Chemopreventive properties of chlorophylls towards aflatoxin B₁: a review of the antimutagenicity and anticarcinogenicity data in rainbow trout

Abstract

The anticarcinogenic activity of chlorophyllin (CHL), a water-soluble derivative of chlorophyll, was first reported in rainbow trout. This review describes certain experiments which set the stage for long-term tumor bioassays, in trout and other species, using CHL and various food-borne carcinogens. Initial work with trout and rat liver enzymes in the Salmonella assay showed that CHL was a potent antimutagen towards heterocyclic amines, polycyclic aromatic hydrocarbons, aflatoxins and other classes of mutagen. Antimutagenic activity was further demonstrated using the corresponding direct-acting mutagens in the absence of an exogenous metabolizing system. Mutagen-inhibitor interaction (molecular complex formation) was identified in spectrophotometry studies, suggesting that CHL acts as an ‘interceptor molecule’. In vivo, CHL reduced hepatic AFB1-DNA adducts and hepatocarcinogenesis when the inhibitor and carcinogen were co-administered in the diet. Finally, co-injection of inhibitor and AFB1 into trout embryos established that CHL was more effective than chlorophyll a in reducing AFB1-DNA adducts 2 weeks after injection, and liver tumors after 1 year.

1. Introduction

Waters et al. [1] recently reviewed an extensive database on the activities of several antimutagens in vitro and in vivo, and summarized the results in the form of detailed ‘activity profiles’. Among the various inhibitors reviewed, chlorophyllin (CHL) was identified as almost uniformly protective against a broad range of direct- and indirect-acting mutagens, including aflatoxins, polycyclic aromatic hydrocarbons, heterocyclic amines, alkylating agents and several miscellaneous compounds. Close inspection of the data presented in the review indicates that the vast majority of experiments using CHL were from assays conducted in vitro, and in many cases these were descriptive rather than mechanistic studies. The antimutagenicity profile for CHL contained only 7 out of a total of 93 data points that were from experiments conducted in vivo. The limited nature of the in vivo data notwithstanding, most of the evidence strongly supported a protective role for CHL [1].

In the time since the review by Waters et al. [1] was submitted and published, several new in vivo studies with CHL have been reported. Perhaps most notable among these is the work that showed CHL to be a potent inhibitor of aflatoxin B₁ (AFB₁)-initiated tumorigenesis in rainbow trout [2]. This was the first evidence that CHL could operate not only as an antimutagen, but also as an anti carcinogen in some experimental protocols. The present paper reviews several of the key experiments with CHL and AFB₁ in trout. In the context of this Special Issue on ‘The Use of Fish and Fish Transgenics in Genotoxicology Studies’, the review serves to illustrate several of the advantages and limitations of rainbow trout as a model for investigating modulators of tumorigenesis and genotoxicity in vivo.

2. In vitro antimutagenic activity of CHL vs. AFB₁ and AFB₁-8,9-epoxide

The first studies with CHL in our laboratory used the Salmonella assay, and were based on preliminary experiments which examined the antimutagenic activity of CHL vs. AFB₁ [3]. A key observation in the studies by Whong et al. [3] was that the inhibitory activity occurred only when CHL and AFB₁ were incubated concurrently in the assay, suggesting that direct interaction between the inhibitor and mutagen might be important for antimutagenic activity. We observed the antimutagenic activity of CHL vs. AFB₁ using trout liver microsomes as the source of metabolic activation [4], but showed the inhibitor to be more effective against the direct-acting mutagen AFB₁-8,9-epoxide (Fig. 1a). These results indicated that the in vitro antimutagenic activity of CHL was not dependent on the presence of a metabolic activation system, and supported an inhibitory mechanism involving direct interaction between CHL and the mutagen, such as electrophile-scavenging or molecular complex formation.

Fig. 1.

Summary of the antimutagenic and anticarcinogenic effects of CHL in trout. (a) and (b) were modified from the original results in Ref. [4] and elsewhere (R.H. Dashwood, unpublished data), whereas (c) was obtained by re-plotting of the original findings in Ref. [2].

3. Inhibition of AFB₁-DNA binding in vivo

Previous work in our laboratory with AFB₁ and the anticarcinogen indole-3-carbinol established that AFB₁-DNA adducts increased linearly with time of carcinogen exposure during a 2-week dietary treatment of trout, and that the slopes of the DNA binding curves decreased with concentration of inhibitor in the diet [5]. Similar results were obtained in studies with CHL, in which trout were co-treated with AFB₁ and the inhibitor for 7 days (Fig. 1b). In these experiments, CHL protected against AFB₁-DNA adducts in a dose-related manner, and the AFB₁-DNA binding levels were inhibited by approximately 70% at a dietary CHL concentration of 2000 ppm [4]. The results from these studies provided the impetus for more extensive experiments in trout that examined the quantitative interrelationships between AFB₁ carcinogen dose, CHL inhibitor dose, target organ DNA adduction and final tumor response (molecular dosimetry studies).

4. Inhibition of AFB₁-DNA adducts and hepatocarcinogenesis (molecular dosimetry studies)

Rainbow trout (1–2 g body weight) were exposed in the diet to 0, 500, 2000, or 4000 ppm CHL and, within each inhibitor group, to one of four different doses of AFB₁. The carcinogen was administered in tritiated form during a 2 week period and 15 fish were selected at random from each tank in order to determine AFB₁-DNA binding levels in the liver. The remaining animals were switched to control diet and after 9 months 100 of the fish in each tank were selected at random in order to determine the incidence of liver tumors. Full details of the experimental conditions and various statistical analyses are included in the original report [2]; for the purposes of this review, we have simply plotted each data-point from the tumor study vs. the corresponding AFB₁-DNA binding data point (Fig. 1c). A straight line relationship was obtained between the 9-month liver tumor incidence and the AFB₁-DNA adducts levels measured after the last dose of AFB₁. The overall effect of the CHL treatment was to shift the data points towards the origin with each increase in inhibitor dose, along a line of slope equal to approximately 1% tumor incidence per μmol AFB₁/mole DNA (see dotted line in Fig. 1c). These results extended our previous molecular dosimetry studies with AFB₁ and indole-3-carbinol [6], and suggest that CHL operates during the initiation phase as an effective ‘blocking agent’ against DNA damage. Mechanism studies conducted both in vitro and in vivo indicated that the blocking activity might involve direct interaction between the carcinogen and the inhibitor, i.e., that CHL operates as an ‘interceptor molecule’.

5. Interceptor molecule hypothesis

Hartman and Shankel [7] reviewed the literature for inhibitors that interact directly with mutagens and carcinogens and serve in effect to ‘sequester’ these compounds from harm’s way. By reducing the bioavailability of many genotoxins, these putative ‘interceptor molecules’ might represent an important first line of defense, perhaps rivaling such mechanisms as induction of detoxification enzymes or inhibition of carcinogen activating enzymes. In the period since the review by Hartman and Shankel was published, a number of papers has described the ability of CHL, chlorophylls, and other porphyrins to form molecular complexes with various carcinogens and mutagens in a manner consistent with the interceptor molecule hypothesis [8–14]. We will briefly review the evidence supporting the formation of a molecular complex between CHL and AFB₁. However, the reader is referred to other work that describes the interactions formed between heterocyclic amines, polycyclic aromatic hydrocarbons, or a number of miscellaneous polycyclic planar mutagens and CHL, chlorophylls, or related porphyrins [8–14].

6. AFB₁-CHL complex formation

Evidence for the formation of an AFB₁–CHL complex was first obtained in titration studies using a double-beam spectrophotometer. With the addition of AFB₁ to the sample cuvette followed by subsequent additions of CHL to both cuvettes, quenching of the absorption spectrum of AFB₁ occurred in a manner consistent with complex formation between the two molecules (Fig. 2). Because both compounds have absorption maxima in the region 200–400 nm, which might cause interference during spectrophotometric titrations, our next experiments with CHL examined the quenching of AFB₁ fluorescence [12]. Scatchard analysis provided evidence for a highly favored complex between AFB₁ and CHL (K_d = 1.4 μM), although this interaction was influenced to some extent by the presence of a metal ion in the porphyrin. Specifically, the complex with AFB₁ was slightly less favored with protoporphyrin IX, an intermediate in the biochemical synthesis of heme, than the metal-containing copper or zinc chlorophyllins [12]. These findings were generally consistent with published data for several antimutagens in the Salmonella assay, in which the order of inhibition was as follows (most potent to least potent): metal-containing porphyrins > free-base porphyrins > ring-opened tetrapyrroles [15].

Fig. 2.

Spectrophotometric titration of AFB₁ with CHL. Starting substrate (AFB₁)concentration, 20 μM in 0.1 M sodium phosphate buffer, pH 7.4; 25°C; ligand concentration, 7.5 μM CHL (each addition); b = 1 cm. From the Benesi–Hildebrand plot, K_d = 1.92 μM (see Ref. [8] for details of the methodology).

Interestingly, the interaction between CHL and AFB₁ was much less susceptible to pH than were the complexes formed between CHL and heterocyclic amine mutagens, which were most stable at pH 7.4 [8]. Computer modeling suggested that this was related to the electrostatic docking of charged carboxyl groups on CHL with the exocyclic amine on the mutagen [14]. However, the most important force stabilizing the complexes between CHL and most aromatic mutagens, including AFB₁, involved the formation of numerous van der Waals interactions [11,12,14]. As shown in Fig. 3a, computer-generated models of the AFB₁-CHL complex provide evidence for the presence of multiple overlapping ring systems, which is an arrangement common to all π–π complexes [16–18]. In our studies, several different orientations were examined for AFB₁ and CHL, and the calculated energies indicated highly favored complexes (−16 to −21 kcal/mol). In the most favored orientation, the 8,9-double bond of AFB₁ was arranged 180° away from the CHL carboxyl groups (Fig. 3b).

Fig. 3.

Computer generated molecular model of the AFB₁-CHL complex. Energy-minimized molecular models of the interaction between AFB₁ and CHL were obtained using HyperChem, as described previously for the complexes between chlorophylls and heterocyclic amines [14] or polycyclic aromatic hydrocarbons [11]. (a) Top view showing the overlapping ring systems of AFB₁ (yellow) and CHL; (b) side view showing the 8,9-double bond of AFB₁ (upper left) oriented 180° with respect to the carboxyl groups in CHL (right); minimized energy of the complex shown, −19.023 kcal/mol.

7. Activity of natural chlorophylls in vitro and in vivo

Computer-modeling studies indicate that the tetra-pyrrole macrocyclic is the most important feature for complex formation with AFB₁, and that natural chlorophylls should be effective inhibitors. However, in contrast to CHL, which is highly water soluble, lipophilic natural chlorophylls complex less effectively with carcinogens under the aqueous conditions of many in vitro assays. A mixture of chlorophylls a and b was shown to exhibit antimutagenic activity in the Salmonella assay by a mechanism involving complex formation [19], and purified chlorophyll a protected against the formation of DNA adducts in isolated hepatocytes by interfering with carcinogen activation [20].

In vivo genotoxicology studies with natural chlorophylls are particularly scarce. Most investigations have tended to avoid the use of highly purified natural chlorophylls for reasons of chemical stability and cost, chlorophyll being approximately four orders of magnitude more expensive than CHL and thus prohibitive in most whole-animal tumor models. Chlorophyll extracted from spinach leaf failed to inhibit the clastogenic activity of potassium dichromate in mice, and was in fact directly clastogenic when administered at a dose of 1.5 mg/kg body weight [21].

In our laboratory, a pilot study was undertaken to compare the anticarcinogenic activities in trout of chlorophyll a, CHL, or highly purified copper–chlorin e₆ (C₆). Approximately 100–150 trout embryos, 21 days post-fertilization, were co-injected with test inhibitor and tritium-labeled AFB₁ or with radiolabeled carcinogen alone. Twenty-four hours later, 3 pools of 5 embryos were selected at random and the levels of total AFB₁-DNA adducts were measured according to previously described methods [22]. The remaining eggs were allowed to hatch and the incidence of liver tumors was determined after 9 months.

As shown in Table 1, inhibitors were tested at three different dose levels (10, 100, or 1000 μg/egg), but none offered protection that was clearly dose-related. Inspection of the data for AFB₁–DNA binding indicated that chlorophyll a was the least effective, and inhibited by <20% at all doses, whereas the water-soluble compounds Ce₆ and CHL inhibited by 17–43% and 31–72%, respectively. These results might reflect the relative distribution of each test inhibitor after injection into the egg. Thus, during the treatment with chlorophyll a it was noted that the solution remained discrete at the site of injection (Fig. 4), whereas CHL and Ce₆ dispersed readily into the yolk sac.

Fig. 4.

Microinjection of chlorophyll a into an egg containing rainbow trout embryo 21 days post-fertilization. The egg was injected with 1 μl of vehicle (ethanol:DMSO, 1:1, v/v) containing 1000 μg chlorophyll a.

The 9-month tumor incidence data obtained for each of the two highest inhibitor doses reflected the poor survival of trout post-hatching, but in groups given 10 μg test compound the data were as follows (percent liver tumor incidence): AFB₁ controls, 73%; AFB₁ plus CHL, 50%; AFB₁ plus Ce₆, 55%; AFB₁ plus chlorophyll a, 65% (Table 1). Thus, the relative order of inhibition from both the DNA binding and tumor incidence data was CHL > Ce₆ > chlorophyll a.

Table 1.

The effects of various chlorophylls on AFB₁-DNA adduction and tumorigenicity when co-injected into rainbow trout embryos^a

Chlorophyll	Dose (μg)	pg AFB1/mg DNAb	Inhibition (%)	Tumor incidence (%)c
None	–	170 ± 25	0	105/144 (73%)
CHL	10	103 ± 24	39	21/42 (50%)
	100	117 ± 11	31	21/43 (49%)
	1000	47 ± 4	72	[21/25]
Chlorin e6	10	138 ± 9	19	24/44 (55%)
	100	142 ± 24	17	[12/20]
	1000	97 ± 14	43	[12/28]
Chlorophyll a	10	142 ± 24	17	32/49 (65%)
	100	164 ± 16	4	[8/9]
	1000	143 ± 10	16	[8/9]

^aFor details of the methodologies for assessing AFB₁-DNA adducts and liver tumor incidence, see Ref. [22].

^bMean ± SD.

^cSquare brackets indicate groups with poor survival.

We are cautious about the full interpretation of these data due to concerns about the toxicity of high doses of chlorophylls to the embryo. These inhibitors also might retard AFB₁ and cause toxicity by delaying the release of carcinogen from the egg. Increased retention of AFB₁ in the egg might be offset by direct inhibition of carcinogen bioactivation, but the dose-dependencies of these two opposing effects would not necessarily match. Moreover, the distribution profiles of AFB₁ and chlorophylls in the egg might differ significantly from those in the gut following an oral dose, which involves the inhibition of carcinogen uptake and increased elimination via the feces [23,24]. Evidently, the embryo model has the advantage of providing a carcinogenicity bioassay system that enables the testing of inhibitors that are expensive (chlorophyll a) or are available in short supply as the highly purified compound (Ce₆). However, further work is necessary to better define the pros and cons of using the trout embryo model for investigations of the anticarcinogenic properties of chlorophylls.