Inhibition of tumor growth by a novel approach: In situ allicin generation using targeted alliinase delivery

Allicin (diallyl thiosulfinate) is the best known biologically active component in the extract of freshly crushed garlic (1, 2). Intact garlic cloves do not contain allicin, but only its odorless precursor alliin (S-allyl-l-cysteine sulfoxide). Alliin is converted to allicin, pyruvate, and ammonia by the pyridoxal 5′-phosphate (PLP)-dependent enzyme alliinase (alliin lyase; EC 4.4.1.4) (3), which is enclosed in compartments within the garlic clove cells (Scheme 1). Crushing the clove exposes alliin to the enzyme, to initiate the following reaction:

Allicin reacts rapidly with free thiol groups and penetrates biological membranes with ease (4, 5). It therefore disappears from the circulation within a few minutes after injection (2, 4–6). This explains why the versatile and valuable activities of allicin, including its potent antibiotic and cytotoxic effects, were demonstrated thus far only in vitro (2, 7–9). We present here a new approach to anticancer therapy based on a localized, site-directed production of allicin. This approach can also be used to develop a wide range of antibiotic treatments. Alliinase is first targeted to the tumor. Conjugating the enzyme to a carrier, in this case, a mAb specific to a tumor-associated surface antigen, enables the targeting process. Alliin is then administrated into the circulation, which results in the formation of allicin only at the site of alliinase localization. The strategy for drug delivery based on antibody-directed enzyme prodrug therapy (ADEPT) was previously described (10–14). As in the ADEPT system, so alliinase-directed therapy uses mAbs specific to tumor cell antigens, to anchor the enzyme onto the cell surface. The second step, however, consists of the administration of alliin, a naturally occurring, inert non-protein amino acid that is converted to the cytotoxic allicin only on interaction with the enzyme, anchored on the tumor cell surface. The system presented here mimics the natural situation in crushed garlic cloves, where allicin formation occurs only after alliin becomes accessible to the enzyme.

The advantages of this approach over the conventional ADEPT are the following: allicin and its precursor alliin which is nontoxic (unlike other drugs) are natural food constituents. Allicin acts swiftly as a powerful anticancer agent and it has an extremely short lifetime. Additionally, its secondary products such as allylmercaptoglutathione and allylmercaptocysteine were shown to bear beneficial antioxidant characteristics (4). Thus, a comprehensive expression of the potent biological activity of allicin is achieved.

3-(2-Pyridyldithio) propionic acid N-hydroxysuccinimide ester (SPDP), DTT, and PLP were purchased from Sigma (St. Louis, MO). Alliin was synthesized according to Ref. (15). Allicin was prepared by applying synthetic alliin onto an immobilized alliinase column (16). Allicin concentration was determined as described in Ref. (17). 2-Nitro-5-thiobenzoate (NTB) was synthesized according to Ref. (18). [Methyl-³H]thymidine was purchased from Amersham Pharmacia Biotech (Amersham, United Kingdom).

Protein concentration was measured at 280 nm, using E₂₈₀ = 77.000 m⁻¹ cm⁻¹ (E₂₈₀^0.1% = 1.54) for alliinase and 210.000 m⁻¹ cm⁻¹ for purified mAb (E₂₈₀^0.1% = 1.4). The number of SPDP residues on the modified proteins was determined according to Ref. (19). Purification of alliinase was done according to Refs. (20, 21). Quantitative assessment of alliin and allicin was performed by high-performance liquid chromatography (17). Determination of alliin content in the serum was done after i.p. injection of alliin (0.2 ml of 30 mg/ml) to mice. Serum aliquots (25 μl) were treated with 90% methanol and proteins precipitated overnight at −20°C. After centrifugation, the supernatants were lyophilized. The dry material was dissolved in water and was assayed for alliin according to Ref. (22).

Alliinase activity was determined by the NTB method in a cell free system according to Ref. (17). An ELISA system using 96-well plates was developed to assay the activity of the enzyme conjugates. Wells were coated with either protein A (5 μg/ml), or with paraformaldehyde-fixed N87 or CB2 cells, at a sub-confluent culture density. Fixation was done with paraformaldehyde (3%) in PBS for 30 min, followed by washing (3×) with PBS. Binding of mAb-alliinase conjugates to the antigen was done at 24°C for 1–2 h in the presence of 2 mm PLP. After removal of unbound protein by washing (3×) with PBS containing 0.05% Tween 20, the activity of adsorbed conjugated alliinase was determined by using NTB (3 × 10⁻⁴M) and alliin (0.6 × 10⁻³ m) in 50 mm phosphate buffer (pH 7.2) containing 2 mm EDTA (0.1 ml/well). The decrease in A_{412 nm} was recorded after 30 min. Wells, to which no enzyme was added, served as controls (17). One unit of activity was defined as the amount of enzyme required to produce 1 μmol of pyruvate per minute.

N28 (IgG1) mAb, anti-ErbB2 (23, 24), was purified from ascites fluid by using caprylic acid (25) or by affinity chromatography on immobilized protein A. F(ab)₂ was prepared from N28, by pepsin digestion (1 mg pepsin/15 mg antibodies) in 0.1 m citrate-phosphate buffer (pH 3.7), for 4 h at 37°C, followed by Superdex-200 (XK16x70) size exclusion chromatography in PBS. HCV-AB^xtl68, a purified human mAb against the E2 protein of the hepatitis C virus, was used to prepare a non-relevant control conjugated with alliinase.

SPDP modification of alliinase: SPDP (50 mm in ethanol) was added at a 5-m excess to alliinase in 25 mm Na phosphate (pH 6.5), containing 50% glycerol for 1 h at room temperature. Excess SPDP was removed by gel filtration [Sephadex G-25 equilibrated with 50 mm phosphate buffer (pH 6.5), containing 10% glycerol]. The protein was collected and concentrated by Centriprep-30 (Amicon, Beverly, MA) at 4°C. The degree of SPDP-alliinase modification was 1–1.4 residues SPDP/E. The modified protein was stored in 50% glycerol at −20°C. SPDP modification of mAb: The only difference from the procedure described above for alliinase was a 20-m excess of SPDP over the mAb. The degree of SPDP-mAb modification was 3–4 SPDP residues/mAb, and the modified mAb was stored in PBS at 4°C.

SPDP-mAb was reduced with DTT (5 mm, 10–20 min). DTT was removed by gel filtration. Protein-containing fractions (mAb-SH) were collected and immediately combined with SPDP-alliinase, at a molar ratio of SPDP-alliinase/mAb-SH (1.1/1). The conjugation mixture was incubated 1 h at 24°C with 20% sucrose, adjusted to 50% glycerol, and stored at −20°C. Before any further treatment , the conjugated alliinase was freed from sucrose and glycerol by gel filtration. Further purification of conjugated proteins from non-conjugated was done by size-exclusion chromatography on Superdex G200 (1.6 × 60; Pharmacia) with 50 mm phosphate buffer (pH 6.5), containing 10% glycerol and 2 mm PLP.

Iodination of alliinase, free mAb, and conjugates was done with ¹²⁵I, using the Chloramine-T method (0.5 mCi Na¹²⁵I/100 μg protein) according to Ref. (26). The specific radioactivity of ¹²⁵I-alliinase, ¹²⁵I-mAb, and ¹²⁵I-mAb-alliinase was 0.51, 2.26, and 0.85 μCi/μg protein, respectively.

N87, a human gastric adenocarcinoma cell line, expressing the ErbB2 receptors (27), HeLa HtTA-1 (28), BHK, baby hamster kidney cells, and NIH 3T3 cells were grown in DMEM supplemented with antibiotics and 10% heat-inactivated FCS; CB2, a Chinese hamster ovary (CHO) cell line transfected with mammalian expression vector ErbB2 (29), was grown in DMEM/F12 (1:1) medium supplemented with antibiotics and 10% heat-inactivated bovine calf serum; 32D, a murine hematopoietic progenitor cell line (30), devoid of the ErbB2 receptors, was grown in suspension in RMPI 1640 supplemented with glutamine, antibiotics, IL-3, and 10% FCS (31).

[Methyl-³H]thymidine (Amersham) incorporation into DNA was measured in 96-well plates. All the experiments performed with cells were carried out in triplicates. CB2 or N87 adhesive cells (1,000–20,000 cells/0.1 ml/well) were grown at 37°C for 6 h after seeding. Growth medium was replaced by fresh medium containing 2 mm PLP. Free mAb or mAb-conjugate was added to the cells for 1–2 h. Cells were then washed (3×) with medium containing 0.4 mm PLP and resupplied with growth medium containing 0.02 mm PLP. 32D cells were grown in suspension in 12-well plates (200,000 cells/ml). Treatment with mAb or conjugates proceeded as described above. Unbound proteins were washed off with medium containing PLP by centrifugation (1200 rpm, 5 min). Treated cells were resuspended with fresh medium and reseeded in 96-well plates (20,000 cells/well). From this stage on, both adhesive and non-adhesive cells were supplemented with [methyl-³H]thymidine (0.6–0.8 mCi/well) and incubated in the presence or absence of alliin (0.6–1.2 × 10⁻³ m). After 16 h incubation at 37°C, plates were frozen (−20°C, 1 h), and harvested following trypsin treatment. Cell viability was determined with trypan blue (0.025%) for 10 min in 24-well plates.

All animal studies and protocols were approved by the Weizmann Institutional Animal Care and Use Committee (WIACUC).

Tumors were generated in Female CD-1 nude mice (6 weeks old) by s.c. injection of N87 (3–5 × 10⁻⁶ cells/mouse) into the back of the animals. Three to 5 days after tumor cell implantation (early tumors), mice were grouped (six to eight/group) according to mode of treatment and subsequent alliin administration (+/−): (1) PBS/−; (2) PBS/+; (3) anti-ErbB2 (N28)/+; (4) anti-ErbB2-alliinase conjugate/−; (5) anti-ErbB2-alliinase conjugate/+; (6) alliinase/+; (7) HCV-AB^xtl68-conjugate/+. All the mice were constantly supplemented with pyridoxine (vitamin B6) in their drinking water (100 mg/l). PLP was injected i.p. (0.2 ml, 20 mm in PBS/mouse) 30 min before each i.v. injection of the alliinase conjugate or its controls. Anti-ErbB2, anti-ErbB2-alliinase conjugate, and HCV-AB^xtl68-alliinase were used at 20 μg mAb/mouse, and alliinase control at 20 μg/mouse. Animals receiving all modes of treatment and controls were injected 5× at 3- to 4-day intervals. Alliin (0.2 ml of 30 mg/ml) was injected i.p. into groups 2, 3, and 5–7 at a 7-h interval, twice a day, during 2–4 weeks. Tumor volume was measured every second day, by using calipers. Mean tumor volume of the different groups was compared by using variance analysis. Due to NIH guidelines and the Animal Welfare Act, mice were sacrificed as soon as the tumor reached 1.5 cm in volume.

This study was carried out with N87-treated nude mice 3 weeks after tumor cell implantation. Mice were injected i.v. with either ¹²⁵I-alliinase, ¹²⁵I-anti-ErbB2, or ¹²⁵I-anti-ErbB2-alliinase conjugate (1 × 10⁶ cpm/mouse). At various time intervals, two mice were sacrificed; organs were excised, washed with PBS, weighed, and their ¹²⁵I content was determined.

Results are shown as the mean values ± SD. The in vitro results are average data of triplicate experiments. The in vivo experiments with anti-ErbB2 and anti-ErbB2-alliinase and PBS have been performed at least twice with similar results. ANOVA and Tukey test were used to analyze statistically significant differences between the groups.

The effects of the inert substrate, alliin, and the active product, allicin, on CB2 cells in culture were determined at different concentrations. The extent of CB2 cell proliferation in the presence of allicin was evaluated by following [³H]thymidine incorporation after incubating both for 16 h at 37°C. Cells were harvested and the amount of [³H]thymidine incorporation was determined (Fig. 1). The results demonstrate that pure allicin inhibits DNA synthesis in a dose-response manner; 50% inhibition is obtained at about 13 μm and 90% inhibition at about 25 μm. A dye exclusion test corroborated the above results. CB2 and N87 cells were incubated for 16 h in the presence of alliin or allicin, and viewed under a light microscope after staining. Lethality was observed at allicin concentrations of about 20–30 μm, whereas alliin was nontoxic at concentrations up to 200 μm.

Covalent conjugates of anti-ErbB2-alliinase were prepared by binding the enzyme to the anti-ErbB2 or to its F(ab)₂ derivative, yielding 1–2 alliinase/mAb (mole/mole). The conjugates of anti-ErbB2-alliinase maintained their target recognition of live and fixed tumor cells as well as their alliinase activity, as determined by allicin formation (Fig. 2A). The activity of the conjugates bound to fixed cells was concentration dependent (Fig. 2B).

We used anti-ErbB2 to establish a general approach of targeted cell killing, employing N87 (a human gastric tumor cell line) and CB2 (a Chinese hamster ovary cell line) expressing ErbB2 receptors, and exploiting the fact that this monoclonal anti-ErbB2 is not internalized after binding to its surface receptor (24). However, on binding to the cell, this antibody enhances cell proliferation at a rate of about 10–20% (24). Therefore, the suppression obtained by targeted alliinase-anti-ErbB2 in the presence of alliin must be compared to that without alliin, rather than to a PBS control. Binding of the anti-ErbB2-alliinase conjugate to ErbB2-expressing cell monolayers (CB2) did not impair the accessibility of alliin to alliinase, and cell proliferation was inhibited on addition of the substrate. The inhibition of [³H]thymidine incorporation in CB2 was dose dependent (Fig. 3A).

In the search for a cell line that does not respond to anti-ErbB2 conjugate, we examined several adherent cells such as HeLa (human), BHK (hamster), and NIH-3T3 (murine). We have screened these cells for the presence of ErbB2 receptors. After cell treatment with the conjugate, two techniques were used to assess its effects: determination of alliinase activity on paraformaldehyde-fixed cells, as well as of thymidine incorporation in live cells. Comparing the above three cell lines to N87 and CB2, we found that the former exhibited a certain degree of conjugate recognition (5–15% of the latter). Therefore, we had to resort to a non-adherent cell line, 32D, devoid of ErbB2 receptors (31). Thymidine incorporation resulting from the specific binding of the conjugate to N87 and to 32D cells was examined. Cells were treated with either anti-ErbB2 or with alliinase-anti-ErbB2 conjugate, in the presence and absence of alliin. Inhibition of [³H]thymidine incorporation in N87 treated with the conjugate in the presence of alliin was higher than 70%, while no inhibition was observed in 32D cells under the same conditions (Fig. 3B). N87 and CB2 cells, treated with mAbN28-alliinase and alliin for 16 h, lost viability as observed from a dye-exclusion test (data not shown). It is worth noting that cells treated with anti-ErbB2 (mAbN28) or anti-ErbB2 alliinase conjugates in the absence of alliin, show accelerated tumor cell growth as previously described (24).

The ability of the anti-ErbB2-alliinase conjugate to bind selectively to tumor cells in vivo was evaluated by using ¹²⁵I-labelled anti-ErbB2-alliinase in mice injected s.c. with human N87 tumor cells. Due to the virulent nature of N87 cells, and their quick proliferation, the time schedule of inhibition experiments and distribution experiments was different. Whereas a large enough time interval was needed to monitor tumor growth and the model studied was an early tumor, the distribution could be performed on mature tumor. Two to 3 weeks after tumor implantation, mice were injected i.v. with the conjugate ¹²⁵I-anti-ErbB2-alliinase, or with its separate two components, ¹²⁵I-anti-ErbB2 and ¹²⁵I-alliinase. The bio-distribution of ¹²⁵I in various organs of the treated mice after 24 h is shown in Fig. 4. After ¹²⁵I-anti-ErbB2-alliinase injection, there is a specific accumulation of ¹²⁵I-labeled protein in the tumor, as compared to other organs. Maximal accumulation of the conjugate occurred after 24 h, and its half-life in the tumor was about 72 h. However, the half-life of the conjugate in the blood was less than 8 h. A similar bio-distribution pattern was observed for the ¹²⁵I-anti-ErbB2-treated mice, but the half-life in the tumor was around 5 days (half-life in the blood was less than 8 h, as in the case of the conjugate). ¹²⁵I-labelled alliinase did not accumulate in the tumor (Fig. 4). The half-life of alliin in the serum was about 4–5 h; therefore, it was administered twice a day.

The effect of the mAb-enzyme conjugate, with and without subsequent alliin administration, on tumor growth in vivo was studied using a human xenograft in mice. An early model was used, in which treatment with conjugates started 3–5 days after tumor cells were injected. The control group treated with PBS (±alliin) represents the non-treated baseline. The rate of tumor growth, in mice treated with the anti-ErbB2-alliinase conjugate, followed by successive administrations of alliin (group 5), slowed down significantly (P < 0.001) as compared to alliin-supplemented control mice that were not treated with conjugate (PBS). Other controls that did not suppress the kinetics of tumor growth were anti-ErbB2 (N28) alone and the anti-ErbB2-alliinase conjugate, without subsequent alliin administration (groups 3 and 4). These controls caused an even somewhat enhanced tumor growth, which actually magnifies the effect of the treatment (Fig. 5). To test the high specificity of targeted allicin production, two more controls supplemented with alliin were tested, and none suppressed tumor growth: alliinase alone and alliinase conjugated to a nonrelevant mAb (HCV-AB^xtl68) (groups 6 and 7, respectively). Alliin itself had no effect on tumor size as seen from control mice (PBS, groups 1 and 2).

Since its discovery in 1944, allicin has been considered as a potent antimicrobial agent (1). A decade after its discovery, its antiproliferative and cytotoxic activities started to be the object of keen interest (32). However, allicin was shown to be unstable and short-lived. Furthermore, after injection, its fast disappearance from the circulation thwarted scientists from attempting to use it as a chemotherapeutic agent.

Here we describe a novel approach designed to overcome features of allicin that hampered its use in vivo. The procedure consists of a two-step process including one that targets alliinase to the surface of tumor cell, followed by another, that makes use of the inert stable compound, alliin. Alliin injection imitates the situation occurring in the crushed garlic clove. Alliin undergoes conversion into allicin in situ, at the location of the targeted alliinase.

In vitro experiments demonstrate that the enzymatic activity of conjugated alliinase, targeted to cell surface, is preserved. The distribution experiments show that the anti-ErbB2 mAb (N28) conjugated to alliinase can specifically target the enzyme to the tumor. Subsequent periodic administrations of the substrate, alliin, to the conjugate-treated mice enable its conversion into allicin in situ, which results in tumor growth arrest. Because mammalian cells do not produce alliinase, alliin is converted into allicin only by the localized conjugate. Using this procedure, we demonstrate, for the first time, the in vivo anticancer effect of allicin.

The natural allicin generating system has several advantages over the traditional ADEPT approach: (a) an extremely low toxicity of alliin, which is a common constituent in our diet, and the convenience of its oral administration in humans; (b) a specific and high anticancer activity of allicin, which is continuously produced in situ; (c) a restricted area of allicin production combined with its short life duration in the body; and (d) the conversion of excess allicin into beneficial derivatives such as allylmercaptocysteine and allylmercaptoglutathione that possess antioxidant and SH-modifying activities (4).

Because the animal model imposes some ethical restrictions, our results endorse, at least for the moment, allicin treatment as a supplementary therapy to surgery and as a preventive of metastases.

This pioneering attempt to use targeted-allicin production in cancer therapy will be followed by further experiments that should investigate the potential of this approach. Various tumors and their related surface recognition molecules will be examined.

In addition to growth inhibition of any type of cancer cell, the targeted-alliinase principle can be used to eliminate pathogens. The only requirement is the availability of highly specific mAbs, or other kinds of target-specific carriers. Ligation of such carriers to alliinase can be done by using chemical, or recombinant fusion methodology.

The mAb N28 to ErbB2 receptor was kindly provided by Dr. E. Hurwitz; CB2, N87, and 32D cell lines were kindly obtained from Prof. Y. Yarden of the Weizmann Institute Rehovot, Israel. The mAb HCV-AB^xtl 68, a purified human mAb against the E2 protein of the hepatitis C virus, was kindly provided by XTL Biopharmaceuticals Ltd., Nes Ziona, Israel.