Tissue inhibitor of metalloproteinases-1 (TIMP-1) mRNA is elevated in advanced stages of thyroid carcinoma

Tumour cell invasion and metastasis is a multistep process that involves the degradation of extracellular matrix proteins by matrix metalloproteinases (MMPs). Tissue inhibitors of metalloproteinases (TIMPs) act as negative regulators of MMPs and thus prevent tumour cell invasion and metastasis by preserving extracellular matrix (ECM) integrity. In the present study we examined the expression of one member of TIMPs, TIMP-1, in 39 thyroid tumour specimens and two thyroid carcinoma cell lines (NPA and SW579). We also investigated the effect of high TIMP-1 expression on the invasive potential of NPA cells. Northern blot analysis showed that TIMP-1 mRNA levels correlated directly with tumour aggressiveness: the highest number of TIMP-1 transcripts was found in stages III and IV vs benign goitre (P < 0.0001). However, TIMP-1 expression was not increased in NPA and SW579 cells, both of which are derived from poorly differentiated thyroid tumours. Immunohistochemical study showed strong TIMP-1 staining in the stroma cells of advanced stages of carcinomas. Overexpression of TIMP-1 by gene transfer resulted in a significant suppression of the malignant phenotype of NPA cells as judged by an in vitro tumour invasion assay. These results suggest that high levels of TIMP-1 transcripts in advanced stages of thyroid carcinoma likely come from stroma rather than thyroid cancer cells, and TIMP-1 may function as a thyroid tumour invasion/metastasis suppressor. © 1999 Cancer Research Campaign

It is not known whether TIMP-1 plays a role in thyroid tumour invasion and metastasis. In the present study, we therefore investigated TIMP-1 gene expression in 39 primary thyroid tumours to see whether there is a correlation between TIMP-1 expression and the aggressiveness of the disease. We also transfected human TIMP-1 cDNA into a papillary thyroid carcinoma cell line (NPA) to study the effect of TIMP-1 expression on its invasive potential using an in vitro tumour invasion assay.

MATERIALS AND METHODS
All tumour specimens were obtained at surgery and were immediately frozen in liquid nitrogen and stored at Ð70°C until processed. The clinical staging of thyroid tumours was based on the TNM classification introduced in 1987 by the International Union Tissue inhibitor of metalloproteinases-1 (TIMP-1) mRNA is elevated in advanced stages of thyroid carcinoma Against Cancer (Hermanek and Sobin, 1987). Briefly, for patients 45 years and over: stage 1 Ð microcarcinoma (< or = 1 cm) without lymph node involvement; stage 2 Ð tumours (> 1 cm) confined to the thyroid without lymph node involvement; stage 3 Ð any T-stage with lymph node involvement or tumours with invasion of perithyroidal tissue; stage 4 Ð distant metastasis and anaplastic tumours; for patients under 45 years: stage 1 Ð any T-stage with or without lymph node involvement; stage 2 Ð distant metastasis. Six multinodular goitres (adenomatous hyperplasia), 28 papillary carcinomas, two follicular carcinomas and three anaplastic carcinomas were studied.
The human papillary thyroid carcinoma cell line, NPA, was kindly provided by Dr James A Fagin, Division of Endocrinology and Metablism, University of Cincinnati Medical Center, Cincinnati, OH, USA, and the squamous thyroid carcinoma cell line, SW579, was obtained from ATCC (Rockville, MD, USA).

RNA extraction and Northern hybridization
Total RNA was extracted from thyroid tumour samples and cell lines by the conventional guanidinium thiocyanateÐphenolÐ chloroform method (Chomczynski and Sacchi, 1987). Twenty micrograms of total RNA were fractionated on a 1% agarose gel containing 2.2 M formaldehyde and blotted onto a nylon membrane (Hybond-N, Amersham) by capillary transfer. The accuracy of RNA loading was monitored by ethidium bromide staining of ribosomal RNA and later by hybridization to an oligoprobe for 18S ribosomal RNA as previously described (Shi et al, 1991). The TIMP-1 cDNA probe was labelled with [α-32 P]dCTP to a specific activity of 10 9 cpm/µg using PharmaciaÕs random primer labelling kit. Hybridization was performed at 42°C for 18 h in 6 × SSPE, 10 mM EDTA, 5 × DenhardtÕs solution, 0.5% sodium dodecyl sulphate, 100 µg ml Ð1 denatured salmon testis DNA, and 50% formamide. The membranes were then washed twice over 15 min in 2 × SSPE at 65°C and exposed to Kodak XAR-5 film at Ð70°C with intensifying screens.
Following autoradiography, band intensity of TIMP-1 mRNA was quantitated by a scanning densitometer (Bio-Rad model GS-670) and normalized by comparison with the intensity of the 18S ribosomal RNA band.

Immunohistochemical procedures
Tumours were frozen and stored at Ð70°C prior to embedding in OCT compound. Sections (5 µm) cut from the frozen tissue blocks were fixed in cold acetone for 10 min before staining. Slides were then incubated with TIMP-1 monoclonal antibody (2 µg ml Ð1 in phosphate-buffered saline) for 30 min at room temperature and stained using mouse UniTect immunoperoxidase staining kit according to the manufacturerÕs instructions (Oncogene Science Inc., Cambridge, MA, USA). Mouse IgG from the kit was used as negative control.

Construction of pTIMP-1/CMV vector
The pTIMP-1/TA was digested with Hind III and Xba I restriction endonucleases to obtain TIMP-1 cDNA. The resulting fragment is oriented such that the 5′-end of the TIMP-1 cDNA is close to the Hind III site. The Hind III/Xba I fragment was then ligated into Hind III/Xba I sites of pRc/CMV, a eukaryotic expression vector from Invitrogen Co. (San Diego, CA, USA). The construct (pTIMP-1/CMV) was verified by DNA sequencing.

Transfection assay
NPA cells were cultured in F-12 medium with 10% fetal calf serum, penicillin (100 U ml Ð1 ), streptomycin (100 µg m Ð1 l), and fungizone (25 µg ml Ð1 ) at 37°C in a humidified atmosphere containing 5% carbon dioxide. Transfection was initiated when the NPA cell culture was 70% confluent. Twenty micrograms of pTIMP-1/CMV were transfected into the cells using calcium phosphate precipitation as described previously (Chen and Okayama, 1987). Seventy-two hours after transfection, 400 µg ml Ð1 G418 (Geneticin; Life Technologies Inc., Grand Island, NY, USA) were added and the cells were cultured for 3 weeks to select stable transfectants. Seven out of 200 surviving clones were characterized by Northern blot analysis and two clones were selected for use in a tumour invasion assay: TP-1 (high TIMP-1 expression), and TP-5 (low TIMP-1 expression). NPA cells transfected with the vector (pRc/CMV) alone were used as a control.

In vitro tumour invasion assay
The invasion assay was performed to quantitate the relative degree of invasiveness of pRc/CMV.NPA (control), TP-1, and TP-5 thyroid carcinoma cells, as described previously (Parhar et al, 1995). Briefly, 3 µm pore size filters in the so-called Transwell insert (Costar, Cambridge, MA, USA) were coated with 25 µl of basement membrane matrigel (1 mg ml Ð1 ) (collaborative research, Bedford, MA, USA) and dried under a hood. The Transwell insert was then placed in a Costar 24-well cluster plate. Sub-confluent cultures were labelled for 30 h with 1 µCi ml Ð1 125 I-deoxyuridine (74 TBq mmol Ð1 , Amersham, Aylesbury, UK). The cells were trypsinized, washed and resuspended in complete F-12 medium. Viable cells (2.5 × 10 5 ) were added to the upper compartment of the chamber in a total volume of 0.5 ml medium. The lower compartment of the chamber was filled with 0.5 ml of medium. After incubation for 96 h at 37°C in a humidified atmosphere containing 5% carbon dioxide, tumour cells that had migrated to the lower surface of the filters were recovered with trypsinÐEDTA and counted in a gamma counter (1272 Clini gamma, Turku, Finland). All experiments were performed in triplicate and repeated three times.

Statistical analysis
The significance of difference between means was analysed by the unpaired StudentÕs t-test (two-tailed test). A value of P < 0.05 was considered significant.

TIMP-1 gene expression in thyroid tumour tissue specimens
The abundance of TIMP-1 mRNA was examined in tissues from six multinodular goitres and 33 malignant thyroid tumours. TIMP-1 was expressed in all the tissues studied. Figure 1 shows Northern blot hybridization results for a representative panel of tumour specimens. The level of TIMP-1 expression was quantitated by a densitometer and compared among tumours of different stages ( Figure 2 and Table 1). Multinodular goitres have a value of 2.60 ± 0.34 (mean ± SEM); stage I differentiated thyroid carcinomas, 4.27 ± 0.62; stage II, 3.95 ± 1.18; stage III, 6.49 ± 0.73; and stage IV, 7.40 ± 0.27. Anaplastic carcinomas have a value of 7.50 ± 0.44. Therefore, TIMP-1 expression started to increase in the early stage of thyroid carcinoma (stage I) as compared with that in benign multinodular goitres (t = 2.56, P < 0.05) ( Figure 1 and Table 1). A significant increase in TIMP-1 expression was observed in samples with advanced stages of thyroid carcinoma (stages III and IV) as compared to multinodular goitres (t = 6.91, P < 0.0001) or stage I tumours (t = 2.75, P < 0.05). We also analysed TIMP-1 expression in both NPA and SW579 cell lines which are derived from poorly differentiated thyroid carcinomas. Interestingly, TIMP-1 expression was not increased in either NPA or SW579 cells as compared to multinodular goitres (Table 1). The level of TIMP-1 expression and relevant clinical data on the patients studied are summarized in Table 1.
In order to address the question of whether tumour cells or stroma cells are the source of the increased TIMP-1 expression, we performed immunohistochemical analysis of seven papillary carcinoma specimens with increased TIMP-1 mRNA levels. All of them demonstrated strong TIMP-1 immunostaining in the stroma cells with no, or weak, tumour cell staining. A representative immunohistochemical staining is shown in Figure 3.

Effects of TIMP-1 overexpression on the invasiveness of NPA cells
The presence of high levels of TIMP-1 expression in advanced stages of thyroid carcinoma prompted us to ask whether increased   Table 1 (  TIMP-1 expression plays a role in counteracting the invasive behaviour of thyroid carcinoma. Gene transfer was used to stably express human TIMP-1 cDNA in a papillary thyroid carcinoma cell line, NPA. Two clones were selected for use in the in vitro invasion assay: one expressing high (TP-1), and the other low (TP-5) levels of exogenous TIMP-1 transcripts ( Figure 4). As shown in Figure 5, high TIMP-1 expression decreased the invasiveness of NPA cells. As compared to the control, the invasive potential of TP-1 and TP-5 cells was reduced by 53% (t = 7.48, P < 0.0001) and 35% (t = 4.86, P < 0.001), respectively. Therefore, higher TIMP-1 expression was associated with increased inhibition of the invasive potential of NPA cells (TP-1 vs TP-5, t = 2.23, P < 0.05).

DISCUSSION
The data presented herein show that TIMP-1 is expressed in both benign and malignant thyroid tumours, and that higher TIMP-1 expression is associated with advanced disease stages (stages III and IV). The findings seem to be contradictory to the proposed role of TIMP-1 as a tumour metastasis suppressor: if TIMP-1 acts as a functional inhibitor of tumour invasion and metastasis, its expression should inversely correlate with the aggressiveness of thyroid tumours. Indeed, such an inverse association has been documented in several (Albini et al, 1991;DeClerck et al 1991DeClerck et al , 1992Ponto et al, 1991;Khoka et al, 1992Khoka et al, , 1994Montgomery et al, 1994) but not all (Kussakowska et al, 1991;Lu et al, 1991;Visscher et al, 1994aVisscher et al, , 1994bNakana et al, 1995;Zeng et al, 1995;Grignon et al, 1996) human and mouse tumours. One possible explanation for increased TIMP-1 expression in advanced stages of thyroid carcinoma is that it may represent a secondary event, i.e.
invasive tumour cells may produce MMPs, which may induce surrounding stroma cells to produce TIMP-1 to contain tumour cells that would be otherwise even more invasive. Although the number of TIMP-1 transcripts is elevated, it may not be effective in both quantity or possibly quality (enzymatic activity) in counteracting MMPs produced by thyroid tumour cells. The high level of TIMP-1 may, therefore, reflect stroma response to elevated production of MMPs, and the imbalance between TIMP-1 and MMPs activities would result in ECM destruction and tumour invasion. A recent study by Soloway et al (1998) using targeted mutagenesis of TIMP-1 further revealed that tumour invasion and metastasis were influenced by TIMP-1 of the tumour and not of the host. Thus, it is likely that TIMP-1 level within tumour cells rather than in stroma cells determines their invasive behaviour. The strong TIMP-1 immunostaining in stroma cells suggests that the high levels of TIMP-1 transcripts in advanced stages of thyroid carcinoma would likely represent a stroma response to tumour cell invasion. This is supported by the low TIMP-1 expression in both NPA and SW579 thyroid cancer cell lines that are derived from poorly differentiated thyroid carcinomas without stroma contamination. Several studies have demonstrated that increased TIMP-1 or TIMP-2 expression in advanced stages of malignancy derives predominantly from stroma cells rather than from cancer cells themselves (Kossakowska et al, 1991;Visscher et al, 1994b;Zeng et al, 1995;Grignon et al, 1996). In the study of elevated TIMP-1 mRNA levels in non-HodgkinÕs lymphoma, Kossakowska et al (1991) showed by in situ hybridization that TIMP-1 transcripts were not localized to the malignant lymphocytes, but instead were present within stromal cells which account for only 20% of the total cells in each sample. Similar results were observed by Zeng et al (1995) in colorectal cancer. These investigators found that TIMP-1 mRNA was present predominantly in tumour stroma within spindle fibroblast-like cells. In invasive bladder and breast cancers, elevated TIMP-2 was mostly localized by immunohistochemistry in the tumour stroma around nests of invasive tumour cells (Visscher et al, 1994b;Grignon et al, 1996). In order to determine whether TIMP-1 suppresses thyroid tumour cell invasion through basement membranes, we transfected NPA cells with human TIMP-1 cDNA and isolated clones with high and low TIMP-1 expression. Their invasive potentials were assessed by an in vitro tumour invasion assay in which reconstituted basement membranes (Matrigel¨) were used. One major advantage of the system is that it allows quantitative analysis of the invasive activity of tumour cells, even though Matrigel¨lacks the structural organization of the intact basement membranes (Albini et al, 1987;Hendrix et al, 1987). The assay measures the capability of a given tumour cell to attach to, degrade, and finally pass through, the matrix. These events are considered to be important steps in tumour cell invasion through basement membranes in vivo. In the present study, we demonstrated that NPA cells with high TIMP-1 expression were less invasive than those with low TIMP-1 level. The data suggest that TIMP-1 may indeed play an inhibitory role in thyroid tumour invasion and metastasis. Whether the expression of TIMP-1 from tumour vs stroma cells makes a difference in this regard is not known and requires further studies.
In summary, we have found elevated TIMP-1 mRNA in advanced stages of thyroid carcinoma. The increased levels of TIMP-1 transcripts likely come from stroma cells as an attempt to counteract tumour invasion and metastasis. NPA cells expressing high TIMP-1 transcripts have significantly less invasive potential in vitro than those with low TIMP-1 expression, supporting the role of TIMP-1 as a metastasis suppressor.