The hybridization signal intensities of 10 MMPs and 4 TIMPs in cDNA microarray were shown in Fig. 1. The expression of MMPs was generally weak in the control.
Fold difference in microarray and in real time QRT-PCR (QRT-PCR ratio) Fold difference in microarray
QRT-PCR ratio
KP/CO
LT/CO
KP/CO
LT/CO
NA 3.5 F 2.0 1.6 F 0.8
NA 4.2 F 2.5 2.1 F 0.5
2.0 F 1.2 39.8 F 18.8 1.9 F 1.4
5.3 F 1.3T 91.9 F 32.0T 1.4 F 0.9
1.0 F 0.3 1.1 F 0.5 0.7 F 0.2
3.5 F 2.0 5.2 F 3.2 2.1 F 1.1
1.9 F 0.6 2.1 F 1.4 0.3 F 0.1TT
2.4 F 0.7 5.3 F 1.7T 0.7 F 0.2T
For statistical analysis of the expression data, the relative value of each gene in a patient obtained from the array is calculated and fold induction or reduction in expression of gene-specific mRNAs between samples is obtained. The QRT-PCR ratio is the relative level of transcripts measured by real time QRT-PCR in one group over that in the other. CO indicates control. T P b .05 LT vs CO. TT P b .05 KP vs CO.
Evaluation of MMPs and their endogenous tissue inhibitors in BA-associated liver fibrosis
1571
Fig. 1 Part of the scanning picture is shown after hybridization of liver tissue cDNAs with the Human Cytokine Expression Array GA001 (R & D Systems). Genomic DNA (GEN) control spots are located at the right upper and lower corner of the array, whereas 9 housekeeping genes (HG) are starting from the second cell on the same column as GEN, followed by 6 negative controls (NC). There are 10 MMPs starting from MMP-1 in the first column from the left-hand side to MMP-15 in the top row of the second column. The strong intensity of TIMP-1, -2, and -3 is located with arrow in the second column, whereas TIMP-4 that locates immediately below TIMP-3 is too weak to be visualized. The most prominent finding is faint MMP-7 expression in the control (arrowhead), which is followed by significant increase in signal intensity in KP. The signal of MMP-7 is most prominent in LT. All the other 9 MMPs continued to be weak despite increasing liver fibrosis from control to LT.
With increasing liver fibrosis, only the intensity of MMP-7 became stronger; its expression in LT was 2- and 5-fold higher than that in KP and control, respectively. The expression of the other 9 MMPs was weak despite progression of liver fibrosis. Nonetheless, there was an increase in expression of MMP-9 and MMP-13 in LT over that in control and KP in the array. The signal intensity of TIMP-1 was the strongest of the 4 members of TIMPs in the array, which was followed by TIMP-3 and TIMP-2 (Fig. 1). The signal intensity of TIMP-4 was too weak to be detected in the array and the difference in signal intensity was not significant between groups. The change of signal intensity of both TIMP-1 and TIMP-2 paralleled the progression of liver fibrosis. There were 2.5- and 3.5-fold increases in signal intensity in TIMP-1 in LT compared with KP and control, respectively. There were 4.7- and 5.2-fold increases in signal intensity in TIMP-2 in LT compared with KP and control, respectively. The signal intensity of TIMP-3 was strongest in control and weakest in KP, resulting in inverse relationship of the expression of TIMP-3 with the active fibrogenesis in BA.
By using real time QRT-PCR, MMP-7 was the only gene among the 10 members of MMPs that demonstrated sturdy and significant increase of transcripts with the progression of liver fibrosis. The QRT-PCR ratio, that is, the relative level of transcripts measured by real time QRT-PCR in one group over that in the other, was 1.7 in LT over KP, up to 39.8 in KP over control, and up to 91.9 in LT over control, respectively. The difference was highly significant ( P b.001). The MMP-9 mRNA expression was higher in KP than in LT and control in real time QRT-PCR. The expression of MMP-13 mRNA in the liver of the study groups was too low to be detected using real time QRT-PCR. Although MMP-2 was not included in this cDNA microarray, we did real time QRT-PCR to quantify MMP-2 mRNA. The results showed a significant upregulation of MMP-2 in LT than in KP and control (Table 2). There was a 2.4-fold increase in TIMP-1 in LT compared with the control. There was a 5.3-fold increase in TIMP-2 in LT compared with the control. The differences were significant in both. The expression of TIMP-3 was highest in control, followed by LT and was lowest in KP, resulting in significantly higher QRT-PCR ratio in LT over KP than in KP over control (Table 2).
1572
Fig. 2 Gel electrophoresis of PCR products after 30 cycles of amplification confirms the findings of QRT-PCR. Two samples from each group are randomly chosen without selection. The intensity of the TIMP-1, TIMP-2, MMP-2, and MMP-7 bands of representative patients in LT group (lane 5-6) is higher than in KP (lane 3-4), and the latter is again higher than in control (lane 1-2). The intensity of MMP-9 bands is highest in KP, followed by LT and control. The intensity of TIMP-3 bands is, on the contrary, lowest in KP, followed by LT and control. The expression of housekeeping gene aˆ-actin is constant between groups.
Gel electrophoresis confirmed the findings of real time QRT-PCR. For genes included in this study, the intensity of the bands correlated well with the QRT-PCR ratio (Fig. 2).
3. Discussion Our findings revealed that TIMP-4 and all 10 MMPs studied were not constitutively expressed in the normal liver. Conversely, there was constitutive expression of TIMP-1, -2, and -3. These findings are consistent with previous reports [18,19]. With increasing liver fibrosis, DNA microarray study demonstrated an increasing signal intensity in MMP-7 only, but not in the 9 other MMPs. Upregulation of MMP-7 was also demonstrated in the other 2 studies of BA using gene chips or cDNA microarray, denoting some role of this unique protease in the progress of liver fibrosis/cirrhosis associated with BA [14,15]. Matrix metalloproteinase 2 was found to be downregulated in BA over normal liver in the study of Chen et al [15], but the difference was less than 2-fold. Although MMP-2 was not included in this cDNA microarray study of BA, we still
C.-S. Hsieh et al. performed real time QRT-PCR to quantify this frequently cited protease. Our findings were consistent with upregulation of MMP-2 mRNA in one study of fibrotic human liver in BA, but contrary to the study of Chen et al [15,20]. We suspect that MMP-2, like other members of MMPs, is expressed in very low level in the liver despite increasing liver fibrosis, which may result in false downregulation when using cDNA microarray study of gene expression without further confirmation with either Northern blotting or QRTPCR. A similar problem happens to MMP-13. Matrix metalloproteinase 13 was found upregulated more than 2-fold in the array, but subsequent real time QRT-PCR failed to detect the MMP-13 transcript because of low copy number. It is known that genes expressed at low or near background levels may result in exceptionally high false-positive rate on the ratios and could be ignored in expression analysis [21]. Significant upregulation of hepatic MMP-9 mRNA was found in the early stage of BA in which the patient received KP in this study. The expression decreases in the late stage of BA, but the level was still higher than that in the control. Similar findings are found in 2 animal studies, which show rapid upregulation of MMP-9 mRNA or the proteolytic activity after experimentally induced liver fibrosis in mice or biliary fibrosis in rats [22,23]. However, a study of fibroproliferative process in the liver during chronic hepatitis C fails to show significant change of MMP-9 expression between patients with or without liver fibrosis [24]. The findings of significant upregulation of TIMP-1 and TIMP-2 in fibrotic liver in BA are consistent with the study of Benyon et al [20]. A recent report from Korea using differential expressed sequence tag screening identifies TIMP-1 overexpressed in biliary cirrhosis [25]. Our study using human cytokine cDNA array demonstrated a greater increase of TIMP-2 than TIMP-1 with increasing fibrosis in BA. The findings were further confirmed with QRT-PCR. For both TIMP-1 and TIMP-2, the fold difference in DNA microarray was correlated with the QRT-PCR ratio. The findings are consistent with in situ hybridization study of TIMP-1 and TIMP-2 expression in rat and human liver fibrosis [26], but not with the report of Kobayashi et al [27]. In the latter study, Kobayashi et al [27] used 1-step sandwich enzyme immunoassays to measure serum MMP-1 and -2 as well as TIMP-1 and -2 in 36 postoperative patients with BA and found no correlation of TIMP levels with the progress of fibrosis between groups. The difference may be explained as that MMPs and TIMPs in the blood samples are contaminated by other sources or mRNA of these cytokines in the liver fails to reflect the protein contents in situ or in the remote site such as blood. In this study, the finding of TIMP-3 expression in the liver is contrary to previous notion that this protease is not present in the liver [28]. We also find that the mRNA levels of TIMP-3 are highest in control and lowest in the early stage of BA. The findings imply significant downregulation of TIMP-3 during the stage of active fibrogenesis in BA. No similar findings have been found in the literature. Because
Evaluation of MMPs and their endogenous tissue inhibitors in BA-associated liver fibrosis the protein of TIMP-3 displays low sequence similarity to human TIMP-1 and TIMP-2 [28], the unique finding of downregulation of TIMP-3 mRNA expression in the early stage of BA implies a different role for human TIMP-3 in human liver tissue remodeling. Although altered balance between MMPs and TIMPs is thought to be important in liver fibrosis, controversy exists as to their role in fibrogenesis. These discordant findings denote a complex change in the local MMP/TIMP balance that underlies liver fibrosis caused by different liver diseases. There has been a growing focus on the role of MMPs in liver disease [29]. However, the complex findings in this study—some consistent with and some contrary to those present in the literature—suggest that the study of this family of zinc- and calcium-dependent proteases and their inhibitors is still far from completion.
Acknowledgments This study was supported by the National Science Council, Taiwan grant NSC91-2314-B-182-065 and the Chang Gung Memorial grants CMRPG8005 and CMRPG8067. We thank the Liver Transplant Team, Chang Gung Memorial Hospital–Kaohsiung, for providing the samples used in this study.
References [1] Karrer FM, Price MR, Bensard DD, et al. Long-term results with the Kasai operation for biliary atresia. Arch Surg 1996;131:493 - 6. [2] Nio M, Ohi R, Shimaoka S, et al. The outcome of surgery for biliary atresia and the current status of long-term survivors. Tohoku J Exp Med 1997;181:235 - 44. [3] Chen CL, Chen YS, de Villa VH, et al. Minimal blood loss living donor hepatectomy. Transplantation 2000;69:2580 - 6. [4] Jain A, Reyes J, Kashyap R, et al. Long-term survival after liver transplantation in 4,000 consecutive patients at a single center. Ann Surg 2000;232:490 - 500. [5] Miller CM, Gondolesi GE, Florman S, et al. One hundred nine living donor liver transplants in adults and children: a single-center experience. Ann Surg 2001;234:301 - 11. [6] Friedman SL, Maher JJ, Bissell DM. Mechanisms and therapy of hepatic fibrosis: report of the AASLD single topic basic research conference. Hepatology 2000;32:1403 - 8. [7] Benyon RC, Arthur MJP. Extracellular matrix degradation and the role of hepatic stellate cells. Semin Liver Dis 2001;21:373 - 84. [8] Faiz Kabir Uddin Ahmed A, Ohtani H, Nio M, et al. In situ expression of fibrogenic growth factors and their receptors in biliary atresia: comparison between early and late stages. J Pathol 2000;192:73 - 80. [9] Ahmed AF, Nio M, Ohtani H, et al. Clinical significance of c-kit expression in biliary atresia. Pediatr Surg Int 2001;17:601 - 3. [10] Liu C, Chiu JH, Chin TW, et al. Expression of aminopeptidase N in bile canaliculi: a predictor of clinical outcome in biliary atresia. J Surg Res 2001;10:76 - 83.
1573
[11] Davenport M, Gonde C, Redkar R, et al. Immunohistochemistry of the liver and biliary tree in extrahepatic biliary atresia. J Pediatr Surg 2001;36:1017 - 25. [12] Rosensweig JN, Omori M, Page K, et al. Transforming growth factor– beta 1 in plasma and liver of children with liver disease. Pediatr Res 1998;44:402 - 9. [13] Lee SY, Chuang JH, Huang CC, et al. Identification of transforming growth factors actively transcribed during the progress of liver fibrosis in biliary atresia. J Pediatr Surg 2004;39:702 - 8. [14] Bezerra JA, Tiao G, Ryckman FC, et al. Genetic induction of proinflammatory immunity in children with biliary atresia. Lancet 2002;360:1653 - 9. [15] Chen L, Goryachev A, Sun J, et al. Altered expression of genes involved in hepatic morphogenesis and fibrogenesis are identified by cDNA microarray analysis in biliary atresia. Hepatology 2003;38: 567 - 76. [16] Ferrell L. Liver pathology: cirrhosis, hepatitis, and primary liver tumors. Update and diagnostic problems. Mod Pathol 2000;13: 679 - 704. [17] Huang CC, Chuang JH, Huang LLH, et al. The human Delta-like 1 homologue is implicated in the progression of liver fibrosis in biliary atresia. J Pathol 2004;202:172 - 9. [18] Cursio R, Mari B, Louis K, et al. Rat liver injury after normothermic ischemia is prevented by a phosphinic matrix metalloproteinase inhibitor. FASEB J 2002;16:93 - 5. [19] Boker KH, Pehle B, Steinmetz C, et al. Tissue inhibitors of metalloproteinases in liver and serum/plasma in chronic active hepatitis C and HCV-induced cirrhosis. Hepatogastroenterology 2000;47:812 - 9. [20] Benyon RC, Iredale JP, Goddard S, et al. Expression of tissue inhibitor of metalloproteinases 1 and 2 is increased in fibrotic human liver. Gastroenterology 1996;110:821 - 31. [21] Baldi P, Long AD. A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes. Bioinformatics 2001;17:509 - 19. [22] Vaillant B, Chiaramonte MG, Cheever AW, et al. Regulation of hepatic fibrosis and extracellular matrix genes by the th response: new insight into the role of tissue inhibitors of matrix metalloproteinases. J Immunol 2001;167:7017 - 26. [23] Kossakowska AE, Edwards DR, Lee SS, et al. Altered balance between matrix metalloproteinases and their inhibitors in experimental biliary fibrosis. Am J Pathol 1998;153:1895 - 902. [24] Lichtinghagen R, Michels D, Haberkorn CI, et al. Matrix metalloproteinase (MMP)-2, MMP-7, and tissue inhibitor of metalloproteinase-1 are closely related to the fibroproliferative process in the liver during chronic hepatitis C. J Hepatol 2001;34:239 - 47. [25] Choe BH, Kim KM, Kwon S, et al. The pattern of differentially expressed genes in biliary atresia. J Korean Med Sci 2003;18: 392 - 6. [26] Herbst H, Wege T, Milani S, et al. Tissue inhibitor of metalloproteinase-1 and -2 RNA expression in rat and human liver fibrosis. Am J Pathol 1997;150:1647 - 59. [27] Kobayashi H, Li ZX, Yamataka A, et al. Clinical evaluation of serum levels of matrix metalloproteinase and tissue inhibitors of metalloproteinase as predictors of progressive fibrosis in postoperative biliary atresia patients. J Pediatr Surg 2002;37:1030 - 3. [28] Uria JA, Ferrando AA, Velasco G, et al. Structure and expression in breast tumors of human TIMP-3, a new member of the metalloproteinase inhibitor family. Cancer Res 1994;54:2091 - 4. [29] Bansal MB, Friedman SL. The enlarging role of matrix metalloproteinases in liver injury—beyond scar degradation. J Hepatol 2002; 37:293 - 4.