Chemopreventive Property of a Soybean Peptide (Lunasin) That Binds to Deacetylated Histones and Inhibits Acetylation¹ Free

Epidemiological evidence suggests that dietary factors play an important role in the etiology of different kinds of cancer (1). For instance, diets rich in soybean products are associated with lower cancer mortality rates, particularly for cancers of the colon, breast, and prostate (2, 3, 4). Components of soybean believed to be capable of suppressing carcinogenesis include the BBI,⁴ inositol hexaphosphate, β-sitosterol, and isoflavones (5). BBI, now in clinical trials, has been shown to suppress carcinogenesis in laboratory animals and in in vitro transformation systems (5).

We isolated and cloned a cDNA encoding a posttranslationally processed 2S albumin (Gm2S-1) from mid-maturation soybean seed (6). The small subunit peptide of Gm2S-1 (lunasin) arrests mitosis, leading to cell death when the lunasin gene is transfected and expressed inside mammalian cells (7). The antimitotic effect of lunasin is attributed to the binding of its poly-aspartyl carboxyl end to regions of hypoacetylated chromatin, similar to that found in centromeres. As a result, the kinetochore complex does not form properly, and the microtubules fail to attach to the centromeres, leading to mitotic arrest and eventually to cell death (7).

The affinity of lunasin for hypoacetylated chromatin suggests a role in chromatin modification, a process implicated in cell cycle control and in the role of tumor suppressors in carcinogenesis (8). The RGD motif (9, 10) in lunasin and its chromatin-binding property point to a potential anticarcinogenic role. Here, we report that the lunasin peptide suppresses foci formation in mice fibroblast cells induced by chemical carcinogens. Biochemical evidence show that lunasin preferentially binds to deacetylated histones and inhibits acetylation. Finally, in the SENCAR mouse skin tumor model, dermal application of lunasin inhibits skin tumorigenesis.

The cell adhesion assay was used as described previously (11). Synthetic Lunasin and lunasin-GRG peptides (Research Genetics) dissolved in 1 ml of 1× PBS were bound to a noncoated 24-well Falcon plate for 3 h at 37°C in a CO₂ incubator. Wells were aspirated, and 0.1% albumin in PBS was added before the plate was incubated for another 30 min. The albumin was aspirated, and approximately 3 × 10⁵ C3H10T½ cells (ATCC), suspended in 1 ml of DMEM (DMEM) + 10% fetal bovine serum (FBS), were added to each well. After 1 h incubation, nonadherent cells were washed by gentle vacuum with 0.3 ml PBS. Cells were fixed by adding 1 ml of 5% formaldehyde and incubating at room temperature for 30 min. Formaldehyde was removed and the cells stained with Giemsa overnight at RT. Cells were washed 5 times with 0.5 ml PBS and dissolved in 200 μl of 0.2% SDS (SDS) in PBS. The absorbances at 630 nm were read in a microtiter plate and quantified in an EL 340 BioKinetics microplate reader.

C3H 10 T1/2 cells (American Type Culture Collection) grown in DMEM + 10% FBS were released from confluency and allowed to grow for 20 h, and lunasin was added to a final concentration of 1 μm. After 4 h, the cells were washed with PBS, trypsinized, transferred into plates with glass coverslips, and incubated overnight at 37°C. Cells on coverslips were fixed with 4% paraformaldehyde/PBS for 30 min at room temperature, washed with PBS, and permeabilized by incubating in 0.2% Triton X-100 on ice for 5 min. Cells were washed twice with PBS, blocked with 3% BSA/PBS for 30 min, and vacuum drained. Cells were incubated with primary antibodies for 1 h at room temperature. For MAD, an XMAD2 polyclonal antibody from Xenopus (Dr. R. H. Chen, Cornell University, Ithaca, NY), with cross-reactivity to murine MAD, was used at 1:200 dilution (12). Polyclonal antibody raised against the 16-amino acid COOH-terminal end of lunasin was used at 1:500 dilution. Cells were washed four times with 3% BSA/PBS before incubation for 30 min with secondary antibodies that were diluted 1:100 (fluorescein-labeled antimouse IgG from donkey to stain lunasin and a Texas Red-labeled antirabbit IgG from goat to stain MAD; Jackson Immunoresearch). Cells were washed four times, stained with 1 ng/ml DAPI for 45 s, and washed with distilled water twice before mounting onto glass slides with Antifade. Cells were examined with an Axiophot microscope (Zeiss) using a UV filter cube for DAPI and band pass filters of 450–490 nm for fluorescein and 530–580 nm for Texas Red. Images were captured using Scion Image and processed using Adobe Photoshop.

The standard transformation assay using C3H10T½ cells was described previously (13). Polycyclic hydrocarbons DMBA and MCA (Sigma Chemical Co.) were dissolved in acetone. C3H cells between the 9th and 15th passage were grown in DMEM + 10% FBS. Confluent plates of C3H cells were trypsinized and resuspended in fresh medium to 300 cells/ml dilution. For the 24-well plate experiments, 1 ml of cell suspension (∼300 viable cells) was added to each well. For the six-well plates, 2 ml were added to each well (∼600 viable cells). Cells were allowed to adhere overnight at 37°C, and after 20 h, the peptides were added. After 4 h, the chemical carcinogen (1.5 μg/ml of DMBA or 5 μg/ml of MCA) was added to the medium. Cells were exposed to the carcinogen for 20 h and washed with 1× PBS, and fresh medium was added. Acetone alone served as negative untreated control. BBI was used for comparison. Medium was changed every 5 days until the cells reached confluency and then once weekly. For the duration of the experiment, lunasin was added (to 1 μm) to fresh medium up to the indicated time point. After 6 weeks, wells in each plate were washed with 0.9% NaCl, fixed with methanol, stained with Giemsa, and scored for transformed foci. Treatments were replicated four times in six-well plates and twice in 24-well plates, a total of 24–48 wells/treatment. Transformation assays were conducted in duplicates at different cell passages and well sizes. Plating efficiency of each treatment was determined by counting (Coulter cell counter) the number of cells (cell size, >8 μm) 5 days after release from confluency, expressed as a percentage of the expected total number of viable cells assuming a generation time of 15.5 h (Ref. 13; i.e., 75,000 cells for 1.9-cm² wells and 150,000 for 9.5-cm² wells).

Wells of Pro-bind microtiter plates (Falcon) were coated with 50 μl (in PBS) of 1 μm synthetic lunasin and modified lunasin peptides (Research Genetics) by incubating at room temperature for 1 h. Residual binding sites were blocked with 100 μl of 3% BSA in PBS for 1 h. Wells were gently aspirated, and 50 μl (in PBS) of 50 nm synthetic NH₂-terminal tails (Upstate Biotechnology) of deacetylated H4 (H4) and tetra-acetylated H4 (H4-Ac) were added and incubated for 1 h. For wells coated with lunasin and lunasin-GRG peptides, 1 mm DTT was added to prevent disulfide linkages between lunasin and H4 and H4-Ac. Negative control wells were treated only with 50 μl of PBS, whereas positive control wells were coated with 50 nm of H4 or H4-Ac and blocked with 3% BSA in PBS. Wells were washed twice with 100 μl of PBS + 0.1% Triton X-100 (PBS-T) and incubated for 1 h with 50 μl of either H4 or H4-Ac primary antibodies diluted 1:1000 (Upstate Biotechnology). Wells were then washed twice with PBS-T before incubating with HRP-labeled antirabbit IgG (1:2000; Amersham) for 30 min. Finally, wells were washed twice with PBS-T before binding was detected by incubating in 100 μl 3,3′,5,5′-tetramethylbenzidine (Sigma Chemical Co.) for 30 min and then adding 50 μl of 0.5 m H₂S0₄. Absorbance at 490 nm was quantified on an EL 340 BioKinetics microplate reader. The percentage of H4 and H4-Ac peptide bound was computed based on the difference in a 490-nm absorbance reading between the peptide treatments and the negative control expressed as a percentage of the difference between the positive and the negative controls.

Acid-extracted proteins enriched for histone proteins were isolated from C3H and MCF-7 cells treated with either 2 μm lunasin and/or 5 mm sodium butyrate, using standard protocols (Upstate Biotechnology). Approximately 0.8 mg (C3H cells) and 1 mg (MCF-7 cells) of acid-extracted proteins were run on SDS-PAGE gels and blotted onto Hybond-ECL membranes (Amersham). H4-Ac and H3-Ac were detected using primary antibodies (Upstate Biotechnology), diluted 1:1000, and HRP-labeled antirabbit IgG secondary antibody, diluted 1:2000. Densitometer readings were taken on autoradiographs using the Bio-Rad Molecular Imaging System GS525 and Molecular Analyst software to measure levels of H4-Ac and H3-Ac.

Confluent plates of C3H cells were trypsinized, and 2.5 × 10⁵ cells were plated with fresh DMEM + 10% FBS medium in 60-mm cell culture plates. For lunasin treatment, 2 μm lunasin was added to the medium. After 20 h, C3H cells were transfected with E1Awt gene and a negative control E1A-ΔCR1 (lacking Rb-binding region) using standard protocols for liposome transfection (Clonfectin; Clontech). After 20 h, transfected C3H cells were observed, and phase contrast images were captured on a Zeiss inverted microscope. Annexin-V staining of apoptotic cells was conducted using standard protocols (Clontech). Annexin V-fluorescein staining of apoptotic, nonadherent cells was captured on a Zeiss Axiophot microscope with a 450–490 band pass filter and processed in Adobe Photoshop.

Two hundred fifty μg of lunasin and 200 μl of ethanol were applied on the shaved backs of two SENCAR (National Cancer Institute) mice. Twenty-four h and 1 week after treatment, a mouse was sacrificed, and a section of skin tissue was placed in formalin. A control skin tissue from a mouse treated only with ethanol was prepared similarly. Fixed tissues were embedded in paraffin, and 5 μm sections were cut and mounted on ProbeOn-Plus slides (Fisher). Immunohistochemical analysis was conducted using lunasin polyclonal antibody and alkaline phosphatase-conjugated secondary antibody with a hematoxylin counterstain to detect chromatin (nuclei).

Skin tumorigenesis was carried out on 12-week-old female SENCAR mice essentially according to Chen et al. (14). Lunasin dissolved in 250 μl of ethanol at three doses, low (2.5 μg/week), medium (25 μg/week), and high (250 μg/week), were applied to the shaved dorsal side of mice (n = 9 for each dose group) twice/week for 1 week before application of the initiator DMBA, 20 μg in 200 μl of acetone. Thereafter, lunasin was applied twice a week, followed 4 h later by the promoter TPA, 1 μg in 200 μl of acetone for 19 weeks. Mice were fed ad libitum on the AIN76A rodent diet (casein protein), and body weights were taken every week. The mice were observed for tumors once per week. Tumor yield in percentage was calculated as the number of tumor-bearing mice divided by the total number of mice in each group, whereas tumor yield/mouse was calculated by dividing the total number of tumors by total number of mice in each group. Positive control group (n = 6) was treated with DMBA and TPA only, whereas the negative control group (n = 8) was treated only with the application solvents. No significant differences were observed in body weights among the groups. Histopathology of the tumors collected at the end of the experiment showed papillomas only; no carcinomas were observed in the 19-week experiment. Tumor yield data were analyzed statistically by the χ² test, and tumor yield/mouse was analyzed by one-way ANOVA followed by Tukey’s test.

The lunasin peptide adhered to C3H cells in a dose-dependent manner, and deletion of RGD (lunasin-GRG) prevented cell adhesion (Fig. 1,A). When applied exogenously, lunasin became internalized, preferentially binding to the telomeres of chromosomes during metaphase (Fig. 1,B, lunasin and composite images). In our earlier work, constitutive expression of the lunasin gene disrupted kinetochore formation in transfected cells (7), leading to mitotic arrest and cell death, whereas here, internalized lunasin did not affect kinetochore assembly (Fig. 1,B), as shown by the presence of the kinetochore protein MAD (12) in the centromere of metaphase chromosomes (Fig. 1,B). Internalized lunasin was initially found in the cytoplasm (not shown), and upon nuclear membrane breakdown at prometaphase, it bound to hypoacetylated regions of the chromosome, such as those in the telomeres (Fig. 1 B). At this stage, kinetochore assembly and microtubule attachment to centromeres had already occurred; thus, mitosis proceeded unimpeded. We believe that these results explain the nondisruptive effect of exogenously applied lunasin on cell division as compared with the antimitotic effect of lunasin constitutively expressed in lunasin-transfected mammalian cells (7).

To determine whether lunasin inhibits carcinogenesis, an in vitro transformation assay was performed on normal, nontumorigenic C3H10T½ cells (C3H) that were allowed to proliferate in 24-well plates. DMBA and MCA were used to induce foci formation (13). The soybean BBI, shown previously to inhibit foci formation using this in vitro assay (15), was used as a control. Addition of 125 nm lunasin reduced transformation efficiency in DMBA-treated C3H cells (0.3%) by 80% compared with the positive control (1.6%), whereas in MCA-treated C3H cells, transformation efficiency was decreased by 64% (0.4% compared with 1.1% in PBS-treated positive control; Table 1, Exp. 1). The antitransformation effect of lunasin was significantly higher than that of equimolar amounts of BBI, which reduced transformation efficiency by 37% in DMBA-treated C3H cells. There was no significant difference in reduction of transformation efficiency between BBI and the positive control in MCA-treated C3H cells. The higher incidence of foci formation in MCA-treated cells can be attributed to a higher plating efficiency (14–19%) compared with DMBA treatment, which was more cytotoxic with plating efficiencies of 8–12%.

Table 1, Exp. 2, shows the dosage effects of lunasin on MCA-induced carcinogenesis. Twenty-four-h treatment with lunasin at 10 nm reduced transformation efficiency by 69% compared with the positive control. Treatment with 100 nm lunasin reduced transformation efficiency further by 84% but an increase to 1 and 10 μm resulted in non-statistically significant reductions of 88 and 90%, respectively. This suggests that there is a finite quantity of lunasin entering the cell, and increasing the amount of lunasin in the external medium may not have a significant effect on the rate of internalization within the 24-h exposure period. The higher reduction of transformation efficiency in Exp. 2 compared with Exp. 1 is explained by the two different batches of C3H cells used in the two experiments.

Six-well plates with 3-fold larger surface area were used to determine the effects of structural modifications in lunasin on MCA-induced transformation (Table 1, Exp. 3). The larger surface is expected to increase the probability of inducing mutations and foci formation attributable to additional cell divisions before postconfluence growth inhibition. Lunasin without the RGD motif (lunasin-GRG) showed no antitransformation effect, suggesting that the RGD motif is required for internalization. A lunasin variant that contains only the helical domain without the RGD motif and the poly-aspartyl end (trLunasin-del) likewise had no antitransformation effect. A form containing RGD, the poly-aspartyl end of lunasin and the helical domain replaced by a nuclear localization signal (NLS-trLunasin), was observed to penetrate the cell and nucleus but did not bind to chromatin.⁵ NLS-trLunasin did not inhibit foci formation, linking the ability to bind chromatin to the antitransformation property of lunasin. Conversely, a truncated lunasin (trLunasin) peptide containing the predicted helix, RGD, and the poly-aspartyl end was observed in immunostaining experiments to bind to chromatin⁵ and concomitantly reduced transformation efficiency by 29%. This was not significantly different from that of BBI.

In vitro immunobinding assay showed that full-length lunasin and lunasin-GRG bound with high affinity to the deacetylated H4 NH₂ terminus but not to the tetra-acetylated H4 (Fig. 2 A), suggesting that the RGD motif is not required for binding. However, deletion of the NH₂-terminal, 22-amino acid sequence (trLunasin) reduced binding by 70%, suggesting that the NH₂ terminus plays some role in facilitating the binding interactions.

Binding to deacetylated H4 of the 10 amino acid trLunasin-del peptide fragment, which spans the predicted helical domain upstream of the poly-aspartyl carboxyl end of the lunasin peptide (16), was not significantly different from that of trLunasin (Fig. 2,A). Substitution of this helix by a NLS in the truncated lunasin peptide (NLS-trLunasin) resulted in a 50% reduction of binding to deacetylated H4 (Fig. 2,A), suggesting an important role of the helix in binding. A homology search revealed structural similarity of the helical region to a short, conserved region of the chromodomain structure (17) found in chromatin-binding proteins such as Drosophila heterochromatin proteins, DmHP1A and DmPc, yeast transcriptional regulator SpSwI6A, and human heterochromatin HuHP1B (Fig. 2,B). Another human heterochromatin protein, HuHP1(p25), colocalizes with the centromeres during mitosis (18) and could be a potential target for constitutively expressed lunasin that leads to mitotic disruption (7). A naturally occurring mutation in Drosophila DmHP1A that substitutes isoleucine for phenylalanine (Fig. 2 B) leads to the disruption of the helix and the consequent loss of chromatin targeting (19).

It is noteworthy that trLunasin has a significantly reduced affinity for deacetylated H4 than the full-length lunasin peptide (Fig. 2,A), which correlates with the reduced efficacy of trLunasin in preventing foci transformation (Table 1, Exp. 3). This provides further evidence linking the binding of lunasin to deacetylated histones and its antitransformation property.

Histone acetylation and deacetylation have been associated with eukaryotic transcriptional regulatory mechanisms (20, 21). To demonstrate the effect of lunasin on histone acetylation in vivo, C3H and human breast cancer MCF-7 cells were treated with a histone deacetylase inhibitor, sodium butyrate (22), in the presence and absence of lunasin. In the absence of lunasin, sodium butyrate treatment increased the level of H4-Ac 98-fold in C3H and 40-fold in MCF-7 cells, 95-fold for H3-Ac in C3H and 35-fold in MCF-7 (Fig. 3). In the presence of lunasin, the increase in H4-Ac was not measurable in C3H and only 10-fold in MCF-7, 2-fold for H3-Ac in C3H, and 14-fold in MCF-7 (Fig. 3). These results suggest that under steady-state conditions in the cell, the core histones are mostly deacetylated (repressed state). Addition of a histone deacetylase inhibitor shifts the equilibrium to an acetylated state, and lunasin significantly inhibits this shift.

The tumor suppressor protein Rb prevents the transcription of a subset of E2F-regulated genes by binding to the transcription factor E2F and recruiting a histone deacetylase (HDAC1) to maintain the chromatin in the region in a hypoacetylated (repressed) state (23, 24, 25, 26). Inactivation of Rb presumably leads to uncoupling of HDAC1, acetylation of histones in the repressed chromatin, binding of the transcription factor E2F to its promoter, and activation of E2F-regulated genes involved in cell proliferation. This eventually leads to carcinogenesis (8, 27). To test this model, C3H cells were transfected with the viral oncogene E1A, known to induce cell proliferation by inactivating Rb (27), in the presence and absence of lunasin. A deleted E1A (E1A-ΔCR1; Ref. 28), without the Rb binding domain, was used as a negative control. C3H cells transfected with E1A-ΔCR1 showed normal cell division 20 h after transfection, both in the presence and absence of lunasin (Fig. 4). Transfection with the E1A gene without lunasin showed cells with normal morphology and the absence of nonadherent, apoptotic cells (Fig. 4). Pretreatment of C3H cells with 2 μm lunasin for 24 h before transfection with E1A resulted in the preponderance of nonadherent cells at 20 h after transfection. Phase contrast image of the nonadherent cells showed characteristic morphology of apoptotic cells, as confirmed by the positive fluorescent staining for Annexin V-FITC (Fig. 4). Treatment of normal C3H and NIH3T3 cells with lunasin did not affect cell proliferation or morphology.⁶

Lunasin dissolved in ethanol penetrated the skin dermal layer and colocalized with basal cell nuclei (Fig. 5,A). The morphology was not different from untreated skin. No significant differences were observed in body weights among the groups. The highest lunasin dose (250 μg/week) reduced skin tumor incidence by ∼70% (P < 0.01), decreased tumor yield/mouse (P < 0.005), and delayed the appearance of tumors by 2 weeks relative to the positive control (Fig. 5, B C). The lower doses (2.5 and 25 μg/week) showed a decreasing trend in tumor incidence and yield but were not statistically significant. The number of tumors/mouse for the different groups was 2.3 (positive control), 1.8 (2.5 μg/week), 1.2 (25 μg/week), and 0.3 (250 μg/week). Histopathology of the tumors collected at the end of the experiment (19 weeks) showed only papillomas; no carcinomas were observed. Carcinomas are usually observed after the 23rd week of DMBA initiation and TPA application (14).

This study demonstrates the chemopreventive property of the lunasin peptide and adds lunasin to the list of anticancer substances from soybean currently being investigated (2, 3, 4, 5). Extraction from soybean and recombinant DNA production through the lunasin gene present two ways of producing lunasin in addition to chemical synthesis. Initial assays in our laboratory show that commercial soy products contain reasonable amounts of lunasin, ranging from 5.48 mg of lunasin/g of protein (defatted soy flour) to 16.52 mg of lunasin/g of protein (soy concentrate). Daily consumption of 25 g of soy protein, recommended by Food and Drug Administration to reduce coronary disease, (29), supplies ∼250 mg of lunasin, but whether this dose is sufficient for chemoprevention still needs to be determined. Synthetic lunasin is heat stable, surviving heat up to 100°C for 10 min. In vitro digestibility studies show that lunasin is digested by pancreatin but protected by chymotrypsin and trypsin inhibitors derived from soybean.⁵ Bioavailability studies of natural and recombinant forms of lunasin should help determine physiologically relevant doses of lunasin for cancer prevention.

Although its physiological significance remains to be established, lunasin appears to be an ideal chemopreventive agent. In the absence of carcinogens, lunasin does not seem to affect cell morphology and proliferation but prevents the transformation in the presence of carcinogens. The proposed chromatin modification mechanism of lunasin is generally supported by the selective induction of apoptosis in E1A-transfected cells by lunasin, in vitro binding of lunasin to deacetylated histones, and the reduction of histone acetylation by lunasin in cells treated with a histone deacetylase inhibitor. We demonstrated recently that lunasin also suppresses foci formation in NIH3T3 cells transfected with the ras oncogene and E1A,⁶ supporting the efficacy of lunasin against different carcinogenic pathways. The observation that tumors induced in rodents by chemical carcinogens DMBA, methylnitrosourea, N′-methyl-N′-nitro-N-nitrosoguanidine), and X-rays have 70% frequency of ras mutations (30) is consistent with our in vitro and in vivo results. Overall, our results suggest that chemical carcinogenesis and viral oncogenesis share common mechanism(s) involving changes in chromatin status that lunasin disrupts to suppress cancer formation. Which genes and proteins are affected globally by lunasin in the presence and absence of carcinogens would be interesting to determine and should be amenable to analysis by microarray gene expression technology and immunoprecipitation techniques.

We thank L. Bjeldanes for the use of his laboratory for the initial transformation experiments; Yi Lam for reading the manuscript and working on the figures; S. Ruzin and D. Schichnes for the microscopy assistance; R-H. Chen for the XMAD2 antibody; H. Land and S. Newman for E1A gene constructs; T. Klein and R. Banatao for the structural analysis of lunasin; G. Wolf for the cell adhesion assay protocol; N. Duldulao and D. Krenz for technical assistance; R. L. Rodriguez (University of California Davis) for critical review of the manuscript; and D. Banatao, FilGen BioSciences, Inc. and UC Biostar Program for financial support of this research.