A novel IL-10 signalling mechanism regulates TIMP-1 expression in human prostate tumour cells

We have previously reported that interleukin 10 (IL-10) signalling stimulated activation of a specific enhancer element, termed HTE-1, to promote tissue inhibitor of matrix metalloproteinase1 (TIMP-1) expression in human bone metastatic PC-3 subclone (PC-3 ML) cells. Recently, we have identified an IL-10 responsive signal molecule, termed IL-10E1, which binds the HTE-1 element and cloned the gene encoding for the 22 kDa protein. In this paper, we have examined the mechanism of IL-10/IL-10 receptor signalling in two distinct human prostate cell lines, a ‘normal’ prostate epithelial cell line, termed NPTX-1532 and highly metastatic PC-3 ML tumour cells. Signalling cascade studies revealed that IL-10 stimulated tyrosine phosphorylation of JAK1 and TYK2 receptor kinases and tyrosine phosphorylation of IL-10E1. Phosphorylation, triggered IL-10E1's rapid translocation to the nucleus by 10–30 min. Deletion analysis combined with transient transfection experiments revealed that the n-terminal domain (∼74 a.a.) of the IL-10E1 protein, the nt-nls peptide, was stimulated by IL-10 to translocate to the nucleus and induce TIMP-1 expression. Site-directed mutagenesis further showed that phosphorylation of two tyrosine moieties (Y57 and Y62) of the nt-nls peptide was required for IL-10 activation of signalling and TIMP-1 expression. The data demonstrate, for the first time, that IL-10 receptor signalling of TIMP-1 expression is regulated by tyrosine phosphorylation of a novel gene, IL-10E1, in human prostate cells.

Changes in the relative levels and molar ratios of TIMPs and the MMPs are implicated in normal and pathological processes such as cancer, which involve connective tissue remodelling or degradation (Leco et al, 1992). Currently, it is thought that differential regulation of these genes might be achieved, in part, by various cytokines and growth factors, which allow responsive elements to either cooperate with or antagonise each other to modulate the expression of these genes (Leco et al, 1992). For example, there is a multiplicity of agents found to stimulate tissue inhibitor of matrix metalloproteinase1 (TIMP-1) expression in a variety of normal and malignant cell cultures. These range from serum, TNF a, bFGF, PDGF, EGF, phorbol esters, interleukin (IL)-1, IL-1 b, Il-6, and IL-11 (MacNaul et al, 1990;Campbell et al, 1991;Carcolo et al, 1991;Wright et al, 1991;Edwards et al, 1992;Roeb et al, 1993). In addition, inhibition of TIMP-1 expression occurs in several malignant cell lines in response to TGF b, all-trans-retinoic acid, female sex hormones, and dexamethasone (Campbell et al, 1991;Wright et al, 1991).
The regulation of TIMP-1 vs MMP expression may depend on specific receptor-mediated transactivating signals and/or promoter elements. For example, IL-6 stimulated TIMP-1 synthesis, but had no effect on MMP synthesis (Lacraz et al, 1993(Lacraz et al, , 1995; TGF b simultaneously increased TIMP-1 mRNA levels while decreasing MMP-2 mRNA content in fibroblasts (Overall et al, 1989); and steroids suppressed the expression of MMPs while simultaneously stimulating TIMP-1 production in uterine cervical fibroblasts (Sato et al, 1991). Similarly, interleukin-1 and TNF a were found to stimulate TIMP-1 production while up regulating MMP synthesis in cultured chondrocytes (Chua and Chua, 1990;Lotz and Guerne, 1991;Martel-Pelletier et al, 1991). However, in fibroblasts interleukin-1 and TNF a had little effect on TIMP-1 expression (MacNaul et al, 1990). The implication is that cell-type-specific signal cascades regulate gene expression.
Studies designed to identify determinant promoter regions and related transacting factors, which control TIMP-1 and MMP expression, would greatly aid in understanding the basis for differential regulation of these proteins in specific cell types. In this regard, we have examined the mechanisms regulating TIMP-1 expression in human prostate lines derived from nonmalignant and malignant tumours. Preliminary studies of the TIMP-1 promoter revealed that TIMP-1 expression was controlled, in part, by an 18 bp enhancer element (CACGATGACTCATCACTG), termed HTE-1, located in intron 1 upstream (À847 to À865 bp) of the 5 0 ATG translational start site. Gel shift and chloramphenicol acetyl transferase (CAT) assays with promoter constructs demonstrated that the enhancer element binds a tentative regulatory protein(s), which was selectively expressed by nonmetastatic PC-3 subclones (i.e. 2 X N.I. PC-3 subclones which fail to metastasise when implanted in SCID mice) and not by the bone metastatic PC-3 subclone PC-3 ML human tumours (Wang et al, 1998). Moreover, and IL-10 independently stimulated a 7 -9-fold increase in pCAT-HTE-1-driven CAT activity and TIMP-1 expression in transfected PC-3 ML cells. In addition, we found that IL-10, IL-4, and IL-6 differentially regulate TIMP-1, MMP-2, and MMP-9 mRNA and protein expression in HPV-18 immortalised human prostate cell lines derived from low and high Gleason score cancer tissue (Stearns et al, 1995). Western blot and Northern blot analysis clearly demonstrated that IL-10, IL-6, and IL-4 all upregulated TIMP-1 expression. In contrast, IL-10 and IL-4 (but not IL-6) downregulated MMP-2 mRNA and protein levels (Stearns et al, 1995). The data suggest that cytokine upregulation of TIMP-1 and coordinate downregulation of MMP-2 expression might control the molar ratio of TIMP-1 and MMP-2 to influence the level of protease activity and invasive behaviour of malignant prostate cancer cells.
In this paper, we have extended the initial studies to investigate the role of IL-10 receptor (IL-10R) signalling in the regulation of TIMP-1 expression in nonmalignant primary human prostate NPTX-1532 cell lines and malignant human prostate PC-3 ML cell lines. A novel IL-10 responsive signal molecule, termed IL-10E1, was identified which binds the HTE-1 element of the TIMP-1 promoter. We found that IL-10 triggered tyrosine phosphorylation of the IL-10R and the IL-10E1 protein to activate translocation of the IL-10E1 to the nucleus. Deletion analysis and site-directed mutagenesis showed that IL-10 specifically triggered tyrosine phosphorylation of tyrosines Y57 and Y62 to activate IL-10E1 signalling events and stimulate TIMP-1 expression. These data demonstrate for the first time that a novel IL-10 signal cascade regulates TIMP-1 expression in human prostate lines.
Note that the PC-3 ML and NPTX-1532 cells were selected for the signalling studies as both cell lines expressed the IL-10R, but only the PC-3 ML cells expressed detectable levels (i.e. in the 10 -30 pg range) of the IL-10E1 protein. Thus, in the IL-10E1 transfection experiments, the translocation of the IL-10E1 protein (or truncated IL-10E1 fragments) to the nucleus could be examined in the presence or absence of endogenous IL-10E1.

Cell cultures
We examined IL-10E1 expression in a variety of primary prostate cell lines. The NPTX-1532 and CPTX-1532 cells (generously provided by Drs Robert Bright and Susan Topalian, NIH-NCI, Bethesda, MD, USA) are normal prostate culture developed from normal prostate tissue extracted from human patients (Bright et al, 1997). These cells were immortalised with E6 and E7 transforming proteins of human papilloma virus serotype 16 and were maintained in Keratinocyte SFM (GIBCO BRL, Grand Island, NY, USA) containing 50 mg ml À1 bovine pituitary extract, 5 ng ml À1 epidermal growth factor (EGF), and 5% foetal bovine serum (Biofluids, Rockville, MD, USA). Both MGEM and Keratinocyte SFM media contained 100 U ml À1 of penicillin G sodium and 100 mg ml À1 streptomycin sulphate. Cells were routinely passaged by mild Trypsin -EDTA (GIBCO BRL, Grand Island, NY, USA) detachment and cultured in a humidified atmosphere of 95% air and 5% CO 2 at 371C. HPV-MLC7 cells (kindly provided by Dr Donna Peehl, Stanford University, CA, USA) are benign prostatic hyperplasia culture immortalised by human papilloma virus serotype 18 (Cohen et al, 1991). The HGPIN cell line was established in our laboratory from human prostate tissue showing high-grade prostatic intraepithelial neoplasia (HGPIN) histology. The HGPIN cultures were immortalised by HPV-18 transfection according to published methods (Wang et al, 1996). The BPH and HGPIN cell lines were maintained in Keratinocyte SFM (Gibco BRL, Grand Island, NY, USA) containing 5 ng ml À1 EGF, 50 mg ml À1 bovine pituitary extract, and 5% foetal bovine serum (Biofluids, Rockville, MD, USA). All the different cell lines were transferred to their respective medium containing 10 nM DHT and 10 mM HEPES buffer, but without supplements (i.e. pituitary extracts and ITS) for 1 h prior to carrying out the IL-10 studies.
The 2 X N.I. PC-3 subclones and the bone metastasising PC-3 ML subclones derived from the parent human prostate PC-3 line were grown as previously described by our laboratory (Wang et al, 1998). In transfection experiments, the cell cultures were transiently transfected overnight and the cells exposed to IL-10 prior to harvesting and protein extraction from the cytoplasmic and nuclear compartments (Stearns et al, 1995). Unless stated otherwise, the cells were treated with IL-10 at 15 ng ml À1 for 1 h in all the experiments.
Cloning the IL-10E1 gene mRNA was isolated from PC-3 ML cells using a mRNA purification kit (PolyATtract system 1000 0 , Promega, Madison, WI). A phage vector ZAP-express vector (Stratagene, La Jolla, CA) was used to construct a cDNA library. Clones were screened three times according to Singh et al (1998) with a 32 P-labeled oligonucleotide containing three repeats of the HTE-1 sequence. Positive phages were treated by in vivo excision to form a phagemid containing the cloned insert using a protocol described by the instruction manual of ZAP-cDNA Gigapack II Gold Cloning kit (Stratagene, La Jolla CA). The IL-10E1 cDNA was cloned in a GST-vector pGEX4T2 vector. The GST -IL-10E1 fusion protein was isolated using a glutathione Sepharose 4B beads, and IL-10E1 was purified utilising thrombin protease (27-0843-01) cleavage to separate IL-10E1 and GST proteins according to the methods of the manufacturer (Pharmacia Biotech, Piscataway, NJ, USA).

Sequencing
DNA sequencing was done using ABI 377 and 373 A Stretch sequencers with Taq FS Dye Terminator or Dye Primer chemistry (DNA Sequencing Facility, University of Pennsylvania, USA). The protein sequence was deduced from the cDNA sequence using a GCG program. Peptide fragments generated from purified IL-10E1 protein were sequenced by the protein sequencing facility (University of Pennsylvania). Protein purity was verified by SDS -PAGE analysis.

Electrophoretic mobility shift assays (EMSAs)
Cytoplasmic and nuclear proteins were prepared using published protocols (Wang et al, 1998). Crude extracts from recombinant phage lysogens were prepared as described by Singh et al (1998). The DNA -protein binding reaction and EMSA were carried out as described by Singh et al (1998). The DNA binding reaction mixture contained B100 000 c.p.m. of the 32 P-end-labelled double-stranded oligonucleotide (B1 ng), plus B5 mg of crude protein extract, 1 mg of poly (dI-dC)-poly (dI-dC) (Sigma) and buffer composed of 10 mM HEPES (pH 7.9), 50 mM NaCl, 1 mM DTT, 1 mM EDTA, and 5% glycerol. Samples were eluted on 8% SDS -PAGE gels and developed by autoradiography overnight.

ELISAs
ELISAs (A 490 nm) were performed according to methods previously described (Stearns et al, 1995) using well-characterised polyclonal antibodies specific for TIMP-1 (Stearns et al, 1995). We used the TIMP-1 antibody at a dilution of 1 : 2000 (500 ng ml À1 ) and recorded the absorbance (A 490) for three different dilutions (B5, 7.5, and 10 mg ml À1 protein) of conditioned medium from NPTX-1532 cells. Cells were plated on human collagen IV (100 mg well À1 in 12 well dishes) at 1 Â 10 6 cells ml À1 for 24 h and exposed to IL-10 for 24 h prior to collecting the conditioned medium. The measurements were normalised to 10 mg ml À1 protein and the total amounts of antigen ml À1 determined from an ELISA standard curve (Stearns et al, 1995). Protein levels were measured according to the methods of Bradford (1976).

Site-directed mutagenesis
Site-directed mutagenesis and deletion analysis were carried out according to the methods described by Sambrook et al (1989). The deleted or mutated fragments were generated from the IL-10E1 cDNA using specific primers and PCR (B35 cycles amplification). The fragments were then subcloned and sequenced (DNA Sequencing Facility, University of Pennsylvania) to confirm the specific deletions and mutations.

Subcloning the IL-10E1 fragments
To generate the different IL-10E1 subclones, ZAP-cDNA synthesis, ligation of the ZAP-cDNA into the Uni-ZAP XR vector arms, and packaging of recombinant phage DNA were performed using a lambda phage vector-ZAP Express cDNA vector and protocol of the ZAP-cDNA Gigapack R II Gold Cloning Kit (Stratagene Inc., San Diego, CA, USA). The cDNA fragments cloned encoded for the following peptide fragments of the IL-10E1 gene, including: Amino acids bolded and underlined were mutated. The nuclear localisation signal (nls) was bolded.

Transient transfection experiments
NPTX-1532 or PC-3 ML cells were transfected overnight with 4 mg of recombinant phagemid using 8 mg of Lipofectin reagent according to protocol described by the manufacturer (Gibco Life Technologies Inc., Boston, MA, USA). Western blots were according to the methods of Towbin et al (1979).

Cloning of IL-10E1 gene
An IL-10 responsive signal protein, termed IL-10E1, was cloned from a phage expression cDNA library utilising a triplicate repeat of the HTE-1 sequence, an enhancer element of the TIMP-1 promoter (Wang et al, 1998). In preliminary studies, EMSAs showed that a protein present in lysogen extracts from rIL-10E1 recombinant phagemid clones binds the HTE-1 oligonucleotide ( Figure 1, lanes 1 -2). The band shift mobility was identical to that observed with the native protein isolated from human prostate PC-3 ML cells ( In control experiments, the nuclear protein extracts were obtained from PC-3 ML cells exposed to IL-10 in the presence of IL-10R antibodies (1 : 100 dilution) ( Figure 1, lane 7), or the protein extracts were preincubated with excess 'cold HTE-1 oligonucleotide' at concentrations ranging from 10 ng ml À1 for 30 min ( Figure 1, lane 8). Note that a total of five separate rIL-10E1 phagemid clones derived from three different cDNA libraries yielded a recombinant protein that bound the HTE-1 probe.

Sequencing
Sequencing of three independently derived IL-10E1 clones revealed that the IL-10E1 gene was B95% homologous with CSRP-2 and hsmLIM domain genes . IL-10E1's homology with CSRP-2 and hsmLIM was reduced to B90% in a region spanning exons 2 and 3 (i.e. B+451 to +490 bp)   7), the nuclear protein extracts were obtained from PC-3 ML cells exposed to IL-10 in the presence of IL-10R antibodies (1 : 100 dilution); or (lane 8) from IL-10-treated cells where the nuclear protein extracts were preincubated (for 30 min) with excess 'cold-HTE-1 oligonucleotide' at concentrations of 10 ng ml À1 , which competitively blocked labelled oligonucleotide binding to rIL-10E1 protein.
Novel IL-10 signalling mechanism M Wang et al . The deduced amino-acid sequences revealed that the IL-10E1 protein had an open reading frame of 192 amino acids (a.a.) (B22 kDa) (Figure 2). A classical LIM-domain sequence (CX2C17HX2CX2C17CX2C) was detected in the n-terminal region with a GY-rich box tail (bolded) and a nuclear localisation signal-(underlined) (YCKSCYGKKYGPK-GYGY).

Western blot studies
Rabbit polyclonal antibodies were generated against a 22 a.a. peptide (YCKSCYGKKYGPKGYGYGQGAG) located near the nterminal domain of the IL-10E1 protein (see diagram 1 in Materials and Methods and Figure 2). Western blot analyses on whole-cell extracts demonstrated that IL-10E1 was expressed at roughly equivalent levels in PC-3 ML and CPTX-1532 cells (Figure 3, lanes 1 -2), but was not expressed in the four different BPH, two HGPIN and the four different NPTX-1532 cell lines ( Figure 3, lanes 3 -12, respectively). Note that CPTX-1532 cells are a malignant primary prostate line derived from the same prostate as the normal NPTX-1532 cells. Western blots further showed that the IL-10E1 antibodies specifically recognised the rIL-10E1 protein (22 kDa, Figure 4A, lane 5).
Truncated peptide fragments of the IL-10E1 protein were produced for studies of the role IL-10E1 in IL-10R signalling (see Diagram 1). Western blots also showed that the IL-10E1 antibodies recognised peptide fragments of the IL-10E1 protein, which, were isolated from recombinant phagemid clones expressing truncated rnt-nls; rnt-nls mC61, and rnt-nls mY57/mY62 peptides ( Figure 4A, lanes 1 -4, respectively). Note that the samples were eluted on a 8 -15% gradient gel and the peptide fragments appear to vary in size slightly as a result of one or two amino-acid differences in the amino-acid sequence ( Figure 4A). Interestingly, the IL-10E1 antibodies recognised the rnt-nls-peptides ( Figure 4B, lane 1) and binding was specifically competed by preabsorbing the antibody with rIL-10E1 protein ( Figure 4B, lane 2). However, the antibodies failed to bind rct and rnt peptides, respectively ( Figure 4B, lanes 3 -4), which were generated from the c-terminal domain of the IL-10E1 protein and missing the 'YCKSCYGK-KYGPKGYGY' region of the rnt-nls peptide.
Western blotting studies further showed that following transfection of NPTX-1532 cells with either rnt-nls, rnt-nls mC61, or rntnls mY57/mY62 constructs ( Figure 4C, lanes 1 -3), respectively, the truncated peptides and the intact protein were expressed in crude cytoplasmic extracts of the cells.
in both cytoplasmic and nuclear protein extracts ( Figure 6, compare lanes 3, 4). In comparison, by 30 min stimulation rntnls was no longer detected in cytoplasmic extracts ( Figure 6, lanes 5 -6), but was detected in the nuclear extracts ( Figure 6, lane 7).
Identical studies were also carried out with PC-3 ML human prostate tumour cells, which normally expressed endogenous IL-10E1 protein. Following transfection with the rnt-nls construct, both nIL-10E1 and rnt-nls were expressed in cytoplasmic extracts (data not shown). Following IL-10 stimulation, the presence of both native nIL-10E1 and the rnt-nls peptide was assayed in nuclear protein extracts. Neither protein was detected in the nuclear extracts after 10 min stimulation ( Figure 7A, lane 1). However, both the intact nIL-10E1 protein and the nt-nls peptide were detected in the nuclear protein fractions after B30 -60 min ( Figure 7A, lanes 2 -3). Control experiments in Figure 7B (lanes 1 -3) showed that in the presence of IL-10R antibodies, IL-10 failed to stimulate nIL-10E1 and rnt-nls translocation to the nucleus. Both the 22 kDa nIL-10E1 protein and 8.4 kDa rnt-nls peptide remained in the cytoplasmic fractions of cells exposed to IL-10 in the presence of IL-10R antibodies for 10, 30, and 60 min ( Figure 7C, lanes 1 -3). The data indicate that the IL-10/IL-10R axis regulates IL-10E1 signalling in nonmalignant and malignant prostate lines and indicates that the nt-nls peptide may, in part, be functionally equivalent to the intact nIL-10E1 protein.

JAK/TYK kinase studies
We postulated that the IL-10 receptor might regulate the IL-10E1 signalling cascade. To test this possibility, we initially employed NPTX-1532 cells transfected with the rIL-10E1 construct. Cells were treated with IL-10 and harvested to prepare crude protein extracts for immunoprecipitation assays with excess phosphotyrosine-specific antibodies (PY20) ( Figure 8A). The immunoprecipitates were then electrophoresed on SDS -PAGE gels and blotted with JAK1-and TYK2-specific antibodies. Both antibodies blotted a B130 kDa band on the gels that corresponded to the JAK1 and TYK2 proteins (arrowhead, Figure 8A, lanes 2 and 6, respectively). In comparison, JAK3 antibodies failed to blot any detectable protein ( Figure 8A, lane 8), indicating that an IL-10-specific receptor was involved. Western blot studies further showed that in cells exposed to IL-10 in the presence of IL-10R antibodies, the PY20 antibodies failed to immunoprecipitate JAK1, TYK2, or JAK3 ( Figure 8A, lanes 1, 5, 7, respectively) from crude cell extracts, presumably as a result of the receptor antibodies blocking IL-10's induction of IL-10 receptor phosphorylation. Figure 8A, lanes 3 and 4 show that secondary antibodies alone failed to blot the 130 kDa JAK and TYK bands. Note that the lower band (double arrowhead) in lanes 2 and 6 appeared to be a nonspecific band inexplicably blotted by both JAK1 and TYK2 antibodies. Western blots with IL-10E1 antibodies ( Figure 8B, lanes 1 -8) confirmed that 22 kDa rIL-10E1 protein was uniformly expressed in the preparations blotted in Figure 8A, and indicated that differences in protein content were not a limiting factor in Figure 8A.
The immunoprecipitation studies were extended to examine whether rIL-10E1 protein in transfected NPTX-1532 cells was also phosphorylated in response to IL-10 ligand. Following IL-10 treatment for 60 min, immunoprecipitates were isolated with PY20 antibodies from nuclear protein extracts. Western blots with IL-10 E1 antibodies showed that the immunoprecipitates from IL-10treated cells contained IL-10E1 protein ( Figure 9A, lanes 1, 3). In addition, immunoprecipitates prepared with IL-10E1 antibodies were blotted with PY20 antibodies which labelled a 22 kDa protein ( Figure 9A, lanes 4, 6), plus 1 -2 associated proteins of unknown composition. In control studies, IL-10 in the presence of IL-10R antibodies failed to induce rIL-10E1 translocation to the nucleus ( Figure 9A, lanes 2, 5). That is, both PY20 and IL-10E1 antibodies failed to immunoprecipitate the 22 kDa protein from nuclear extracts of these cells, and Western blots with IL-10E1 ( Figure 9A, lane 2) and PY20 ( Figure 9A, lane 5) antibodies failed to detect the 22 kDa protein in the respective immunoprecipitates. Control experiments where the cells were incubated with IL-10 plus secondary IgG antibody showed that the secondary antibodies did not block IL-10 activation of the IL-10R. In these latter assays, the IL-10E1 protein was immunoprecipitated with both PY20 ( Figure 9A, lane 3) and IL-10E1 ( Figure 9A, lane 6) antibodies and blotted with IL-10E1 and PY20 antibodies, respectively. Western blots of whole-cell extracts of the rIL-10E1-transfected cells confirmed that rIL-10E1 protein was uniformly expressed by the NPTX-1532 cells in these experiments ( Figure 9B, lanes 1 -6).

Site-directed mutagenesis studies
In attempts to identify whether specific tyrosine groups were phosphorylated in response to IL-10 ligand, PC-3 ML cells were transfected with either rnt-nls ( Figure 10A, lanes 1 -2) or mutant rnt-nls mY57/mY62 constructs ( Figure 10A, lanes 3 -4). In rnt-nlstransfected cells, gel autoradiography of immunoprecipitates isolated with the PY 20 antibodies showed that the nuclear protein extracts from untreated cells did not contain either nIL-10E1 or rnt-nls ( Figure 10A, lane 1). Following IL-10 treatment, the nuclear extracts contained both nIL-10E1 (arrowhead) and rnt-nls peptide (double arrowhead) ( Figure 10A, lane 2). In PC-3 ML cells transfected with the rnt-nls mY57/mY62 expression vector, PY20 antibodies failed to immunoprecipitate the nt-nls mY57/mY62 peptide from nuclear extracts of untreated ( Figure 10A, lane 3) and IL-10-treated cells ( Figure 10A, lane 4). In comparison, the nIL-10E1 protein was barely detectable in nuclear extracts of untreated cells ( Figure 10A, lane 3), but was significantly elevated in nuclear extracts of IL-10-treated cells ( Figure 10A, lane 4, arrowhead). Western blots with IL-10E1 antibodies of cytoplasmic extracts verified that nIL-10E1 and rnt-nls mY57/mY62 peptides were both expressed in the cytoplasmic protein extracts of untreated PC-3 ML cells (data not shown). Western blots with IL-10E1 antibodies further showed that the immunoprecipitates prepared with PY20 antibodies from cytoplasmic and nuclear protein fractions of the IL-10-treated cells did not contain detectable levels of rnt-nls mY57/mY62 peptide, indicating that the PY20 antibodies failed to recognise nonphosphorylated protein. The nIL-10E1 protein was detected in Western blots of the PY20 antibody immunoprecipitates prepared from the nuclear extracts of IL-10-treated cells. The truncated rnt-nls mY57/mY62 peptide was not detected in these assays (data not shown), presumably because IL-10 failed to induce its translocation to the nucleus. Identical experiments were carried out with normal NPTX-1532 cells transfected with either rnt-nls or rnt-nls mY57/mY62 constructs ( Figure 10B). Gel autoradiography of immunoprecipitates prepared with PY20 antibodies showed that nuclear extracts from cells exposed to IL-10 for 120 min (i.e. NPTX-1532 cells transfected with the rnt-nls mY57/mY62 construct) did not contain the rnt-nls mY57/mY62 peptide ( Figure 10B, lane 1). In contrast, the rnt-nls peptide was detected in nuclear extracts (i.e. from NPTX-1532 cells transfected with the rnt-nls construct) following exposure of the cells to IL-10 for 120 min ( Figure 10B, lane 2). That is, the phosphorylated rnt-nls peptide was immunoprecipitated with PY20 antibodies from the nuclear extracts of these cells ( Figure 10B, lane 3). Control experiments showed that the rnt-nls peptide was not immunoprecipitated from nuclear extracts of NPTX-1532 cells (transfected with the rnt-nls construct) exposed to IL-10 in the presence of IL-10R antibodies ( Figure 10B, lane 3). Western blots with IL-10E1 antibodies of whole-cell extracts from untreated NPTX-1532 cells in Figure 8A, lanes 1 -6, respectively.
Western blots with IL-10E1 antibodies of cytoplasmic extracts from the untreated NPTX-1532 cells confirmed that both rnt-nls ( Figure 10C, lane 1) and rnt-mY57/mY62 peptides ( Figure 10C, lane 2) were expressed in the NPTX-1532 cells lines transfected with their respective constructs. The rnt-nls peptide was not detected in NPTX-1532 cells transfected with vector alone ( Figure  10C, lane 3). These data indicate that an intact rnt-nls peptide (i.e. with Y57 and Y62 tyrosines) was required for the nuclear localisation of the rnt-nls peptide and the IL-10E1. Note that the 2 -3 bands at the top of the gel in Figure 10A and B were phosphorylated proteins of unknown identity, which were uniformly precipitated with the PY20 antibodies from extracts of both untreated and IL-10-treated cells.

Regulation of TIMP-1 expression
Preliminary studies were carried out to assess if IL-10 activation of rIL-10E1 or the rnt-nls peptide induced TIMP-1 expression in transfected NPTX-1532 cells. ELISA measurements showed that NPTX-1532 cells (Figure 11, -) and mock-transfected NTPX-1532 cells (Figure 11, X) (i.e. cells transfected with vector alone) normally expressed (secreted) barely detectable levels of TIMP-1 protein in the absence or presence of increased levels of IL-10. However, following transient transfection overnight with either the IL-10E1 or the nt-nls construct, IL-10 (0 -60 ng ml À1 for 24 h) induced TIMP-1 expression in a dosage-dependent manner in these NPTX-1532 subclones (Figure 11, m'). IL-10 receptor antibodies blocked IL-10 induction of TIMP-1 expression by cells transfected with the rnt-nls construct (Figure 11, K) and the rIL-10E1 expression vector (data not shown). In comparison, NPTX-1532 subclones transfected with the rnt-mY57/mY62 construct failed to secrete TIMP-1 in response to IL-10 at increased dosages for 24 h (Figure 11, E). These experiments indicate that IL-10 activation of the rIL-10E1 protein or the truncated rnt-nls peptide stimulated TIMP-1 production. Since IL-10 failed to induce TIMP-1 expression in NPTX-1532 cells transfected with the nt-mY57/mT 62 peptide, we suggest that the intact nt-nls peptide was required for the signalling cascade and that IL-10-induced phosphorylation of specific tyrosine groups (e.g. Y57 and Y62) may regulate activation of the IL-10E1 protein and nt-nls peptide to control TIMP-1 synthesis.

DISCUSSION
In two earlier papers, we found that the IL-10 stimulated activation of the HTE-1 enhancer element of the TIMP-1 gene to activate TIMP-1 expression in malignant human prostate cell lines (Wang et al, 1998(Wang et al, , 2002. In this paper, the data significantly extended these studies and showed that IL-10 ligand binding to the IL-10  Figure 11 ELISA measurements of TIMP-1 secreted over 24 h by NPTX-1532 cells in the presence of IL-10 at concentrations ranging from 10 to 60 ng ml À1 for 24 h. NPTX-1532 cells were (À) not transfected; (x) mock-transfected with vector alone; or transfected with (m) rIL-10E1; (') rnt-nls; and (~) rnt-mY57/mY62 expression vectors. NPTX-1532 cells were transfected with (K) rnt-nls and exposed to IL-10 in the presence of IL-10R antibodies. Values represent the mean+1 s.d. from at least three experiments and triplicate measurements per experiment.
receptor activated a signal cascade, which rapidly activated TIMP-1 expression in normal and malignant prostate cell lines. That is, IL-10 binding to the IL-10R induced tyrosine phosphorylation of the JAK 1/TYK2 kinases, tyrosine phosphorylation of Y57 and Y62 tyrosines in the 'n-terminal' domain of the IL-10E1 protein, and the rapid transport to the nucleus of the IL-10E1 protein.
Immunoprecipitation assays showed that phosphorylation of Y57/Y62 tyrosines triggered redistribution of the intact IL-10E1 protein and the nt-nls fragment in both normal and malignant human prostate cell lines. The nt-nls mY57/mY62 fragments containing mutated tyrosines were not phosphorylated following IL-10 treatment of transfected cells and the nt-nls peptide failed to migrate to the nucleus in these cells. In addition, nuclear translocation of the nt-nls fragment (and presumably the intact IL-10E1 protein as well) required the nls, but not the remaining cterminal domain of the IL-10E1 protein. This indicates that the mechanism of activation of the nls region may involve changes in protein -protein interaction, or perhaps conformational changes in the IL-10E1 protein mediated by tyrosine phosphorylation of Y57 and Y62 groups. In sum, the data from the sitedirected mutagenesis and deletion analysis studies uniformly confirmed that the n-terminal domain, and specifically the region spanning about 17 a.a.'s (e.g. YCKSCYGKKYGPKGYGY), were primarily responsible for nuclear translocation of the IL-10E1 protein. We believe that this domain may possibly control binding to the HTE-1 element and activation of TIMP-1 expression. The data presented in Figure 10 clearly showed that IL-10 activation of IL-10E1 signalling serves to upregulate TIMP-1 in NPTX-1532 cells, but the precise mechanism controlling the molecular process has yet to be defined. Preliminary EMSA studies in our laboratory have shown that the nt-nls fragment binds the HTE-1 site, but the truncated n-terminal fragments lacking the YCKSCYG domain failed to bind the HTE-1 site. In addition, other EMSA studies have indicated that IL-10-induced tyrosine phosphorylation of the intact nt-nls peptide or the IL-10E1 protein promoted a 4 -5 fold increase in binding activity compared to nonphosphorylated protein and recombinant protein from phagemid expression vectors (data not shown). The significance of these observations is currently under study in hopes of understanding how binding interactions of the YCKSCYG domain with specific nucleotide sequences of the HTE-1 element serve to regulate TIMP-1 expression.
Interestingly, the sequence data reported in this paper indicate that IL-10E1 is a 'LIM-domain' protein with a 495% homology to CRP2 and hsmlim genes . Sequence comparisons from GenBank indicated that there was 100% homology in the 5 0 region spanning the first LIM domain of the CRP2 and hsmlim genes. However, the region spanning the second LIM-domain (B +451 to +490 bp)(i.e. exons 2 -3) exhibited B90% homology with CRP2 and hsmlim genes . The deduced amino-acid sequence revealed that unlike CRP2 and hsmlim proteins (Dawid et al, 1995(Dawid et al, , 1998Weiskirchen et al, 1997), which expressed two LIM domains, the IL-10E1 protein consisted of a single LIM domain at the n-terminal region of the protein. Presumably, this difference between IL-10E1 and CRP2 or hsmlim might explain why IL-10E1 is overexpressed in prostate cancer cells and functions as an IL-10-responsive signal molecule.
In brief, the LIM motif is a 'helix-loop-helix' domain that was originally identified in three developmentally regulated transcription factors, Caenorhabditis elegans Lin-11, rat Isl-1, and C. elegans Mec-3 (Freyd et al, 1990;Karlsson et al, 1990), but has since been identified in a variety of proteins. The LIM domain consists of a 52 amino-acid sequence CX 2 CX 16 -23 HX 2 CX 2 CX 2 CX 16 -21 CX 2 -3 (C/H/D) motif, which typically has been associated with DNA-binding homeodomains and kinase domains (Freyd et al, 1990;Karlsson et al, 1990;Dawid et al, 1995Dawid et al, , 1998. The LIM domain has been recognised as a distinct protein motif within LIM homeodomain proteins by both genetic and biochemical methods (Freyd et al, 1990;Karlsson et al, 1990;Way and Chalfie, 1994). The LIM domain has been found to mediate 'protein -protein' interactions of cytoplasmic and nuclear proteins. NMR analysis of the solution structure of the c-terminal LIM domain of two of these proteins, CRP1 (Perez-Alvarado et al, 1994) and CRP2 (Konrat et al, 1997;Weiskirchen et al, 1997;Wadman et al, 1994), has shown that two zinc-fingers constitute a LIM domain, which represent two separated structural entities held together in a distinct configuration by hydrophobic interactions (Dawid et al, 1998). Moreover, the c-terminal LIM domain in CRP1 and CRP2 has been found to closely resemble the zinc-finger in the DNA-binding domain of GATA1, suggesting that the LIM domain might specifically bind DNA (Weiskirchen et al, 1995).
There is a potpourri of evidence indicating that LIM homeodomain proteins or LHX proteins function as transcription factors and genetic evidence has supported this idea (Dawid et al, 1998). Several studies have shown that LIM-homeodomain proteins can activate the expression of specific genes (i.e. insulin) or direct cellular differentiation in cooperation with other nuclear factors (Perez-Alvarado et al, 1994). Other examples include, E47/Pan-1 and Lmx1.1 transcription factors, which have been shown to interact with LMO2, a LIM only protein, and with TAL1, LYL1, and LYL2 to regulate gene activation (Sadler et al, 1992;Warren et al, 1994;Larson et al, 1996;Johnson et al, 1997;Dawid et al, 1998). Similarly, the rat cysteine-rich intestinal protein (CRIP), the human cysteine-rich protein (CRP), and rhombotin-1 and -2 have been implicated as protein crosslinkers in transcription regulation (Dawid et al, 1998). The IL-10E1 protein belongs to the LIM only homeodomain family of proteins, which bind specific cognate DNA elements to regulate gene expression (Dawid et al, 1998). Part of the LIM domain, the YCKSCYG region, appears to directly mediate binding, but the entire LIM domain and the 'helix -loop -helix' conformation of the 52 a.a. LIM domain is probably important for specific binding to the TIMP-1 promoter and the stimulation of TIMP-1 expression.
In conclusion, we have shown for the first time that IL-10 signalling in prostate cell lines involves a signal pathway distinct from the STAT 1/STAT 3 pathway identified in white blood cell lines. The signal cascade activates a novel LIM only protein, termed IL-10E1, which is phosphorylated at two specific tyrosine groups, Y57/Y62, to activate signalling and binding to the TIMP-1 promoter. As a consequence, TIMP-1 expression is upregulated. Since TIMP-1 overexpression has been associated with the blockade of metastasis in human prostate tumour xenograph model studies (Stearns et al, 1999), we suggest that IL-10 may indirectly prevent tumour metastasis by this mechanism. IL-10 does not appear to alter IL-10E1 levels since ELISA measurements indicate that PC-3 ML cells express pictogram quantities (10 -30 pg ml À1 ) independent of IL-10. Also, in NPTX-1532 cells which do not normally express detectable IL-10E1, IL-10 failed to induce expression.
The preliminary studies further indicate that the IL-10/IL-10R axis and activation of IL-10E1 stimulates the upregulation of TIMP-1 production in NPTX-1532 cells. Thus, the IL-10/IL-10R axis is of import in controlling tumour metastasis and the development of therapeutic approaches which target this pathway is ongoing in our lab. Recently, green fluorescent protein has been employed to tag regulatory proteins and analyse the mechanisms controlling nuclear translocation in real-time. For example, Hagting et al (1999) showed in real-time studies that the translocation of fluorescently tagged wild-type (wt) cyclin B1 was controlled by phosporylation. In these studies, we will utilise cells transfected with wt and mutant IL-10E1 proteins tagged with green fluorescent protein to identify (in real time) therapeutic agents that specifically block signalling in response to specific ligands (IL-10, IL-4, INF).