The propeptide domain of lysyl oxidase induces phenotypic reversion of ras-transformed cells.

Lysyl oxidase is an extracellular enzyme critical for the normal biosynthesis of collagens and elastin. In addition, lysyl oxidase reverts ras-mediated transformation, and lysyl oxidase expression is down-regulated in human cancers. Since suramin inhibits growth factor signaling pathways and induces lysyl oxidase in ras-transformed NIH3T3 cells (RS485 cells), we sought to investigate the effects of suramin on the phenotype of transformed cells and the role of lysyl oxidase in mediating these effects. Suramin treatment resulted in a more normal phenotype as judged by growth rate, cell cycle parameters, and morphology. beta-aminopropionitrile, the selective inhibitor of lysyl oxidase enzyme activity, was remarkably unable to block suramin-induced reversion. By contrast, ectopic antisense lysyl oxidase demonstrated that lysyl oxidase gene expression mediated phenotypic reversion. Since lysyl oxidase is synthesized as a 50 kDa precursor and processed to a 30 kDa active enzyme and 18 kDa propeptide, the effects of these two products on the transformed phenotype of RS485 cells were then directly assessed in the absence of suramin. Here we report, for the first time, that the lysyl oxidase propeptide, and not the lysyl oxidase enzyme, inhibits ras-dependent transformation as determined by effects on cell proliferation assays, growth in soft agar, and Akt-dependent induction of NF-kappaB activity. Thus, the lysyl oxidase propeptide, which is released during extracellular proteolytic processing of pro-lysyl oxidase, functions to inhibit ras-dependent cell transformation.

Lysyl oxidase catalyzes oxidative deamination of peptidyl lysine and hydroxylysine residues in collagens, and peptidyl lysine residues in elastin. The resulting peptidyl aldehydes spontaneously condense and undergo oxidation reactions to form the lysine-derived covalent cross-links required for the normal structural integrity of the extracellular matrix (1)(2)(3). Lysyl oxidase is synthesized as a 48 -50 kDa proenzyme, secreted into the extracellular environment where it is then processed by proteolytic cleavage to a functional 30 kDa enzyme and an 18 kDa propeptide (4). Evidence supports that 30 kDa lysyl oxidase is active whereas the 50 kDa proenzyme is enzymatically inactive (5)(6)(7). Procollagen C-proteinases are active in processing pro-lysyl oxidase and are products of the Bmp1 gene and the structurally related Tll1 and Tll2 genes (6 -8).
Lysyl oxidase gene expression was found to inhibit the transforming activity of ras and was hence named the "ras recision gene" (rrg) (9,10). Lysyl oxidase is down-regulated in rastransformed cells and in many cancer cell lines. Reduced lysyl oxidase levels are also observed in human cancers (9,(11)(12)(13)(14)(15), whereas in spontaneous revertants or upon induced phenotypic reversion higher normal levels of lysyl oxidase are again seen (9,14). Conversely stable phenotypic revertants of ras-transfected NIH3T3 cells return to a transformed phenotype upon transfection with an antisense lysyl oxidase vector (9,10,16). Antisense lysyl oxidase transfection triggers transformation of normal rat kidney fibroblasts (17). Thus, the lysyl oxidase gene has tumor suppressor activity. Recently, we showed that ectopic expression of pro-lysyl oxidase in ras-transformed cells inhibits the activities of the phosphatidylinositol 3-kinase (PI3K), 1 Akt, and MEK kinases that lead to the activation of NF-B (18).
Suramin is a polysulfonated naphthylurea, initially used in the treatment of trypanosomiasis and onchocerciasis (19). Its anticancer activity was later identified, and suramin has been introduced into clinical trials for various forms of cancer (20 -24). Suramin interrupts autocrine growth factor pathways by inhibiting the binding of growth factors to their receptors (19,(25)(26)(27)(28). We have recently shown that lysyl oxidase is dramatically up-regulated by suramin in c-Ha-ras-transformed NIH3T3 cells (RS485 cells) because of its inhibition of an FGF-2-mediated autocrine pathway (29). The question is now raised whether suramin causes phenotypic reversion of RS485 cells, and whether this reversion depends on lysyl oxidase expression or activity. In addition, the role of lysyl oxidase enzyme activity in mediating phenotypic reversion was investigated. The results indicate that suramin-induced phenotypic reversion requires lysyl oxidase expression as expected, but reversion surprisingly does not require lysyl oxidase enzyme activity. Data show that reversion is mediated instead by the lysyl oxidase propeptide. The ability of the lysyl oxidase propeptide, and not active lysyl oxidase, to stimulate phenotypic reversion was confirmed in studies of ras-transformed cells performed in the absence of suramin, and data suggest that the lysyl oxidase propeptide inhibits the ras-dependent PI3K/PDK1/Akt pathway. These findings identify the 18 kDa lysyl oxidase propeptide as a novel inhibitor of ras-mediated transformation of fibroblasts.

EXPERIMENTAL PROCEDURES
Chemicals-Suramin was either kindly provided by the Division of Cancer Treatment Diagnosis and Centers, NCI, National Institutes of Health and Parke-Davis, or was purchased from Sigma. All other chemicals and reagents were purchased from Sigma or Invitrogen.
Cell Culture-RS485 cells are transformed by overexpression of c-Ha-ras in NIH3T3 cells (30). PR4 cells are stable phenotypic revertants of RS485 cells obtained after treatment with ␣/␤ interferon (31), and AS-3B cells are re-transformed after transfection of PR4 cells with antisense lysyl oxidase (9,10). Cells were plated onto 100-mm cell culture plates in Dulbecco's modified Eagle's medium, containing 10% fetal bovine serum (FBS) plus 1% nonessential amino acids, 100 units/ml penicillin and 100 g/ml streptomycin. Cultures were maintained at 37°C in a fully humidified atmosphere of 5% CO 2 in air. Cells in logarithmic growth phase, were dissociated with trypsin/EDTA, and inoculated at a desired density for each experiment.
Growth Curves-To study cell growth rates, cells were plated in 6-well plates at a density of 35,000 cells/well and were grown in complete medium containing 10% FBS, as described above. Additions of suramin or ␤-aminopropionitrile (BAPN), when appropriate, were initiated 24 h after plating. Media were changed every 3 days in the continuous presence of suramin or BAPN, as indicated for each experimental design in the "Results." Cell density was determined in triplicate every day by crystal violet staining, as described (32,33). Cells were fixed with 10% formalin in PBS, washed with PBS, and then stained for 30 min with 0.1% crystal violet at room temperature with shaking. Unbound dye was then removed by washing with water until washes were colorless. Bound dye was then eluted with 10% acetic acid, and quantitated by measuring the absorbance at 590 nm. For quantitative analyses of growth rates, the logarithmic value of absorbance versus time was plotted Ϯ S.D. and the rates were calculated by linear regression analyses. In addition, data were plotted as total absorbance Ϯ S.E. versus time. Experiments were performed three times each with consistent findings.
Lysyl Oxidase Enzyme Activity-PR4 cells were plated in 100-mm cell culture plates and then grown and re-fed every 2 days as described above in the constant presence of 0, 200, and 400 M BAPN for 7 days until visually confluent. Cells were then re-fed with serum-free medium supplemented with 0.1% bovine serum albumin still in the constant presence or absence of BAPN. After 24 h conditioned 0.3-ml aliquots of media samples were assayed in quadruplicate using a tritiated recombinant human tropoelastin as substrate as previously described (4). Incubations were performed at 37°C for 90 min (34), and data were expressed as total cpm released Ϯ S.E. per culture.
Cell Cycle Analysis-RS485, NIH3T3, and PR4 cells were plated on 100-mm plates and were grown until confluent with 0 or 150 M suramin. Cells (1.5-2 ϫ 10 6 ) were then trypsinized, washed with PBS, and fixed by washing with ice-cold 70% ethanol. Cells were stained with propidium iodide (50 g/ml) in PBS containing 2% FBS and was analyzed by flow cytometry using a FACScan flow cytometry with CELLQUEST acquisition and analysis software (BD Biosciences).
RNA Isolation and Northern Blot Analysis-Total RNA was isolated using the RNeasy-RNA isolation kit (Qiagen, Valencia, CA). 10-g samples of denatured RNA were electrophoresed on a 1% agarose gel containing 18% formaldehyde. Gels were transferred in 10ϫ SSC by capillary blotting overnight to Gene Screen nylon membranes (PerkinElmer Life Science, Boston, MA). The membranes were hybridized at 42°C as previously described (35) with labeled mouse lysyl oxidase probe (9,10), prepared by random primer labeling (36). For normalization and as a measure of constant loading of gels, blots were stripped, and rehybridized with a radiolabeled 18 S rRNA probe (37). Autoradiograms were assessed and normalized by densitometric scanning on a Versa Doc Model 3000 Gel Documentation System and Quantity One Software (Bio-Rad).
Stable Transfection of RS485 Cells with an Antisense Lysyl Oxidase Expression Vector-Cells were grown in 100-mm cell culture dishes. At ϳ70% confluence, they were transfected with the antisense murine lysyl oxidase expression vector pCLO3 (17) (20 g DNA/plate) using the calcium phosphate precipitation method (38). As a control, RS485 cells were transfected with empty vector (pcDNA3). The transfected cells were selected using G418 (geneticin) at a final concentration of 400 g/ml in the medium. Colonies were isolated from antisense lysyl oxidase and empty vector transfected RS485 cells using cloning cylinders (39) and cultured in 400 g/ml G418. Cells were then plated onto 100-mm plates (250,000 cells/plate) without G418 and treated with 0 or 150 M suramin. After 24 h, cells were prepared for cell cycle and Northern analysis as described above.
Lysyl Oxidase Propeptide-coated Cell Culture Plates-Rat lysyl oxidase propeptide was expressed in E. coli and purified as described (40). The propeptide (200 -400 g/ml) was then dialyzed against 16 mM phosphate buffer, pH 7.8 for 5 h and 6-well plates were coated with 0, 1, 5, or 10 g of propeptide in 1 ml of water per well and left overnight under UV light in the cell culture hood to completely dry. AS-3B or RS485 cells were then plated at a density of 35,000 cells per well on the propeptide-coated plates and cultured until visual confluence. Cells were then prepared for cell cycle analysis as described above. To study the effect of propeptide on the growth rate of AS-3B and RS485 cells, 24-well plates were coated with 0, 0.2, 1, 2, or 4 g of propeptide in 350 l of water per well, as above, and the cells plated at a density of 7,000 cells/well. Cell density was determined in triplicate every day, by crystal violet staining as described above. In selected experiments, mature 30 kDa lysyl oxidase (36) was dialyzed against 16 mM potassium phosphate buffer, pH 7.8 for 5 h, and 4 g coated in 24-well plates at the same time as described for the propeptide.
Focus Formation Assay in Soft Agar-RS485 cells and Myctransformed M158 cells were plated, in duplicate, at 10 4 cells/ml in top plugs consisting of complete Ham F-12 nutrient mixture medium and 0.4% SeaPlaque agarose (FMC Bioproducts, Rockland, Maine) in the presence of 2.5 g of purified bovine aorta lysyl oxidase enzyme (30 kDa form) (41), or with 2.5 g recombinant rat lysyl oxidase propeptide (18 kDa form) (40), or the same volume of vehicle potassium phosphate (16 mM, pH 7.8). After 2 weeks of incubation in a humidified incubator at 37°C, the colonies were stained with 0.5 ml of 0.0005% crystal violet and photographed using a digital camera coupled to a dissection microscope (ϫ50 magnification). Three random fields were counted from each of two duplicate samples, and average values presented Ϯ S.D.
Transient Transfections, Luciferase Assays, and Fluorescence Microscopy-NIH3T3 and RS485 cells were plated in 35-mm culture dishes. Cells were transfected overnight, in triplicate, with the indicated expression vectors by using FuGENE 6 transfection reagent (Roche Applied Science) in Dulbecco's modified Eagle's medium containing 0.5% FBS. The plasmids used are pCMVneo-Myr-Akt kindly provided by Z. Luo (Boston University Medical School, Boston, Mass.), NF-B-luciferase which was a gift from G. Rawadi (Hoechst-Marion-Roussel, Romainville, France), pEGFP-C1-PDK1 kindly provided by J. Chung (Korean Advanced Institute of Science and Technology, Taejon, Republic of Korea), and pcDNA3.1 (ϩ)/LOPP propeptide and pcDNA4-LO enzyme expression vectors. The expression vector for the lysyl oxidase propeptide pcDNA3.1 (ϩ)/LOPP was generated from pSV40 PolyACOD (5) by PCR, using forward primer: 5Ј-ACTGGATCCCGA AGAGGTCTCCCTC-CTTCGCG-3Ј and reverse primer 5Ј-TACGAATTCTCAGCCCACCAT-GCGATCTACGTGGCTG-3Ј. The DNA was digested with BamHI and EcoRI and gel-purified and cloned into pcDNA3.1 (ϩ) (Invitrogen), resulting in pcDNA3.1 (ϩ)/LOPP. This construct contains the rat cDNA sequence (Ϫ94 to ϩ486) that includes a portion of the 5Ј-UTR, the signal peptide, the entire rat lysyl oxidase propeptide coding region and no mature lysyl oxidase sequence. The insert was directly confirmed by DNA sequencing. The expression vector for mature lysyl oxidase was accomplished by excision of nucleotides encoding amino acid residues 23-157 from a construct of murine lysyl oxidase cDNA Ϫ33 to ϩ1234, and then cloned into pcDNA4 as previously reported (42).
For luciferase assays, 1 g of NF-B-dependent luciferase reporter plasmid and 0.5 g of pSV40-␤-Gal reporter gene were co-transfected with the indicated DNAs. Cells were stimulated with addition of FBS to a final concentration of 10%, and total cell extracts were prepared after 48 h. The resulting extracts were normalized for ␤-Gal expression and used in a luciferase activity assay, according to the manufacturer's instructions (Promega kit). Results are expressed as the fold induction of luciferase activity calculated under stimulation conditions (10% FBS) compared with starvation conditions (0.5% FBS). The results are expressed as the mean Ϯ S.D. For fluorescence microscopy, cells were co-transfected with a vector expressing green fluorescent proteintagged PDK1 protein (GFP-PDK1), and either lysyl oxidase propeptide or lysyl oxidase enzyme expression vectors or parental empty vectors for 48 h. Localization of the GFP-PDK1 was determined using an Anxiovert 200M fluorescent microscope (Carl Zeiss MicroImaging, Inc., Thornwood, N.Y.). Analyses and photographs were performed using Axiovision (v.3.1 software; Carl Zeiss MicroImaging, Inc.). (29). Here we investigated the effects of suramin on the rate of growth and morphology of RS485 cells, as an initial measure of cell phenotype. Cells were plated in 6-well plates and cultured for 24 h and then grown in the continuous presence of 0, 100, 125, or 150 M suramin. As control, the growth of phenotypically normal NIH3T3 cells in the absence of suramin was analyzed at the same time. Cell growth was determined by daily crystal violet staining of replicate wells (Fig. 1). RS485 cells grew more rapidly than NIH3T3 cells, as expected (30). Suramin significantly decreased the growth rate in a dose-dependent manner. Data show that 100, 125, and 150 M suramin decreased the growth rate by 38, 49, and 56%, respectively, calculated from linear regression analyses of log of absorbance versus time. The growth rate of RS485 cells treated with 150 M suramin was similar to that of NIH3T3 cells. Furthermore, treatment with suramin caused a dose-dependent change in the morphology of RS485 cells with cells appearing less transformed, i.e. flatter and contact inhibited in the presence of 150 M suramin (inset, Fig. 1).

Effect of Suramin on RS485 Cell Phenotype-Treatment of c-Ha-ras-transformed NIH3T3 cells (RS485 cell line) with suramin leads to the induction of lysyl oxidase
We next determined the effects of suramin on cell cycle progression of RS485 cell cultures. As shown in Table I, fluorescenceactivated cell sorting (FACS) indicated that treatment with 100 or 150 M suramin increased the percentage of RS485 cells in G 1 and decreased that in S phase in a dose-dependent manner. The proportions of cells in G 1 and S phases in cultures treated with 150 M suramin are nearly identical to untreated phenotypically normal NIH3T3 cells (Table I). Thus, treatment with 150 M suramin changed the morphology, cell cycle, and growth rate of RS485 cells resulting in a more normal phenotype.
Inhibition of Lysyl Oxidase Expression Reduces Phenotypic Reversion Induced by Suramin-Data presented above indicate that suramin causes phenotypic changes in RS485 cells. To determine that the suramin-induced phenotypic changes depend upon lysyl oxidase expression, we generated stable antisense lysyl oxidase transfected RS485 cell clones. Nine anti-sense lysyl oxidase transfected clones and six empty vector transfected clones were generated and grown in the presence and absence of 150 M suramin, and after 24 h cells were fixed and subjected to cell cycle analysis. Non-transfected RS485 cells were analyzed as an additional control. Fig. 2 shows the differences in the percentage of cells in G 1 and S phase as a function of suramin treatment. In empty vector transfected clones, suramin increased the average number of cells in G 1 by 20.8%, and decreased the average number of cells in S phase by 13.5% (bar N in Fig. 2). As expected, these values are not significantly different from non-transfected cells (Fig. 2, compare V to N). In contrast, antisense lysyl oxidase transfected clones show only a 10.6% increase in the percentage of cells in G 1 phase and a 6.9% decrease in the S phase after suramin treatment. These changes are significantly smaller than either of the control groups of cells (Fig. 2 To confirm that lysyl oxidase expression is actually diminished by antisense lysyl oxidase transfection, the ability of 150 M suramin to induce low steady state mRNA levels of lysyl oxidase in clones was assessed after 24 h of treatment by Northern blot analysis with normalization to 18 S rRNA signals. Suramin treatment led to an average 8-fold increase in lysyl oxidase mRNA levels in the six empty vector-transfected clones assayed, consistent with previous studies on non-transfected RS485 cells (29). In contrast, an average 1.8-fold in-  crease in lysyl oxidase mRNA levels occurred in the nine antisense lysyl oxidase transfected clones. A Northern blot of RNA from representative empty vector and antisense transfected clones is shown in Fig. 3. Thus, the induction of steady state lysyl oxidase mRNA levels by suramin is inhibited by the antisense lysyl oxidase transfection, as expected. Taken together, these data demonstrate that antisense lysyl oxidase transfected cells have significantly diminished suramin-induced cell cycle changes compared with those of empty-vector transfected or non-transfected RS485 cells. These data indicate that lysyl oxidase expression plays a role in mediating the phenotypic effects of suramin on RS485 cells.

BAPN-mediated Inhibition of Lysyl Oxidase Enzyme Activity Fails to Prevent Phenotypic Reversion Induced by Suramin-It
is generally assumed, although never directly tested, that lysyl oxidase-dependent phenotypic reversion and tumor suppressor activity depends on its enzyme activity. To directly test the role of lysyl oxidase enzyme activity, we measured the effects of the lysyl oxidase inhibitor, BAPN (43), on the phenotypic changes in RS485 cells following treatment with 150 M suramin. We have previously shown that suramin induces lysyl oxidase activity by about 2.5-fold (29). RS485 cell growth was assessed in the absence or presence of either 150 M suramin or 400 M BAPN, or a combination of both 150 M suramin and 400 M BAPN. This concentration of BAPN effectively inhibits lysyl oxidase (see below and Ref. 44). As shown in Fig. 4A, suramin decreased the growth rate of RS485 cells whereas BAPN had no effect on the growth rate. Surprisingly, BAPN did not reverse or affect in any detectable way the suramin-mediated inhibition of RS485 cell growth.
Studies performed with the stable phenotypic revertant cell line PR4 demonstrated that lysyl oxidase expression specifically is required to maintain the normal phenotype (9,10). These studies utilized antisense transfection methodology to reduce lysyl oxidase expression resulting in transformation, but did not directly investigate the role of lysyl oxidase enzyme activity in phenotypic reversion. If lysyl oxidase enzyme activ-ity were required for the normal phenotype of PR4 cells, then BAPN would cause re-transformation. Thus, growth curves were generated for PR4 cells in the presence of 0 or 400 M BAPN. As shown in Fig. 4B, BAPN did not affect the growth rate of PR4 cells. Furthermore, BAPN did not change the morphology of PR4 cells (data not shown). Assays of PR4 cell culture media confirmed that cells grown without BAPN produce easily detectable lysyl oxidase enzyme activity (14,000 Ϯ 400 dpm ϫ 10 6 cells), whereas no lysyl oxidase enzyme activity was detected in the medium of cultures grown at the same time in the continuous presence of both 200 M and 400 M BAPN using a highly sensitive assay for lysyl oxidase enzyme activity (4) (data not shown). Taken together, these findings show that growth inhibition of RS485 cells by suramin does not depend on lysyl oxidase enzyme activity. Similarly, inhibition of lysyl oxidase enzyme activity does not affect growth of stable phenotypic revertants that require lysyl oxidase expression for the normal cell phenotype. vector (9,10) and should be sensitive to features of lysyl oxidase that cause phenotypic reversion. Cells were plated on 6-well plates that had been coated with 0, 1, 5, or 10 g of propeptide per well. This experimental approach was taken due to the poor solubility of the propeptide in cell culture media and physiologic buffers. After 4 days, cells (1-2 ϫ 10 6 ) were harvested and subjected to cell cycle analysis. For AS-3B cells ( Table II) the percentage of cells in G 1 phase increased in the presence of the propeptide by about 6% while the percentage of cells in S phase decreased by 5.7%, suggesting that the lysyl oxidase propeptide has a role in altering the cell cycle of AS-3B cells. Similarly, in RS485 cells (Table III) the percentage of cells in G 1 phase increased by 7.6% in the presence of the propeptide, with a corresponding decrease in the percentage of cells in S phase. Results suggest that propeptide affects cell cycle progression of both AS-3B and RS485 cells.

Lysyl Oxidase Propeptide and Not the Active Enzyme Causes Phenotypic Reversion of Antisense Lysyl Oxidase-transfected PR4 Cells (AS-3B) and RS485 Cells-The
We next investigated the effects of the lysyl oxidase propeptide and of the mature 30 kDa enzyme on the growth of AS-3B and RS485 cells using coated 24-well plates. As shown in Fig. 5, A and B, propeptide decreased the growth of both cell lines in a dose-dependent manner. Linear regression analyses of plots of the log of absorbance versus time demonstrated dose-dependent growth inhibition of 6.0, 9.3, 13.6, and 17.6% for AS-3B cells and 4.1, 5.6, 8.9, and 11.9% for RS485 cells with 0.2, 1, 2, or 4 g of lysyl oxidase propeptide, respectively. Moreover, the lysyl oxidase propeptide did not affect the plating efficiency of these cells, as initial crystal violet absorbance values were essentially identical for cells plated on propeptide compared with no propeptide (Fig. 5). No obvious effect of propeptide on cell morphology was observed. As an additional control in Fig. 5A, AS-3B cells were grown at the same time on mature 30 kDa lysyl oxidase enzyme. In contrast to the effects of the lysyl oxidase propeptide, lysyl oxidase enzyme did not inhibit the growth rate of AS-3B cells (Fig. 5A). In fact, growth on lysyl oxidase enzyme appeared to be slightly increased compared with the control. As seen in Fig. 5A, this is because of higher plating efficiency of cells on the mature enzyme. Linear regression analyses showed that the rate of cell growth on lysyl oxidase enzyme was essentially unaffected (increased by 2%), even though plating efficiency was increased. Taken together, the data indicate that the lysyl oxidase propeptide has a specific inhibitory effect on cell growth and cell cycle progression that contributes to phenotypic reversion.
Lysyl Oxidase Propeptide, and Not the Active Enzyme, Inhibits Growth of RS485 Cells in Soft Agar-A hallmark of transformed cells is the ability to grow in soft agar and to form colonies, whereas non-transformed cells are unable to grow when suspended in soft agar. The respective effects of active 30 kDa lysyl oxidase enzyme, and of the 18 kDa lysyl oxidase propeptide versus vehicle control on the ability of RS485 cells to grow in soft agar were determined. Lysyl oxidase propeptide was strongly inhibitory, whereas the 30 kDa lysyl oxidase enzyme was unable to inhibit growth of RS485 cells in soft agar (Fig. 6). In two separate experiments an average 80% reduction in colony formation was observed. Lysyl oxidase-mediated reversion appears to be selective for ras-transformed cells (9,10). To investigate the specificity of the propeptide, its growth inhibitory on c-myc-transformed M158 fibroblasts was determined. Neither lysyl oxidase propeptide nor the 30 kDa lysyl oxidase enzyme inhibited the growth in soft agar of c-myctransformed M158 fibroblasts (Fig. 6). Thus, the 18 kDa lysyl oxidase propeptide, and not the active lysyl oxidase enzyme, inhibits ras-dependent transformation.
Lysyl Oxidase Propeptide and Akt-mediated NF-B Activity-NF-B is highly activated in ras-transformed NIH3T3 cells, and transfection with a full-length lysyl oxidase expres-

FIG. 5. Lysyl oxidase propeptide, but not enzyme, inhibits growth of transformed cells; AS-3B cells (A) and RS485 cells (B).
Cells were plated in 24-well plates (7,000 cells/well) precoated with 0 g (f), 0.2 g (Ⅺ), 1 g (OE), 2 g (E), or 4 g (q) recombinant rat lysyl oxidase propeptide/well. In A, cells were plated in addition on 4 g of mature 30 kDa lysyl oxidase (‚). Growth was assessed as described above in Fig. 1. Each data point is the average of values from three wells Ϯ S.D. In B the inset is the same data directly plotted as absorbance Ϯ S.D. versus time to more clearly show the growth inhibitory effect of the lysyl oxidase propeptide.

TABLE II
Effects of the lysyl oxidase propeptide on cell cycle progression of AS-3B cells AS-3B cells were grown on 0, 1, 5, or 10 g of propeptide-coated 6-well plates and cultured for 4 days. Cells (1.5-2 ϫ 10 6 ) were fixed with 70% ethanol and stained with propidium iodide and analyzed by flow cytometry. Data shown are the averages Ϯ S.D. of experiments performed three times.  cell cycle progression of RS485 cells RS485 cells were grown on 0, 1, 5, or 10 g of propeptide-coated 6-well plates and cultured for 2 days. Cells (1.5-2 ϫ 10 6 ) were fixed with 70% ethanol and stained with propidium iodide and analyzed by flow cytometry. Data shown are the averages Ϯ S.D. of experiments performed three times. sion vector reduces the NF-B activity in these cells (18). Moreover, this reduction is mediated in part via inhibition of PI3K/ Akt activities, which are induced by the ras in the phosphoinositide-dependent kinase-1 (PDK1) signaling pathway (18,45). We wished to determine the respective abilities of the lysyl oxidase propeptide, and the lysyl oxidase enzyme to inhibit NF-B dependent reporter constructs and the functional role of Akt in mediating these effects. Control samples show that the NF-B reporter is about 3 times more active in RS485 cells than NIH3T3 cells transfected with empty expression vectors in 10% FBS, as expected (Fig. 7). Transfection with the lysyl oxidase propeptide expression vector returned luciferase activity to control levels, whereas there was only a small reduction in NF-B activity in cells transfected with mature lysyl oxidase expression vector (Fig. 7). To verify that the inhibition of the Akt signaling cascade was mediating these effects, a vector expressing a constitutively active, myristoylated Akt was employed. In the presence of constitutively active Akt, the lysyl oxidase propeptide no longer significantly reduced luciferase activity (Fig. 7). Little effect was seen with the lysyl oxidase enzyme. These results support the notion that the lysyl oxidase propeptide, and not the lysyl oxidase enzyme, inhibits NF-B activation via an Akt-dependent mechanism, thus implicating the ras-dependent PI3K/Akt pathway as a target for the lysyl oxidase propeptide. PI3K activity results in the formation of phosphatidylinositol phosphates that then activate PDK1. In turn, PDK1 phosphorylates and activates Akt. In this pathway, active PDK1 is recruited to the plasma membrane where it complexes with Akt. Pro-lysyl oxidase transfection has previously been shown to inhibit plasma membrane localization of GFP-labeled PDK1 (18). We next investigated the ability of lysyl oxidase enzyme versus the lysyl oxidase propeptide to alter PDK1 plasma membrane localization, as judged by the staining pattern of ectopic GFP-PDK1. Data in Fig. 8 indicate that RS485 cells have a high level of staining of the plasma membrane seen as a distinct fluorescent border outlining these cells, whereas wild type NIH3T3 cells exhibit predominantly cytoplasmic staining, as expected. Transfection of RS485 cells with mature lysyl oxidase expression vector did not alter the plasma membrane localization of PDK1, whereas transfection with the lysyl oxidase propeptide eliminated this feature. Overall, these data support the conclusion that the lysyl oxidase propeptide, and not the enzyme, inhibits the PI3K/Akt pathway that is downstream of ras. DISCUSSION This report shows for the first time that the ability of lysyl oxidase to revert the phenotype of ras-transformed fibroblasts depends substantially on the propeptide domain, and not on lysyl oxidase enzyme activity. Since diminished lysyl oxidase expression in some way contributes to the transformed phenotype, it has generally been assumed that lysyl oxidase enzyme activity is related to the tumor suppressor activity of lysyl oxidase, and, therefore, that diminished lysyl oxidase activity promotes the transformed phenotype. However, BAPN, the specific inhibitor of lysyl oxidase enzyme activity, did not prevent suramin-mediated reversion of the transformed phenotype, which is accompanied by increased lysyl oxidase expres- sion. These findings were confirmed in the stable phenotypic revertant cell line PR4, that requires lysyl oxidase expression for normal phenotype maintenance; yet inhibition of lysyl oxidase activity with BAPN failed to re-transform these cells. Similarly, BAPN failed to block the ability of ectopic lysyl oxidase expression to prevent growth of ras-transformed fibroblasts in soft agar (data not shown). The lack of effect of BAPN on lysyl oxidase-dependent phenotype control is interesting. Intracellular localization of mature lysyl oxidase has been shown to occur via normal extracellular processing of pro-lysyl oxidase, followed by uptake of mature lysyl oxidase (46). Given that BAPN is an irreversible inhibitor of lysyl oxidase (43), it follows that both extracellular and intracellular lysyl oxidase activity are susceptible to inhibition by BAPN. These findings, therefore suggest that neither extracellular nor intracellular lysyl oxidase activity contribute significantly to inhibiting the transformed cell phenotype.
The importance of lysyl oxidase expression in maintaining a normal cell phenotype in suramin treated RS485 cells was supported by antisense transfection studies. Most important, using recombinant lysyl oxidase propeptide, we demonstrated that the lysyl oxidase propeptide itself directly stimulates phenotypic reversion of ras-transformed cells, as judged by rate of proliferation, cell cycle colony formation in soft agar, and PDK1/Akt signaling to activate NF-B. This contrasts with the absence of an effect of the lysyl oxidase enzyme on both the growth rate of AS-3B cells, and on colony formation in soft agar of RS485 cells. These studies identify an activity of the lysyl oxidase propeptide that may ultimately prove to be of therapeutic significance in the treatment of cancers in which rasdependent pathways are abnormally active.
A concept that has gained increasing experimental support is that many proteins have multiple biological functions (47). A recently reported example is the housekeeping cytoplasmic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which has been unexpectedly found in the nucleus as a component of a transcription complex (48). It is unknown whether GAPDH enzyme activity plays a role in its activity as a transcription complex component, though its co-activator activity does depend on NAD ϩ binding. Lens crystallins are structural proteins found in the lens that in other contexts serve as enzymes (49). Similarly, there is an increasing appreciation for biological activities of propeptides of structural proteins that are released as a result of biosynthetic proteolytic processing and maturation. C-propeptides of type I and type II collagen are ligands for ␣ 2 ␤ 1 integrins and they inhibit collagen gene transcription (50 -53). The C-propeptide of type I collagen promotes attachment of osteoblasts, and is chemotactic for endothelial cells (54,55). A variant of the N-terminal propeptides of type II collagen binds and modulates the activity of TGF-␤ 1 and BMP-2 in developing cartilage (56). Endostatin is a 20 kDa cationic protein derived from C-terminal extensions of type XVIII procollagen that inhibits angiogenesis (57), an activity that is receiving much attention as a therapeutic approach to treat cancer. The finding reported here that the lysyl oxidase propeptide has biological function as an inhibitor of cell transformation provides a new and important example of a distinct biological activity derived from an extracellular protein precursor.
Structural features of the lysyl oxidase propeptide are interesting. The biosynthesis of lysyl oxidase occurs by secretion of a 50 kDa precursor, followed by extracellular proteolytic processing to form active 30 kDa lysyl oxidase and the 18 -20 kDa propeptide (5-7). Unlike the anionic C-terminal region of prolysyl oxidase that becomes the active enzyme after processing (6,7), the N-terminal propeptide region is rich in arginine and is cationic with a calculated pI of 12.5 for the mouse, rat, and human proteins. We hypothesize that the highly basic character of the lysyl oxidase propeptide could facilitate its uptake by cells where it might exert its biological function, possibly entering cells even in the absence of a specific receptor. Cell membranes are permeable to arginine-rich basic proteins, and uptake of these basic proteins is mediated by heparin sulfate proteoglycans (58). The arginine-rich highly basic propeptide region of lysyl oxidase is less well conserved between species than the mature enzyme (59), but it contains blocks of 33 and 37 amino acids residues in length, respectively, that are nearly perfectly conserved between mouse and human, and highly conserved in chicken lysyl oxidase. These regions are residues 26 -59 and 77-114, respectively, in the mouse lysyl oxidase sequence. This high degree of similarity suggests that biological activities of the lysyl oxidase propeptide reside in these conserved sequences. Lysyl oxidase is a member of a multigene family, and it is notable that the sequence of the lysyl oxidase propeptide region is not well conserved among other lysyl oxidase family members, whereas the catalytic domains are well conserved. Lysyl oxidase itself, and not the lysyl oxidase like genes, has been consistently identified in screens for tumor suppressors and is expressed at low levels in transformed cells and at higher levels in phenotypically normal cells (9,12). The finding of phenotype modulating activities occurring in regions of lysyl oxidase that are located in the unique propeptide domain may help to explain why lysyl oxidase itself is a tumor suppressor. Comparisons of the lysyl oxidase propeptide sequence with data bases have so far not revealed clues regarding the mechanisms by which the lysyl oxidase propeptide functions to inhibit cell transformation. However, our data do suggest that the lysyl oxidase propeptide inhibits the ras-dependent PI3K/PDK1/Akt signal transduction pathway similar to what was found in ras-transformed cells transfected with fulllength lysyl oxidase (18). It is interesting to note that NF-B activity in propeptide transfected cells was inhibited as early as 48 h after transfection, whereas this effect was not observed until 72 h after transfection with the full-length lysyl oxidase proenzyme. Although our data so far do not directly address whether full length lysyl oxidase proenzyme that contains the propeptide is active in promoting phenotypic reversion, these results suggest that it is the released propeptide itself that inhibits NF-B activation. Future studies will directly investigate the structure/ function relationships and mechanism of action of the lysyl oxidase propeptide in cell phenotype control.