The Extracellular Matrix Protein ITIH5 Is a Novel Prognostic Marker In

Home , Aachen

Oncogene (2008) 27, 865–876 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ONCOGENOMICS The extracellular matrix protein ITIH5 is a novel prognostic marker in invasive node-negative breast cancer and its aberrant expression is caused by promoter hypermethylation

J Veeck1,6, M Chorovicer1,6, A Naami1, E Breuer1, M Zafrakas1, N Bektas1,MDu¨ rst2, G Kristiansen3, PJ Wild4, A Hartmann5, R Knuechel1 and E Dahl1

1Molecular Oncology Group, Institute of Pathology, University Hospital of the RWTH Aachen, Aachen, Germany; 2Department of Gynecology and Obstetrics, Friedrich-Schiller-University, Jena, Germany; 3Institute of Pathology, Universita¨tsmedizin Berlin, Berlin, Germany; 4Institute of Pathology, University of Zu¨rich, Zu¨rich, Switzerland and 5Institute of Pathology, University of Regensburg, Regensburg, Germany

Inter-a-trypsin inhibitors (ITIs) are protease inhibitors promoter methylation-mediated loss of ITIH5 expression stabilizing the extracellular matrix. ITIs consist of one is associated with unfavourable outcome in breast cancer light (bikunin) and two heavy chains (ITIHs). We have patients, and thus ITIH5 could be used as a prognostic recently characterized ITIH5, a novel member of the marker, although this marker is not multivariate inde- ITIH gene family, and showed that its messenger RNA is pendent due to its close association with ER expression. lost in a high proportion of breast tumours. In the present Our data indicate that ITIH5 is a candidate class II study, an ITIH5-specific polyclonal antibody was gener- tumour suppressor gene and could be involved in tumour ated, validated with western blot and used for immuno- progression, invasion and metastasis, as its absence is histochemical analysis on a tissue microarray; ITIH5 was associated with increased proliferation rates and a strongly expressed in epithelial cells of normal breast prognostic value indicating poor clinical outcome. (n ¼ 11/15), while it was lost or strongly reduced in 42% Oncogene (2008) 27, 865–876; doi:10.1038/sj.onc.1210669; (92/217) of invasive breast cancers. ITIH5 expression in published online 23 July 2007 invasive carcinomas was associated with positive expression of oestrogen receptor (P ¼ 0.008) and histological Keywords: inter-a-trypsininhibitor heavy chain(ITIH); grade (P ¼ 0.024). Correlation of ITIH5 expression with breast cancer; prognostic marker; predictive marker; tumour clinical outcome revealed that patients with primary invasion; metastasis tumours retaining abundant ITIH5 expression had longer recurrence-free survival (RFS; P ¼ 0.037) and overall survival (OS; P ¼ 0.044), compared to those with reduced Introduction expression (mean RFS: 102 vs 78 months; mean OS: 120 vs 105 months). Methylation-specific PCR analysis The inter-a-trypsininhibitors (ITIs) comprise a family of frequently showed strong methylation of the ITIH5 protease inhibitors found in the extracellular matrices of promoter in primary breast tumours (41%, n ¼ 109) and various organs, as well as in the blood circulation. breast cancer cell lines (n ¼ 6). Methylation was signifi- Owing to their original isolation in complexes with cantly associated with mRNA loss (Po0.001; n ¼ 39), and hyaluronan (HA), ITIs are also referred to as SHAPs ITIH5 expression was induced after treatment of tumour 0 (Serum-derived HA-associated proteins) (Yoneda et al., cell lines with the demethylating agent 5-aza-2 -deoxycy- 1990). It has beenpreviously shownthat interactionof tidine. Moreover, ITIH5 promoter methylation was ITIs with HA leads to stabilizationof the extracellular significantly associated with reduced OS (P ¼ 0.008). matrix (Chen et al., 1994). ITI molecules consist of three The cellular function of ITIH5 was evaluated by forced protein chains: one ITI light chain, also referred to as ONCOGENOMICS expression of a full-length ITIH5 complementary DNA in bikunin, and two ITI heavy chains (ITIHs) (Enghild the breast cancer cell line MDA-MB-231, which does not et al., 1991). The transfer of ITIHs onto HA requires endogenously express ITIH5. ITIH5-expressing clones tumour necrosis factor a-induced protein 6 (TNFAIP6), showed a 40% reduced proliferation rate compared to also known as TNF-stimulated gene 6 (Jessen and mock-transfected cells. Overall, these data show that Odum, 2003). TNFAIP6 forms a stable complex (Rugg et al., 2005) with ITIH and HA during the transester- Correspondence: E Dahl, Molecular Oncology Group, Institute of ificationreaction(Sanggaard et al., 2005). This forma- Pathology, University Hospital Aachen, RWTH Aachen, Pauwels- tionof ITIH–HA complexes is thought to play an strasse 30, 52074 Aachen, Germany. important role in the stabilization of HA-rich extra- E-mail: [email protected] 6These two authors contributed equally to this work. cellular matrices, with TNFAIP6 acting as an essential Received 21 March 2007; revised 4 June 2007; accepted 13 June 2007; cofactor and catalyst (Rugg et al., 2005). To date, published online 23 July 2007 five distinct ITIHs have been identified, encoded by five ITIH5 expression in breast cancer J Veeck et al 866 genes located on two different chromosomes (Diarra- in breast cancer cell lines and primary breast tumour Mehrpour et al., 1989; Himmelfarb et al., 2004). tissue samples and the biological role of ITIH5 was ITIH1, -3 and -4 have beenmapped to chromosome evaluated with functional in vitro assays. Clinicopatho- 3p2.11–12, and ITIH2 and -5 are located onchromo- logical patient characteristics were statistically corre- some 10p14–15 (Bost et al., 1998; Himmelfarb et al., lated with expressionandmethylationdata andrevealed 2004). ITIH1, -2 and -3 are synthesized primarily in liver an unfavourable prognosis in case of methylation- as polypeptide precursors, which undergo extensive mediated loss of ITIH5 expression. post-translational processing; they all contain a con- served cleavage site, which enables them to form covalent bonds to bikunin via glycosaminoglycan (Enghild et al., 1991). Results Various studies (Kobayashi et al., 1995; Bourguignon et al., 1999; Paris et al., 2002; Zhang et al., 2004) have Characterization of the anti-ITIH5-specific antibody shown involvement of ITIs in tumour biology using an The specificity of the rabbit polyclonal antiserum raised expressed sequence-tag-based bioinformatics approach against a synthetic peptide corresponding to amino acids (Dahl et al., 2005). We have previously identified ITIH5 207–220 of humanITIH5 proteinwas evaluated by as a novel gene differentially expressed in breast cancer. western blot analysis. The antibody is able to detect After cloning, mapping and determination of its humanITIH5 proteinexpressed in ITIH5-transfected genomic organization, we analysed the expression of COS7 cells (Figure 1a, lane 1) as well as in ITIH5- ITIH5 at the messenger RNA level in a panel of normal transfected MDA-MB-231 breast cancer cell lines human tissues and a small set of normal and malignant (Figure 1b; lanes 1, 3, 5 represent different clones). No breast tissue samples. Initial analyses showed that proteinwas detectable inmock-transfected COS7 cells ITIH5 was predominantly expressed in human female (Figure 1b; lane 2) and mock-transfected MDA-MB-231 reproductive tissues and strongly downregulated in cells (Figure 1b; lanes 2, 4, 6), arguing that the ITIH5 breast tumours, suggesting a potential role in breast antibody does not detect any protein, that is not a cancer development (Himmelfarb et al., 2004). Inthe modificationof the ITIH5 protein.ITIH5 proteininthe present study, protein expression of ITIH5 was analysed transfected cell lines was approximately 100 kDa in size, ina large set of breast tumours usingtissue microarrays inaccordancewith its deduced molecular weight. (TMAs). Furthermore, the epigenetic configuration of Immunocytochemical staining of transfected cells the ITIH5 promoter was comprehensively investigated showed abundant ITIH5 protein in the cytoplasm of

175 ITIH5 83 62 123456 47

32 100 kDa ITIH5

β-Actin β-Actin

Figure 1 Characterizationof the ITIH5 antibody.( a and b) Detectionof ITIH5 proteinof lysates from stable ITIH5 clones of COS7 cells (a) and MDA-MB-231 cells (b). Lane 2 in (a) and lanes 2, 4, 6 in (b) represent lysates from mock-transfected cells (pBK-CMV). A specific signal of the ITIH5 proteinapproximately 100 kDa insize is seenonlyinthe ITIH5 clones (lane 1 in (a) and lanes 1, 3, 5 in (b)), but not in the mock- transfected cells. b-Actin was detected to prove equal loading of protein. (c and d) Immunocytochemistry of MDA-MB-231 cells transfected either with ITIH5 cDNA (c) or with empty vector (d). The ITIH5 antibody specifically detects ITIH5 protein in the ITIH5-transfected cell line. Magnification: Â 200. ITIH, inter-a-trypsin inhibitor heavy chain.

Oncogene ITIH5 expression in breast cancer J Veeck et al 867

Figure 2 Expression of ITIH5 in normal breast tissue, non-invasive and invasive breast tumours. (a) Placental tissue, known to abundantly express ITIH5 (Himmelfarb et al., 2004) served as positive control. In negative controls (b), primary ITIH5-antibody was omitted. (c) Strong ITIH5 expressioninnormalbreast tissue. ( d) Scale-up of specimenshownin( c). The magniﬁcation demonstrates abundant ITIH5 expression in luminal-epithelial breast cells. (e) Very abundant ITIH5 expression in a ductal carcinoma in situ of the breast. (f) Scale-up of specimenshownin (e). (g) Invasive ductal breast cancer without detectable ITIH5 expression (IRSo2). (h) Scale-up of specimenshownin( g). (i) Invasive ductal breast cancer with moderate ITIH5 expression (IRS ¼ 4). (j) Scale-up of specimenshowninshownin( i). (k) Invasive ductal breast cancer with abundant ITIH5 expression (IRS ¼ 12). (l) Scale-up of specimenshownin( k). Magniﬁcations: (c, e, g, i and k): Â 100; (j and l): Â 200; (a, b, d, f and h): Â 400. IRS, immunoreactivity score; ITIH, inter-a-trypsin inhibitor heavy chain.

MDA-MB-231 breast cancer cells (Figure 1c). Mock- Correlation of ITIH5 expression with clinicopathological transfected MDA-MB-231 cells did not exhibit ITIH5 parameters and patient survival staining (Figure 1d). Clinicopathological characteristics were correlated with ITIH5 immunohistochemistry results for descriptive data analysis (Table 1). ITIH5 immunohistochemical ITIH5 protein is downregulated in the course of breast staining was significantly associated with positivity of tumour progression oestrogenreceptor (ER) ( P ¼ 0.008) and low (G1/G2) Immunohistochemical analysis was applied to investi- histological grade (P ¼ 0.024). Recurrence-free survival gate ITIH5 protein expression in normal and malignant (RFS) and overall survival (OS) were compared between breast tissue using a TMA containing 217 invasive invasive breast tumours showing strong ITIH5 expres- breast carcinomas, 10 ductal carcinomas in situ (DCIS) sion(IRS ¼ 12) and all other invasive tumours by and 15 normal breast tissue samples. Intensity and univariate log-rank statistics (Table 2). Strong ITIH5 quantity of immunohistochemical staining was evalu- expressionwas clearly associated with longerRFS and ated using a semiquantitative immunoreactivity score OS, as shown by Kaplan–Meier analysis (Figure 3). (IRS) (Remmele and Stegner, 1987). ITIH5 protein was Patients with strong ITIH5 expression in the tumour strongly expressed in 73% (11/15) of normal breast had anestimated meanRFS of 102 months(95% tissue samples (Figures 2c and d) and was localized in confidence interval (CI): 88–116) compared to 78 luminal epithelial cells of the normal breast with absence months (95% CI: 68–88) in patients with loss of ITIH5 in myoepthelial cells, fibroblasts, adipocytes and en- expression( P ¼ 0.037) (Figure 3a). Strong ITIH5 dothelial cells. Subcellularly, ITIH5 was localized to the expressionwas also associated with longerOS of 120 cytoplasm and the cell membrane. In DCIS (Figures 2e months (95% CI: 108–132) compared to 105 months and f), ITIH5 expression was as prominent as in normal (95% CI: 96–114) inpatientswith partial or total loss of breast tissue. However, invasive carcinomas showed ITIH5 expression( P ¼ 0.044) (Figure 3d). Ina stratified strongly reduced or complete loss (IRSp2) of ITIH5 univariate analysis, the prognostic value of ITIH5 expression in 42.4% (92/217) (Figures 2g and h) and became even more pronounced in the clinically im- a moderate expression(IRS 3–9) in28.1% (61/217) portant subgroup of node-negative patients (Figures 3b of cases (Figures 2i and j). Interestingly, 29.5% and e) for RFS and OS alike. However, Cox regression (64/217) of invasive breast tumours had an IRS of models including tumour grade (G1 and G2 vs G3), 12, thus presenting very abundant ITIH5 expression pT-stage (pT1 and pT2 vs pT3 and pT4), node status (Figures 2k and l). (negative vs positive), hormone receptor status (ER/PR

Oncogene ITIH5 expression in breast cancer J Veeck et al 868 Table 1 Clinicopathological and immunohistochemical factors in Table 2 Univariate analysis of clinicopathological factors regarding relationto ITIH5 immunoreactivity recurrence-free survival and overall survival Variable ITIH5 immunoreactivity Variable RFS OS

na IRS o12 IRS ¼ 12 P-value na Events P-value n Events P-value

Clinicopathological factors: Clinicopathological factors: Tumour stageb Tumour stageb pT1–2 172 127 45 1.0 pT1–2 168 59 o0.0001 173 47 0.0001 pT3–4 41 24 17 pT3–4 37 23 41 24 Lymph node statusb Lymph node statusb pN0 97 68 29 1.0 pN0 96 16 o0.0001 98 14 o0.0001 pN1–3 109 77 32 pN1–3 105 61 109 51 Histological grade Histological grade G1/G2 120 77 43 0.024 G1 22 5 0.0001 23 5 0.0005 G3 93 73 20 G2 91 28 97 23 Oestrogen receptor status G3 92 49 94 43 Negative (IRS 0–2)c 55 46 9 0.008 Histology Positive (IRS 3–12) 120 76 44 Ductal 168 71 0.2895 173 58 0.8779 Progesterone receptor status Lobular 16 5 18 6 Negative (IRS 0–2)c 122 89 33 1.0 Other 19 5 21 6 Positive (IRS 3–12) 62 38 24 Oestrogen receptor status p53 nuclear staining Negative 56 29 0.0239 56 24 0.0161 o5% 102 68 34 0.229 Positive 115 38 120 32 X5% 67 51 16 Progesterone receptor status HER2 IHC Negative 115 54 0.0029 123 51 0.0006 Negative (0;1+) 143 97 46 0.849 Positive 62 15 62 10 Positive (2+;3+) 41 29 12 Hormone receptor status (combined) Negative 49 27 0.0110 49 21 0.0330 Abbreviations: IHC, immunohistochemistry; IRS, immunoreactivity Positive 116 38 121 35 score; ITIH, inter-a-trypsin inhibitor heavy chain. Significant P-values ITIH5 marked inbold face. aOnly female patients with primary, unilateral, IRSc ¼ 12 63 18 0.0369 63 14 0.0436 invasive breast cancer were included. bAccording to UICC: TNM IRSo12 142 64 151 57 classification by Sobin and Wittekind (2002). cIRS according to Remmele and Stegner (1987). Abbreviations: IRS, immunoreactivity score; ITIH, inter-a-trypsin inhibitor heavy chain; OS, overall survival, RFS, recurrence-free survival. Significant P-values marked inbold face. aOnly female patients with primary, unilateral, invasive breast cancer were included. combined, double negative vs any positive) and ITIH5 bAccording to UICC: TNM classification by Sobin and Wittekind (2002). cIRS according to Remmele and Stegner (1987). (IRS 12 vs 0–9) failed significance in confirming the prognostic value of ITIH5 as an independent marker (Tables 3a and b) due to the close association with ER. MCF12A and Hs578T cells and absence of ITIH5 mRNA inBT20, MCF7, SKBR3, T47D andMDA- MB-231 breast cancer cells (Figure 4c). This association ITIH5 downregulation in breast cancer is caused by was further supported by treatment of four cell lines promoter hypermethylation lacking ITIH5 expressionwith 5-aza-2 0-deoxycytidine Analysis of the ITIH5 gene promoter using the genomic (DAC) and trichostatin A (TSA): RT–PCR and real- DNA information contained in ENSEMBL contig time PCR analyses 24 h after treatment (Figures 4d and ENSG00000123243 based onNCBI Build 36.1 showed e) showed expressionof ITIH5 inall cell lines three CpG-rich islands between genomic positions (P ¼ 0.014, one-tailed U-test). 7747928 and 7749039 (À1012 to þ 99 relative to the Consistently, analysis of 109 invasive human breast expected transcription start site) on chromosome 10p. cancer and ten matching normal breast tissue samples Figure 4a gives anoverview of the analysed promoter by MSP showed ITIH5 promoter hypermethylationin sequence and the regulatory region analysed by methy- 77 breast tumours (70.6%), and a very weak methyla- lation-specific PCR (MSP). tion signal in only one normal breast tissue sample. Of Using MSP (Herman et al., 1996), a nonmalignant the methylated samples, 32/77 (29.4% of all) had only (MCF12A) and six malignant breast cell lines (BT20, weak methylation and 45/77 (41.3% of all) had strong MCF7, SKBR3, T47D, MDA-MB-231 and Hs578T) methylationdetectable. Representative results are shown were initially studied (Figure 4b): The nonmalignant cell inFigure 5a. line exhibited lack of ITIH5 promoter methylationin the analysed promoter region, whereas five of six malignant cell lines showed ITIH5 promoter methyla- Correlation of promoter methylation with ITIH5 mRNA tion—Hs578 T was the only exception. Reverse tran- expression scription–PCR (RT–PCR) analysis showed an inverse To further prove that promoter methylationabrogates associationbetweenpromoter methylationand ITIH5 ITIH5 expression, we analysed in parallel 39 samples by expression, that is, abundant ITIH5 expressionin real-time PCR and also by MSP. Figure 5b illustrates

Oncogene ITIH5 expression in breast cancer J Veeck et al 869

Figure 3 Univariate survival analysis of ITIH5 immunoreactivity according to the Kaplan–Meier (KM) method concerning recurrence-free (a–c) and overall survival (d–f). In( a)and(d) the KM-Analysis of all cases is shown. (b)and(e) illustrate the strong prognostic value of ITIH5 in the clinically relevant group node-negative patients (pN0), which is not demonstrable in node-positive patients (pN1–3) (c)and(f). Grey line: strong ITIH5 staining (IRS ¼ 12); black line: weaker ITIH5 staining. Crosses in each graph indicate censored patients. IRS, immunoreactivity score; ITIH, inter-a-trypsin inhibitor heavy chain.

Oncogene ITIH5 expression in breast cancer J Veeck et al 870 Table 3a Cox multivariate analysis for RFS and ITIH5 immunohisto- Reduced proliferation rates of MDA-MB-231 breast chemistry cancer cells after forced expression of ITIH5 P-value Hazard 95.0% confidence A full-length ITIH5 cDNA derived from normal breast ratio (HR) interval of HR tissue was cloned into the eukaryotic expression vector pBK-CMV (Himmelfarb et al., 2004). A proliferation Lower Upper assay was performed for two independent ITIH5- ITIH5 0.192 0.658 0.351 1.235 transfected clones in MDA-MB-231 cells and two pTa 0.020 2.008 1.114 3.619 independent mock-transfected clones using XTT (2,3- G 0.012 2.075 1.174 3.666 Bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium- a pN 0.001 2.914 1.559 5.447 5-carboxanilide) assays (Figure 6). ITIH5 protein ER/PR 0.080 0.613 0.354 1.061 expressioninthe ITIH5-transfected clones and lack of Abbreviations: ER, Oestrogen receptor status; G, histological grading; expressioninthe mock-transfected cloneshad been ITIH, inter-a-trypsin inhibitor heavy chain; PR, progesterone receptor verified by western blot (Figure 2b, lanes 3 and 5 and status; pN, lymph node status; RFS, recurrence-free survival. lanes 2 and 4, respectively). The optical density Significant P-values marked inbold face. aAccording to UICC: measured at the first time point (24 h) was set equal to 1. TNM classification by Sobin and Wittekind (2002). We observed an increasing growth retardation of the ITIH5-transfected compared with the mock-transfected Table 3b Cox multivariate analysis for OS and ITIH5 immunohisto- cells, which reached 39.8% growth reduction96 h after chemistry plating. P-value Hazard 95.0% confidence ratio (HR) interval of HR

Lower Upper Discussion ITIH5 0.407 0.752 0.384 1.475 pTa 0.015 2.134 1.158 3.932 Thus far, numerous studies have shown involvement of G 0.057 1.756 0.984 3.134 ITIs invarious pathological processes, includingforma- a pN 0.003 2.976 1.463 6.052 tion of urinary tract stones (Atmani and Khan, 1999; ER/PR 0.025 0.500 0.272 0.916 Janssen et al., 2001), female infertility (Zhuo et al., Abbreviations: ER, Oestrogen receptor status; G, histological grading; 2001), atherosclerosis and vascular disease (Onda et al., ITIH, inter-a-trypsin inhibitor heavy chain; OS, overall survival; PR, 1999; Fujita et al., 2004), and inflammation (Balduyck progesterone receptor status; pN, lymph node status. Significant et al., 2000; Yingsung et al., 2003; Pineiro et al., 2004). P-values marked inbold face. aAccording to UICC: TNM classifica- Furthermore, several lines of evidence suggest a tion by Sobin and Wittekind (2002). potential role of ITIs in tumour biology, particularly inthe metastatic cascade andtumour progression (Kobayashi et al., 1995; Bourguignon et al., 1999; Paris the distributionof ITIH5 expressionamongthe et al., 2002; Zhang et al., 2004). groups of unmethylated, weakly methylated and We have previously described cloning and RNA strongly methylated samples. Comparison of the median expression analysis of a novel member of the ITIH gene expression level of each group reveals a significant family, ITIH5, in normal and malignant breast tissues coherence with promoter methylation, since a stable (Himmelfarb et al., 2004). We have now generated a decrease of ITIH5 mRNA with increasing promoter polyclonal anti-ITIH5 antibody to evaluate ITIH5 methylationwas detected. expression at the protein level employing a comprehen- sive TMA. Consistent with our previously published RNA Correlation of ITIH5 methylation with expression data, we found that ITIH5 was down- clinicopathological parameters and patient survival regulated at the proteinlevel ina large number of Clinicopathological characteristics were correlated breast tumours as compared to normal breast tissue, with ITIH5 methylationfor descriptive data analysis suggesting that ITIH5 expression is lost in the course of (Table 4). ITIH5 promoter methylationwas not tumour progression. Correlation with clinicopathologi- significantly associated with tumour stage, lymph node cal data of the invasive breast tumours analysed in the status or histological grade. RFS and OS were compared present study showed that strong ITIH5 immunohisto- betweenmethylated andunmethylated ITIH5 alleles by chemical staining was associated with longer RFS univariate log-rank statistics. ITIH5 methylationdis- and OS. Intriguingly, ITIH5 remains a significant played a trend towards association with occurrence of prognostic factor in the clinically important subgroup relapse (P ¼ 0.067; data not shown) and was highly of patients with node-negative tumours, as shown by significantly associated with shorter OS (P ¼ 0.008) univariate analysis, although this finding was multi- (Table 4) as also illustrated by Kaplan–Meier analysis variate insignificant when the hormone receptor status (Figures 5c and d). A Cox regression model was was included in the analysis. It remains to be shown in calculated onOS, but all parameters included failed confirmatory studies, if ITIH5-positivity can help to significance, although ITIH5 methylationwas of border- identify those patients with very low risk of tumour line significance (Table 5; P ¼ 0.055). recurrence.

Oncogene ITIH5 expression in breast cancer J Veeck et al 871

Figure 4 ITIH5 methylation and expression analysis in breast cell lines. (a) Schematic representation of the human ITIH5 promoter region. A B1.2-kb fragment is shown. A, B and C indicate predicted binding sites of regulatory transcription factors (A: CCAAT/enhancer-binding- protein-b; B: MYC-MAX-binding site; C: GC-box elements). The CpG-rich islands (black bars) are shown relative to transcriptional start ( þ 1). Double-head arrow represents the amplified region by methylation-specific PCR (b) Methylationstatus of the ITIH5 promoter in nonmalignant (MCF12A) and malignant (BT20, MCF7, SKBR3, T47D, MDA-MB-231 and Hs578-T) breast cell lines. Bands labelled with U and M indicate unmethylated and methylated alleles, respectively. (c) Expressionof ITIH5 mRNA inbreast cell lines.Cells with a methylated promoter lack mRNA expression, whereas unmethylated MCF12A and the very weakly methylated Hs578T cells exhibit strong expression. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as loading control. (d) RT–PCR showing ITIH5 mRNA expressionbefore and after incubation with 1 mM 5-aza-20-deoxycytidine (DAC) and 300 nM trichostatinA (TSA). ITIH5 mRNA expressionis inducedafter demethylating treatment. Expression of GAPDH served as a control for equal starting amounts of cDNA. (e) Semiquantitative real-time PCR of the same experiment. Grey bars represent nontreated cells, black bars indicate DAC/TSA-treated cells. Gain of ITIH5 mRNA expressionis indicated as fold-change (X) relative to each baseline expression. Relative Y-axis scaling is related to nontreated BT20 cells (set to 1). Level of induction is significant (P ¼ 0.014). Error bars were derived from triplicate experiments. ITIH, inter-a-trypsin inhibitor heavy chain.

A growing number of genes have been reported to be demethylating and histone acetylating agents led to silenced by promoter hypermethylation in breast cancer restorationof ITIH5 expression. Moreover, analysis by effectively silencing expression of their corresponding MSP showed clear ITIH5-promoter methylationin41% proteins. They fall into several broad categories of of breast tumours, matching very well with a complete fundamental cellular networks, such as cell cycle control loss of the ITIH5-proteinin42% of breast tumours or steroid receptor regulation(for a review, see Yang as analysed by immunohistochemistry. Since ITIHs et al., 2001), thus representing critical genes exerting a represent the matrix-adhesive component of ITIs, these tumour suppressive function. Few studies demonstrated findings support the fundamental role of loss of that cell adhesion molecules and proteins involved in adhesionmolecules intumorigenesis, like it has also cell–matrix interactions are also target of epigenetic beenshownfor hypermethylationof the E-cadherin gene downregulation in breast cancer (Graff et al., 1995; (Graff et al., 1995), the tissue inhibitor of metalloprotei- Bachman et al., 1999; Shao et al., 2006). Decrease incell nase 3 (Bachman et al., 1999) or the betaig-h3 gene adhesion and increase of proteolytic activity of extra- (Shao et al., 2006). cellular matrix components are thought to contribute To elucidate if the ITIH5 methylationstatus is equally largely to a tumour’s invasive phenotype. discriminatory in patient-survival estimation as the To analyse the cause of downregulation of ITIH5, we ITIH5 proteinlevel is, we performed statistical evalua- investigated the epigenetic configuration of its gene tions and found that methylation of the ITIH5 promoter promoter in breast cell lines and in primary breast was also significantly associated with reduced OS carcinomas. Expression analysis of normal and malig- (P ¼ 0.008). The biological role of ITIH5 was evaluated nant breast cell lines showed that methylation of the by a functional assay that showed a 40% reduced ITIH5-promoter was associated with absence, whereas proliferationrate in ITIH5-transfected cells as com- demethylationwas associated with presenceof ITIH5 pared to mock-transfected control cells. Taken together, mRNA. Furthermore, treatment of cell lines with these findings provide evidence that ITIH5 acts as a

Oncogene ITIH5 expression in breast cancer J Veeck et al 872

Figure 5 Methylationanalysisof the ITIH5 promoter in primary breast cancers and correlation with ITIH5 gene silencing. (a) Representative results of three matched pairs (T ¼ tumour, N ¼ normal breast epithelium) and 11 invasive breast cancer specimens are shown. Bands labelled with U represent positive results from PCR reactions amplifying unmethylated promoter sequence; bands labelled with M indicate methylated promoter sequence. Bisulphite-converted DNA from the benign breast cell line MCF12A and from in vitro methylated DNA (IVD) (Esteller et al., 1999) served as positive controls for U- and M-reactions, respectively. H2O indicates the negative control. (b) Expressionlevels of breast cancer specimens (n ¼ 39) were analysed by real-time PCR and grouped according to the strength of their promoter methylation. Comparing the median expression levels (grey bars) of each group indicates a coherence of methylation with expression loss. All samples were related to an mRNA sample derived from normal breast tissue containing >50% breast epithelium (set to 1 on X-axis scaling). (c) Kaplan–Meier survival curves displaying patient overall survival and (d) recurrence-free survival in relation to ITIH5 promoter methylation. ITIH, inter-a-trypsin inhibitor heavy chain.

Oncogene ITIH5 expression in breast cancer J Veeck et al 873 Table 4 Clinicopathological/immunohistochemical parameters in relation to ITIH5 promoter methylation and univariate analysis of factors regarding overall survival Variable Categorization ITIH5 Methylation Overall survival (OS)

na No Yes P-value n Events P-value

Clinicopathological factors: Tumour stageb pT1–2 78 23 55 0.275 65 17 0.5597 pT3–4 18 3 15 18 5 Lymph node statusb pN0 38 12 26 0.407 30 5 0.5414 pN1–3 43 10 33 37 10 Histological grade G1/G2 54 16 38 0.529 42 7 0.0374 G3 42 10 32 41 15 Immunohistochemistry (IHC): Oestrogenreceptor status Negativec (IRS 0–2) 37 7 30 0.114 31 10 0.0484 Positive (IRS 3–12) 59 20 39 50 11 Progesterone receptor Negativec (IRS 0–2) 36 7 29 0.146 30 10 0.0323 Positive (IRS 3–12) 60 20 40 51 11 ITIH5 promoter: No methylation— — — — 25 1 0.0080 Weak/strong methylation — — — 60 21

Abbreviations: IRS, immunoreactivity score; ITIH, inter-a-trypsin inhibitor heavy chain; OS, overall survival. Signiﬁcant P-values marked inbold face. aOnly female patients with primary invasive breast cancer were included. bAccording to UICC: TNM Classiﬁcation of Malignant Tumours (2002). cIRS according to Remmele and Stegner (1987).

Table 5 Cox multivariate analysis for OS and ITIH5 methylation P-value Hazard 95% conﬁdence ratio (HR) interval of HR

Lower Upper

ITIH5 0.055 7.267 0.957 55.180 G 0.073 2.582 0.916 7.284 ER 0.617 0.739 0.225 2.422 PR 0.537 0.686 0.208 2.267

Abbreviations: ER, Oestrogen receptor status; G, histological grading; ITIH, inter-a-trypsin inhibitor heavy chain; OS, overall survival; PR, progesterone receptor status. tumour-suppressor gene in normal breast tissue. Loss of ITIH5 expressionmay be involved intumour development, particularly since aberrant promoter hypermethylationof tumour suppressor genesis a well-established mechanism of tumour progression (Das and Singal, 2004; Esteller, 2005). Moreover, ITIH5 may represent a Figure 6 ITIH5 expressionreduces proliferationinthe breast class II tumour suppressor gene whose altered expres- cancer cell line MDA-MB-231. Stable clones have been generated with a full-length cDNA of ITIH5 (ITIH5 clone 4 and 8) or with sionis caused by epigenetic changes(class II) rather empty pBK-CMV vector control (Mock clone 1 and 2). An XTT thanby mutation(class I) (Sager, 1989). Class II tumour assay was performed at four subsequent time points. The baseline suppressor genes are particularly interesting drug level at 24 h for each clone was set to 1. ITIH, inter-a-trypsin targets, since reversing the block of their gene expression inhibitor heavy chain; XTT, 2,3-Bis(2-methoxy-4-nitro-5-sulfophe- could lead to tumour regression. nyl)-2H-tetrazolium-5-carboxanilide. Interestingly, the results on the expression analysis of ITIH5 in normal and malignant breast tissue are in line with those of previous studies investigating in general lung cancer mouse model, overexpression of ITI protein the role of ITIs inepithelial tumours. A role of ITIs in chains led to inhibition of tumour progression and tumour invasion has been postulated on the basis of metastatic spread (Bourguignon et al., 1999). their protease inhibitor function mediated by bikunin Inconclusion,our study indicatesthat ITIH5 may (Kobayashi et al., 1995), and their capacity to bind to represent a class II tumour suppressor gene whose loss the extracellular matrix component HA mediated by the of expressionis associated with short RFS andOS in heavy chains (Huang et al., 1993; Zhao et al., 1995). Ina breast cancer patients. In the clinically very important

Oncogene ITIH5 expression in breast cancer J Veeck et al 874 group of patients with node-negative breast cancer, (Pierce, Rockford, IL, USA) for 30 minonice. Approximately ITIH5 could be used as a predictive marker to identify 20 mg of proteinwere resolved by sodium dodecylsulphate– those patients who would not beneﬁt from systemic polyacrylamide gel electrophoresis and blotted to nitrocellu- chemotherapy and thus could be spared from its adverse lose ﬁlters. Membranes were incubated with primary antibody 1 effects. To our knowledge, this is one of only a few (polyclonal anti-ITIH5 200 ng/ml) overnight at 4 C. After washing with phosphate-buffered saline Tween-20 (PBS-T), studies published so far identifying a biomarker membranes were incubated with horseradish peroxidase potentially valuable for prognosis prediction on both (HRP)-conjugated anti-rabbit IgG (Chemicon International, the expressionlevel andthe epigenetic level. Hampshire, UK; 1:30 000) for 90 minat room temperature. Antibody detection was performed with the ‘ECL plus’ western blotting detection system (Amersham Life Science, Materials and methods Buckinghamshire, UK).

Breast cancer tissue micro array Immunocytochemical staining ITIH5 proteinexpressionwas assessed usinga TMA with 217 After fixationwith 4% paraformaldehyde, mock- andITIH5- breast cancer cases that have been described previously (Dahl transfected MDA-MB-231 cells were permeabilized using et al., 2006). The TMA contained one tissue core from 0.05% Tween. Endogenous peroxidase activity was blocked nonselected, formalin-fixed, paraffin-embedded primary breast by adding 3% H2O2 for 10 min. A polyclonal anti-ITIH5 cancer specimens diagnosed between 1994 and 2002 at the antibody (1:200) was added for 1 h. After washing twice with Institute of Pathology, University of Regensburg, Germany. PBS, cells were incubated with the 4plus Universal Immuno- Histologically, all tumours were graded according to Elston peroxidase DetectionSystem (Biocarta, Hamburg, Germany), and Ellis (1991). Clinical follow-up data were available for all followed by 10 min incubation with a streptavidin–HRP 217 breast cancer patients with a median follow-up period of complex. 3-amino,9-ethyl-carbazole (AEC) chromogen sub- 78 months (0–148 months). The Institutional Review Board of strat (RomulinAEC; Biocarta) was used for antibody the participating centres approved the study. detection. Samples were counterstained with hemalaun and examined by phase contrast microscopy (Leica TCS) (Leica, Bensheim, Germany). Cryoconserved patient samples Breast tissue samples for methylationandmRNA expression analysis were obtained from patients treated by primary ITIH5 immunohistochemistry surgery for breast cancer at the Department of Gynecology Paraffin-embedded tissue sections (2mm) were subjected to at the University Hospitals of Aachen, Jena and Regensburg, immunostaining using the Envision system (DAKO, Hamburg, Germany. All patients gave informed consent for retention and Germany), following the manufacturer’s instructions. Antigen analysis of their tissue for research purposes. Tumour material retrieval was performed by pre-treatment in citrate buffer was snap-frozen in liquid nitrogen immediately after surgery. (pH 6) ina microwave oven(30 min).The sectionswere Hematoxylin and eosin-stained sections were prepared for incubated for 1 h at room temperature with primary antibody assessment of the percentage of tumour cells, only samples (polyclonal anti-ITIH5, 1:200). Slides were incubated for 10 min with >70% tumour cells were selected. For patient character- with secondary antibody (biotinylated polylink; Biocarta). istics, see Supplementary Table 1. AEC chromogensubstrate (RomulinAEC; Biocarta) was used for antibody detection. An experienced breast cancer pathologist (E.B.) scored the immunohistochemical staining Cell lines intensity according to the scoring system suggested by All breast cell lines were obtained from the American Type Remmele and Stegner (1987). Culture Collectionandcultured as described previously (Veeck et al., 2006). For Hs578T, medium was additionally supple- Nucleic acid extraction mented with 1 mM sodium pyruvate and 10 mg/ml insulin (Sigma-Aldrich, Deisenheim, Germany). Genomic DNA from cell culture and primary invasive breast tumours were isolated using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) and RNA was extracted by use of Expression constructs and transfection TRIzol reagent (Invitrogen, Carlsbad, CA, USA), according to The coding sequence of ITIH5 was amplified from a human the manufacturers’ recommendations. normal breast tissue cDNA (Acc. No. AY238437) and cloned into pBK-CMV expression vector as described previously RT–PCR (Himmelfarb et al., 2004). For transfection experiments, cells A1mg portion of total RNA was reverse transcribed using the were grown at 60–70% confluence. Transfection was carried reverse transcription system (Promega, Madison, WI, USA). out using the jetPEI transfection reagent (Biomol, Hamburg, To improve transcription rate, we mixed oligo-dT and pdN - Germany), following the manufacturer’s instructions. (6) Primers 1:2. PCR was carried out as described previously (Veeck et al., 2006) using ITIH5 and GAPDH primers as given Generation of anti-ITIH5 antibody inSupplementaryTable 2. An ITIH5 polyclonal antiserum was generated at Eurogentec (Herstal, Belgium) by immunizing rabbits with a synthesized Semiquantitative real-time PCR peptide corresponding to amino acids 207–220 of the human Semiquantitative PCR was performed using the LightCycler ITIH5 protein. The antibody was then extracted and affinity system together with the LightCycler DNA Master SYBR purified. Green I Kit (Roche Diagnostics, Mannheim, Germany) as described elsewhere (Veeck et al., 2006). Primer sequences for Western blot analysis ITIH5 and GAPDH are listed inSupplementaryTable 2. To Protein extracts were prepared from mock-transfected and ensure experiment accuracy, all reactions were performed in ITIH5-transfected cell lines by incubation with RIPA buffer triplicates.

Oncogene ITIH5 expression in breast cancer J Veeck et al 875 Bisulphite-modification and MSP P-values were o0.05. A Mann–Whitney U-test was employed For MSP analysis, breast cell lines and cryoconserved breast to analyse differences in expression levels. A statistical cancer specimens (see Supplementary Table 1) were used. The association between clinicopathological and molecular para- DNA modification Kit (Chemicon, Ternecula, CA, USA) was meters was tested using a two-sided Fisher’s exact test. RFS applied according to the manufacturer’s recommendations. and OS were calculated according to the Kaplan–Meier For MSP (Herman et al., 1996), 1 ml of modified DNA was method. Multivariate survival analysis was calculated using a amplified using MSP primers (Supplementary Table 1), which Cox regression model including all parameters that were specifically recognized either the unmethylated (product size significantly prognostic in univariate analysis. 221 bp) or methylated (219 bp) ITIH5 promoter sequence after bisulphite conversion (Suzuki et al., 2002). For MSP reaction conditions, see Veeck et al. (2006). Normal DNA from human Conflict of interest appendix vermiformis was treated in vitro with SssI methyl- transferase (New England Biolabs, Beverly, MA, USA), to The authors have declared that no competing interests exist. generate a positive control for methylated alleles (Esteller et al., 1999).

Abbreviations DAC and TSA treatment A demethylating treatment of breast cancer cell lines was CI, confidence interval; DAC, 5-aza-20-deoxycytidine; DCIS, performed as previously described (Veeck et al., 2006). ductal carcinoma in situ; ER, oestrogenreceptor; IRS, immunoreactivity score; ITI, inter-a-trypsininhibitor;ITIH, inter-a-trypsin inhibitor heavy chain; MSP, methylation- Proliferation assay specific PCR; OS, overall survival; PR, progesterone receptor; The XTT proliferationassay from Biological Industries RFS, recurrence-free survival; TMA, tissue microarray; TSA, (Frankfurt, Germany) was used. Cells were plated and trichostatinA. cultivated ina flat 96-well plate (1 Â 103 cells/well). To each well, 100 ml of growth medium was added. Proliferationwas assayed after 24, 48, 72 and 96 h of incubation: 50 ml of XTT Acknowledgements reagent solution was added to each well and the plate was incubated for 4 h at 371C. The absorbance of the samples was We thank Sonja von Sere´ nyi, Sevim Alkaya and Inge Losen measured at 450 nm. for excellent technical assistance and Monika Klinkhammer- Schalke and Armin Pauer from the Tumor Registry Regens- burg for continuous help in obtaining clinical follow-up data. Statistical analysis This work is a research project withinthe GermanHuman For statistical evaluation, the SPSS software version 10.0 Genome Project and has been supported by the BMBF Grants (SPSS GmbH Software, Munich, Germany) was used. 01KW0401 to ED and a grant from the RWTH Aachen Differences were considered statistically significant when (START program project ITIH5).

References

Atmani F, Khan SR. (1999). Role of inter-alpha-inhibitor and its Dahl E, Sadr-Nabavi A, Klopocki E, Betz B, Grube S, Kreutzfeld R related proteins in urolithiasis. Purification of an inter-alpha- et al. (2005). Systematic identification and molecular characteriza- inhibitor related protein from the bovine kidney. Urol Res 27: 57–61. tion of genes differentially expressed in breast and ovarian cancer. BachmanKE, HermanJG, CornPG, Merlo A, Costello JF, J Pathol 205: 21–28. Cavenee WK et al. (1999). Methylation-associated silencing Das PM, Singal R. (2004). DNA methylation and cancer. J Clin Oncol of the tissue inhibitor of metalloproteinase-3 gene suggest a 22: 4632–4642. suppressor role in kidney, brain, and other human cancers. Cancer Diarra-Mehrpour M, Bourguignon J, Sesboue R, Mattei MG, Passage Res 59: 798–802. E, Salier JP et al. (1989). Humanplasma inter-alpha-trypsin Balduyck M, Albani D, Jourdain M, Mizon C, Tournoys A, Drobecq inhibitor is encoded by four genes on three chromosomes. Eur J H et al. (2000). Inflammation-induced systemic proteolysis of inter- Biochem 179: 147–154. alpha-inhibitor in plasma from patients with sepsis. J Lab Clin Med ElstonCW, Ellis IO. (1991). Pathological prognostic factors inbreast 135: 188–198. cancer. I. The value of histological grade in breast cancer: experience Bost F, Diarra-Mehrpour M, MartinJP. (1998). Inter-alpha-trypsin from a large study with long-term follow-up. Histopathology 19: inhibitor proteoglycan family—a group of proteins binding and 403–410. stabilizing the extracellular matrix. Eur J Biochem 252: 339–346. Enghild JJ, Salvesen G, Hefta SA, Thogersen IB, Rutherfurd S, Pizzo Bourguignon J, Borghi H, Sesboue R, Diarra-Mehrpour M, Bernaudin SV. (1991). Chondroitin 4-sulfate covalently cross-links the chains JF, Metayer J et al. (1999). Immunohistochemical distribution of of the humanblood proteinpre-alpha-inhibitor. J Biol Chem 266: inter-alpha-trypsin inhibitor chains in normal and malignant human 747–751. lung tissue. J Histochem Cytochem 47: 1625–1632. Esteller M, Sanchez-Cespedes M, Rosell R, Sidransky D, Baylin SB, ChenL, Mao SJ, McLeanLR, Powers RW, LarsenWJ. (1994). HermanJG. (1999). Detectionof aberrantpromoter hypermethyla- Proteins of the inter-alpha-trypsin inhibitor family stabilize the tion of tumor suppressor genes in serum DNA from non-small cell cumulus extracellular matrix through their direct binding with lung cancer patients. Cancer Res 59: 67–70. hyaluronic acid. J Biol Chem 269: 28282–28287. Esteller M. (2005). DNA methylation and cancer therapy: new Dahl E, Kristiansen G, Gottlob K, Klaman I, Ebner E, Hinzmann B developments and expectations. Curr Opin Oncol 17: 55–60. et al. (2006). Molecular profiling of laser-microdissected matched Fujita Y, Ezura Y, Emi M, Sato K, Takada D, Iino Y et al. (2004). tumor and normal breast tissue identifies karyopherin alpha2 as a Hypercholesterolemia associated with splice-junction variation of potential novel prognostic marker in breast cancer. Clin Cancer Res inter-alpha-trypsin inhibitor heavy chain 4 (ITIH4) gene. J Hum 12: 3950–3960. Genet 49: 24–28.

Oncogene ITIH5 expression in breast cancer J Veeck et al 876 Graff JR, HermanJG, Lapidus RG, Chopra H, Xu R, Jarrard DF covalent transfer of heavy chains onto hyaluronan. J Biol Chem et al. (1995). E-cadherinexpressionis silencedby DNA hyper- 280: 25674–25686. methylation in human breast and prostate carcinomas. Cancer Res Sager R. (1989). Tumor suppressor genes: the puzzle and the promise. 55: 5195–5199. Science 246: 1406–1412. HermanJG, Graff JR, MyohanenS, NelkinBD, BaylinSB. (1996). Sanggaard KW, Karring H, Valnickova Z, Thogersen IB, Enghild JJ. Methylation-specific PCR: a novel PCR assay for methylation status (2005). The TSG-6 and I alpha I interaction promotes a of CpG islands. Proc Natl Acad Sci USA 93: 9821–9826. transesterification cleaving the protein-glycosaminoglycan-protein Himmelfarb M, Klopocki E, Grube S, Staub E, KlamanI, Hinzmann (PGP) cross-link. J Biol Chem 280: 11936–11942. B et al. (2004). ITIH5, a novel member of the inter-a-trypsin Shao G, Berenguer J, Borczuk AC, Powell CA, Hei TK, Zhao Y. inhibitor heavy chain family is downregulated in breast cancer. (2006). Epigenetic inactivation of Betaig-h3 gene in human cancer Cancer Lett 204: 69–77. cells. Cancer Res 66: 4566–4573. Huang L, Yoneda M, Kimata K. (1993). A serum-derived hyaluronan- SobinLH, WittekindCH (eds) (2002). UICC: TNM Classification of associated protein(SHAP) is the heavy chainof the interalpha- Malignant Tumours 6th edn. New York: Wiley-Liss. trypsininhibitor. J Biol Chem 268: 26725–26730. Suzuki H, Gabrielson E, Chen W, Anbazhagan R, van Engeland M, Janssen U, Thomas G, Glant T, Phillips A. (2001). Expression of inter- Weijenberg MP et al. (2002). A genomic screen for genes alpha-trypsin inhibitor and tumor necrosis factor-stimulated gene 6 upregulated by demethylation and histone deacetylase inhibition in inrenalproximal tubular epithelial cells. Kidney Int 60: 126–136. humancolorectal cancer. Nat Genet 31: 141–149. JessenTE, Odum L. (2003). Role of tumour necrosisfactor stimulated Veeck J, Niederacher D, AnH, Klopocki E, WiesmannF, Betz B et al. gene 6 (TSG-6) in the coupling of inter-alpha-trypsin inhibitor to (2006). Aberrant methylation of the Wnt antagonist SFRP1 in hyaluronan in human follicular fluid. Reproduction 125: 27–31. breast cancer is associated with unfavourable prognosis. Oncogene Kobayashi H, Gotoh J, Hirashima Y, Fujie M, Sugino D, Terao T. 25: 3479–3488. (1995). Inhibitory effect of a conjugate between human urokinase Yang X, Yan L, Davidson NE. (2001). DNA methylation in breast and urinary trypsin inhibitor on tumor cell invasion in vitro. J Biol cancer. Endocr Relat Cancer 8: 115–127. Chem 270: 8361–8366. Yingsung W, Zhuo L, Morgelin M, Yoneda M, Kida D, Watanabe H Onda H, Kasuya H, Takakura K, Hori T, Imaizumi T, Takeuchi T et al. (2003). Molecular heterogeneity of the SHAP-hyaluronan et al. (1999). Identification of genes differentially expressed in canine complex. Isolation and characterization of the complex in synovial vasospastic cerebral arteries after subarachnoid hemorrhage. fluid from patients with rheumatoid arthritis. J Biol Chem 278: J Cereb Blood Flow Metab 19: 1279–1288. 32710–32718. Paris S, Sesboue R, Delpech B, Chauzy C, Thiberville L, MartinJP Yoneda M, Suzuki S, Kimata K. (1990). Hyaluronic acid associated et al. (2002). Inhibition of tumor growth and metastatic spreading with the surfaces of cultured fibroblasts is linked to a serum-derived by overexpression of inter-alpha-trypsin inhibitor family chains. 85-kDa protein. J Biol Chem 265: 5247–5257. Int J Cancer 97: 615–620. Zhang Z, Bast Jr RC, Yu Y, Li J, Sokoll LJ, Rai AJ et al. (2004). Pineiro M, Andres M, Iturralde M, Carmona S, Hirvonen J, Pyorala S Three biomarkers identified from serum proteomic analysis et al. (2004). ITIH4 (inter-alpha-trypsin inhibitor heavy chain 4) is a for the detectionof early stage ovariancancer. Cancer Res 64: new acute-phase protein isolated from cattle during experimental 5882–5890. infection. Infect Immun 72: 3777–3782. Zhao M, Yoneda M, Ohashi Y, Kurono S, Iwata H, Ohnuki Y et al. Remmele W, Stegner HE. (1987). Recommendation for uniform (1995). Evidence for the covalent binding of SHAP, heavy chains of definition of an immunoreactive score (IRS) for immunohistochem- inter-alpha-trypsin inhibitor, to hyaluronan. J Biol Chem 270: ical estrogenreceptor detection(ER-ICA) inbreast cancertissue. 26657–26663. Pathologe 8: 138–140. Zhuo L, Yoneda M, Zhao M, Yingsung W, Yoshida N, Kitagawa Y Rugg MS, Willis AC, Mukhopadhyay D, Hascall VC, Fries E, Fulop et al. (2001). Defect inSHAP-hyaluronancomplexcauses severe C et al. (2005). Characterizationof complexes formed between female infertility. A study by inactivation of the bikunin gene in TSG-6 and inter-alpha-inhibitor that act as intermediates in the mice. J Biol Chem 276: 7693–7696.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene