JOURNAL TRANSCRIPT

---

Miyamoto T et al. (Revised) Page 1

Original contribution

Laser captured microdissection-microarray analysis of the genes involved in endometrial carcinogenesis: stepwise up-regulation of lipocalin2 expression in normal and neoplastic endometria, and its functional relevance

Running title: Endometrial carcinoma and lipocalin2

Tsutomu Miyamoto MD, PhD a, Hiroyasu Kashima MD, PhD a, Akihisa Suzuki MD a, Norihiko Kikuchi MD, PhD a, Ikuo Konishi MD, PhD b, Naohiko Seki PhD c, Tanri Shiozawa MD, PhD a,*

a

Department of Obstetrics and Gynecology, Shinshu University School of Medicine, 3-1-1

Asahi, Matsumoto 390-8621, Japan b

Department of Gynecology and Obstetrics, Kyoto University Faculty of Medicine,

Yoshida-Konoe-cho,Sakyo-ku, Kyoto 606-8501, Japan c

Department of Functional Genomics, Chiba University Graduate School of Medicine, 1-8-1

Inohana, Chuou-ku, Chiba 260-8677, Japan

* Corresponding author. Professor Tanri Shiozawa, Department of Obstetrics and Gynecology, Shinshu University School of Medicine, 3-1-1 Asahi, Matsumoto 390-8621, Japan

Tel: 81-263-37-2718

Fax: 81-263-34-0944

e-mail: [email protected]

Key words: endometrium, endometrial carcinoma, lipocalin2, microarray, laser-captured microdissection 1

Miyamoto T et al. (Revised) Page 2

Abstract Background: Endometrial carcinoma often arises from normal endometrial glandular cells via a precursor, atypical endometrial hyperplasia. However, the genetic changes involved in this carcinogenetic process are not fully understood. Methods: Differentially expressed genes were selected from glandular cells of normal proliferative phase endometria, atypical endometrial hyperplasia and endometrial carcinoma using laser-captured microdissection (LCM) and microarray. Results: The microarray analysis revealed a total of 51 genes to be up-regulated, and 23 genes to be down-regulated in neoplastic endometrial epithelia. We focused on lipocalin2 (LCN2), which showed the largest magnitude of up-regulation. Immunostaining for lipocalin2 confirmed a stepwise increase in its expression in endometrial hyperplasia and carcinoma. In addition, elevated expression of lipocalin2 was correlated with the poor outcome of endometrial carcinoma patients. The subcellular distribution of lipocalin2 was both cytoplasmic and nuclear, despite reports that lipocalin2 is a secretory protein. Treatment of endometrial carcinoma cells with 5-azacytidine increased the expression of lipocalin2, suggesting the expression to be controlled by methylation of the promoter. The forced expression of lipocalin2 resulted in the enhanced cell proliferation and invasion in vitro. Conclusions: The expression of lipocalin2 increased with the endometrial carcinogenesis, and accumulation of the protein conferred biological aggressiveness to endometrial carcinoma cells. These results suggest lipocalin2 to be a novel target in the treatment of endometrial carcinoma.

2

Miyamoto T et al. (Revised) Page 3

Introduction Uterine endometrial carcinoma is one of the most common malignancies in the female genital tract (1), accounting for approximately 25% of all deaths ascribed to cancer of the female genital tract in developed countries (2). The number of patients with this tumor has been increasing rapidly in Japan (3). Thus, further understanding of this malignancy in terms of the carcinogenetic process and biological characteristics is important for better management of this disease. Endometrial carcinoma is clinicopathologically classified into two conceptional subgroups; type 1 and type 2 (1). The type 1 tumor is dominant, making up approximately 80% of all cases, and thought to arise from normal endometrial glandular cells via a precursor, atypical endometrial hyperplasia. Previous papers have reported the accumulation of genetic abnormalities, such as mutations of the PTEN, K-Ras, and p53 genes and microsatellite instability, to be involved in this tumorigenetic cascade (4). However, the percentage of endometrial carcinomas in which such genetic abnormalities have been identified is reportedly 50-60%, therefore, the involvement of other unknown genetic abnormalities has been presumed. Difficulty in the detection of genetic abnormalities in endometrial carcinogenesis, especially those involved in the early stages of the hyperplasia-carcinoma sequence, is partly attributed to the histological structure of the normal endometrium. Normal endometrial glandular cells, from which endometrial carcinomas develop, are surrounded by ample stromal cells. The early step of neoplastic change is believed to occur in limited loci in endometrial glands as observed in endometrial hyperplasia. Therefore, if DNA is extracted from a tissue block, subtle changes may be missed due to the intermingling of the intervening stroma and normal glands. In this regard, the precise collection of tissue samples from neoplastic glands is mandatory to analyze the early and subtle genetic changes. To solve this

3

Miyamoto T et al. (Revised) Page 4

problem, we employed the laser captured microdissection (LCM) technique, which enabled us to obtain histologically specific cells under microscopic observation (5). Using this technique, we exclusively collected normal endometrial glandular cells, hyperplastic cells and carcinoma cells from tissue sections of the same patients to abrogate the effect of gene polymorphism. In addition, to detect the genes responsible for the neoplastic process, differences in gene expression among the three cell types were investigated using a microarray. Accordingly, a group of genes whose expression was up-regulated stepwise as the disease progressed was identified. Among them, we focused on lipocalin2 (LCN2), because it had the largest magnitude of up-regulation in the neoplastic tissues. Lipocalin2, a small protein of 25kD, reportedly functions as a mediator of inflammation and iron ion transport (6~8). In this study, we examined the expression and possible functions of lipocalin2 in endometrial tissues.

Materials & Methods Fresh tissue samples and laser-captured microdissection (LCM) Immediately after the uterus was removed, fresh tissue of normal and neoplastic parts of approximately one hundred endometrial carcinomas was extirpated. The normal part was obtained from a macroscopically “flat” area with underlying myometrium and neoplastic part was mainly from the border of the tumor. These tissues were embedded in OCT compound, and flash frozen and stored at -80oC prior to use. The histological diagnosis of these samples was made using hematoxylin and eosin (H &E)-stained frozen sections. Regarding the diagnostic criteria, endometrial hyperplasia was defined by marked glandular crowding (back to back structure) with cellular atypia such as round and enlarged nucleus, granular chromatin pattern and nucleoli. Endometrial carcinoma was defined by apparent stromal invasion as shown by multiple cribriform patterns. Cases lacking these findings were excluded.

4

Miyamoto T et al. (Revised) Page 5

Consequently, three cases that simultaneously contained normal proliferative, hyperplastic and cancerous areas were identified. All three carcinomas were endometrioid adenocarcinoma, grade 1. Likewise, there were only 3 cases that included normal proliferative and hyperplastic sites in the same patient. From these 6 cases, ten to thirty serial sections (10μm-thick for LCM, 5μm-thick for H & E) of each frozen tissue sample were cut using a Cryostat (Sakura Seiki, Tokyo, Japan), fixed in 70% ethanol. One of 5 sections was stained with H&E for reference, and the remainders were stained hematoxylin and used for LCM. Normal, hyperplastic and malignant glands were precisely collected using LCM (LM100, Olympus, Tokyo, Japan). At least 20,000 laser shots were performed to obtain the respective target tissue sample. Each fresh tissue sample was used with the approval of the Ethics Committee of Shinshu University, after obtaining written consent from the patients.

Microarray analysis Total RNA was extracted from the tissues obtained by LCM, using an RNeasy Micro Kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions, and then a quality of it was checked. The total RNA was then subjected to two rounds of T7-based aRNA amplification and aminoallyle labeling using an Amino Allyl MessageAmp aRNA Kit (Ambion, Austin, TX) according to the instructions. The amplified RNA (aRNA) was coupled with Cye-Dye. Cy5-coupled aRNA of neoplastic cells and cy3-coupled aRNA of normal glandular cells of the same patients were mixed and hybridized with AceGene Human oligo chip array 30k, subset A (Hitachi software, Tokyo, Japan) , containing 10,368 probe sets for 9881 genes, according to the instructions. The hybridized arrays were scanned using a Packard GSI Lumonics Scan Array 4000 (PerkinElmer, Boston, MA) and the results were quantitatively analyzed using QuantArray software (GSI Lumonics, Unterschleissheim, Germany). Data from each microarray were normalized using LOWESS normalization. We 5

Miyamoto T et al. (Revised) Page 6

selected genes which were expressed greater than 1.5-fold or less than 0.67-fold in neoplastic tissues compared with corresponding normal tissues.

Validation of gene expression using real time-quantitative PCR Fresh frozen tissues from four samples each of normal proliferative endometrial glands, atypical endometrial hyperplasia and grade 1 endometrial carcinoma were subjected to analysis. Total RNA was extracted from the tissues using LCM and an RNeasy Micro Kit and treated with DNase I. cDNA was synthesized from each total RNA sample using PrimeScript RT reagent (Takara Bio Inc. Otsu, Japan). Reactions were carried out with the SYBR Green method. A SYBR Green PCR master mix (Takara Bio Inc.) was used in 20-μl reaction mixtures set up in triplicate with each primer. Reactions were run at 95°C for 10 seconds followed by 45 cycles of 95 °C for 3 seconds and 62 °C for 25 seconds. A dissociation curve was then used to ensure that the fluorescence signal was not derived from the formation of primer-dimer. The threshold number of cycles was determined. Gene expression was quantified by the comparative Ct method using β-actin (ACTB) as the internal control. Primer sets used for real time RT-PCR are summarized in Table 1 (Table 1).

Immunohistochemistry Formalin-fixed and paraffin-embedded tissue specimens of the endometrium obtained by hysterectomy or biopsy were selected from the pathology files of Shinshu University Hospital, and used for immunohistochemistry. One hundred thirty-one cases were endometrioid adenocarcinoma treated between 1987 and 2004 with known age (from 29 to 90 years of age, median 57.3), stage (80: stage I, 12: stage II, 25: stage III, 14: stage IV), histological grade (80: grade 1, 25: grade 2, 26: grade 3), and follow-up survival data. Ten cases were atypical endometrial hyperplasia. Forty cases were normal endometrium (30: proliferative phase, 10:

6

Miyamoto T et al. (Revised) Page 7

secretory phase), obtained by hysterectomy for uterine myoma. Each tissue sample was used with the approval of the Ethics Committee of Shinshu University, after obtaining written consent from the patients. Indirect immunohistochemical staining was performed using a goat-polyclonal anti-LCN2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and Histofine SAB-PO detector kit (Nichirei, Tokyo, Japan) with microwave pretreatment as described previously (9). Granulocytes in the same tissue section were used as a positive control. Addition of the blocking peptide (Santa Cruz Biotechnology) prepared by the manufacturer diminished the staining. The immunoreactivity of normal or neoplastic endometrial glandular cells was semi-quantitatively evaluated according to the percentage of positive cells among 500 cells in 5 high power fields by two independent reviewers (T.M. and T.S.), and these results were described as a positivity index (PI), with a maximal score of 100. The significance of differences in PI was examined using the Kruskal-Wallis rank test and Sheffe's test. A P value of less than 0.05 was considered significant. Cumulative survival was also analyzed using the Kaplan-Meier method. The log-rank test was used to evaluate the significance of lipocalin2 for survival. These analyses were made using the SPSS Statistics system (SPSS Inc., Chicago, IL). To confirm the subcellular localization, immunohistochemical expression of lipocalin2 was re-examined using rat-monoclonal anti-human lipocalin2 antibody (R & D systems, Minneapolis, MN) in additional 40 cases of endometrial carcinoma.

Immunofluorescent staining Immunofluorescent staining for cultured cells was performed as described previously (10), using a goat polyclonal anti-lipocalin2 antibody (x 50) and a Fluorescein Isothiocyanate (FITC)-conjugated anti-goat IgG antibody (x 100, green, Sigma-Aldrich, Saint Louis, MO).

7

Miyamoto T et al. (Revised) Page 8

Cell culture and transfection The endometrial carcinoma cell lines Ishikawa, and HEC1A and HEC1B were gifts from Dr. H. Nishida (Kasumigaura Medical Center, Tsuchiura, Japan) and Dr. H. Kuramoto at Kitazato University (Sagamihara, Japan), respectively. HHUA was purchased from the Riken Cell Bank (Saitama, Japan) with the permission of Dr. Ishiwata at the Ishiwata Laboratory (Mito, Japan). KLE and RL95-2 were purchased from American Type Culture Collection (Rockville, MD). Ishikawa and HHUA were derived from well differentiated endometrial adenocarcinoma. HEC1A and HEC1B were from moderately differentiated endometrial adenocarcinoma. RL95-2 and KLE were from moderately differentiated adenosquamous carcinoma and poorly differentiated endometrial carcinoma, respectively. Primary cultures of normal proliferative phase endometrial glandular cells were prepared from the surgically resected endometrial tissues according to a previous study (9) with written consent from the patients. The human LCN2 expression vector, pCEP4-LCN2, was kindly provided by Dr. Kornelia Polyak (11). To establish endometrial carcinoma cells that stably express lipocalin2, Ishikawa and HEC1B cells were transfected with pCEP4-LCN2 or pCEP4 alone by lipofection according to the supplier’s instructions (Lipofectamine 2000, Invitrogen, Carlsbad, CA). At 48 hours after transfection, the medium was changed to selection medium and a stable expression clone was selected.

Western blot analysis Proteins extracted from sub-confluent cultures of endometrial carcinoma cells (Ishikawa, HHUA, HEC1A, HEC1B, KLE) and cultured normal endometrial glandular cells were subjected to a Western blot analysis as described previously (9) using goat-polyclonal

8

Miyamoto T et al. (Revised) Page 9

anti-LCN2 antibody for primary antibody. Proteins of cytoplasmic and nuclear fractions were extracted using a NE-PER nuclear and cytoplasmic extraction Kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions. Blocking was performed with 5% nonfat milk or 3% bovine serum albmin in PBS-T for 1 hour at room temperature. The membranes were blotted with primary antibody at 4oC overnight and then incubated with a peroxidase-conjugated secondary antibody. Bound antibodies were visualized using the ECL western blot detection reagent (Amersham, Piscataway, NJ). To confirm the results obtained using goat-polyclonal anti-LCN2 antibody, the expression of lipocalin2 protein was re-examined using rat-monoclonal anti-human lipocalin2 antibody in Ishikawa cells.

RT-PCR Endometrial carcinoma cell lines (Ishikawa, HHUA, HEC1A, HEC1B, KLE) and cultured normal proliferative phase endometrial glands were subjected to RT-PCR. Total RNA extracted by TRIzol reagent (Invitrogen) according to the manufacturer’s instructions and primers for lipocalin2 and β-actin (Table 1) were used for RT-PCR. In brief, 1μg of total RNA was treated with 1 U/10 µl DNase I (Life Technologies, Gaithersburg, MD). RT was performed using an RNA PCR Kit (Takara Bio Inc.). The corresponding cDNA fragments were denatured at 94°C for 3 minutes, then subjected to 28 cycles of denaturing at 94 °C for 10 seconds, annealing at 58°C for 10 seconds and extension at 72°C for 20 seconds for lipocalin2, and 24cycles for β-actin.

Effect of the inhibition of methylation on the expression of lipocalin2 Endometrial carcinoma HHUA, Ishikawa, HEC1A, HEC1B, KLE and RL-95-2 cells were treated with a methylation inhibitor, 5-azacytidine (5-aza-C, Sigma-Aldrich), at 5μM for 72 hours. Cells were harvested and RNA was extracted using TRIzol agent. The expression of 9

Miyamoto T et al. (Revised) Page 10

lipocalin2 mRNA was evaluated using a semi-quantitative RT-PCR.

WST-1 assay Ishikawa and HEC1B cells transfected with pCEP4-LCN2 or pCEP4 alone were plated at a density of 2.0 x 103 cells into 96-well plates and cultured under optimal conditions (37 oC in a 5% CO2 incubator) for measurement. The cell viability was measured by a modified 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (WST-1 assay) according to the manufacturer’s instructions in post-plated day 1 (D-1), 3 (D-3), and 5 (D-5). A450 was measured using a microplate reader (Multiskan JX, Thermo Fisher Scientific).

Matrigel invasion assay Matrigel invasion was performed according to a previous study (12). In brief, polycarbonate membranes (8.0 µm pore size) of the upper compartment of transwell culture chambers were coated with 5% Matrigel (Becton Dickinson Labware, Bedford, MA), and 1×105 Ishikawa or HEC1B cells transfected with pCEP4-LCN2 or pCEP4 alone were plated on the upper chamber. Twenty two hours later, the number of migratory cells on the lower surface was counted in arbitrarily selected 10 high power fields. Each experiment was performed in three wells and repeated three times. The statistical analysis was conducted with the Mann-Whitney U test.

Results According to the laser captured microdissection-microarray analysis, a total of 51 genes were up-regulated, and 23 genes were down-regulated in hyperplastic and cancerous endometrial epithelia compared with normal glandular cells (Fig. 1). Among them, 25 genes were selected according to the possible oncogenetic characteristics on the basis of GenBank data

10

Miyamoto T et al. (Revised) Page 11

(http://www.ncbi.nlm.nih.gov/Genbank/), and subjected to real-time RT-PCR to quantitatively ascertain the level of their expression. Accordingly, we selected nine up-regulated genes and one down-regulated gene (Fig. 2) that were expressed greater than 2-fold or less than 0.5-fold in more than three-fourth cases of hyperplasia and/or carcinoma compared with normal endometria. Of the nine up-regulated genes, lipocalin2 (LCN2) was eventually selected for further analysis, because it showed the largest magnitude of amplification in carcinoma tissues compared to normal and hyperplastic endometria.

Immunohistochemical expression of lipocalin2 protein Immunoreactivity for lipocalin2 protein using goat polyclonal antibody in formalin-fixed, paraffin-embedded sections was observed both in the cytoplasm and in the nucleus (Figs. 3a-e). Because the nuclear staining of lipocalin2 has not been reported, immunohistochemical expression of lipocalin2 was re-examined using another antibody; rat-monoclonal anti-human lipocalin2 antibody. The result also indicated the similar cytoplasmic and nuclear staining (Fig. 3f). The positivity index (PI) for cytoplasmic lipocalin2 in normal endometrial glands in the proliferative and secretory phases was 1.6 ± 2.1 and 1.6 ± 2.5 (mean ± standard deviation, SD), respectively (Fig. 3g). The cytoplasmic PI of lipocalin2 increased in a stepwise manner, and the PI of atypical endometrial hyperplasia and endometrial carcinoma was 4.7 ± 4.7 and 10.3 ± 14.1, respectively. The cytoplasmic PI of lipocalin2 in carcinoma was significantly higher than that in the proliferative phase (p=0.020, Fig. 3g). There were a few cells with nuclear staining in normal endometrium and hyperplasia, whereas clear nuclear staining was observed in 22 cases of endometrial carcinoma (Fig. 3e). The cytoplasmic PI of grade 2 tumors was significantly greater than that of grade 1(p=0.009, Fig. 3 h). The nuclear PI of grade 3 tumors and stage III-IV tumors were significantly greater than that of grade 1 11

Miyamoto T et al. (Revised) Page 12

(p=0.048) and stage I-II (p=0.002), respectively (Fig. 3h). Regarding the prognostic value of lipocalin2, endometrial carcinoma patients with elevated level of lipocalin2 protein (PI in the cytoplasm≧10, 47/131) had significantly shorter survival periods (p=0.012, Fig. 4a). Likewise, elevated nuclear staining (PI in the nucleus≧10, 22/131 ) was associated with the poor survival (p