As a result of the largely negative results obtained in the previous work, it was decided to initiate a new experiment with a fresh approach. The aim was to obtain an assessment of the in vivo activity of GST in response to the toxicant, chlorothalonil (2,4,5,6-tetrachloroisophthalonitrile, TCIN). The development of a technique for analysing concomittant changes in GST and GSH was considered to be useful. This was because of the close inter-relationship between the two in response to xenobiotics that was identified in the preceding literature review (see Section 2).

Chlorothalonil has also been shown to induce GST activity towards CDNB (Davies, P.E., 1985), and so it was chosen as a test substrate to see if similar modulations in mussel GST and GSH could be observed at concentrations known to be potentially toxic to fish. To give an idea of changes in in vivo GST activity, the Km of gill GST was also determined.

A total of 178 Swan Mussels (Anodonta cygnae) were purchased from a commercial supplier (World of Water, Reading). Two groups were available, the first consisted of large individuals which had arrived that morning, the second consisting of smaller individuals which had been held as stock for several days. Mussels from the first group were chosen preferentially, although around 50 to 60 had to be taken from the second group. It is worth noting that the individuals from the large population used in this study were very much (5 times) larger than those used in previous GST work with Anodonta (Polak, M., 1992 and Johnson et al., 1992). On arrival they were placed into a 280 l stock tank containing aerated, flowing ground water. The largest 168 were numbered and randomly assigned to the treatments given below. The remainder were used in development of the procedures.

1. Gill water content

2. Gill GST Km

3. Response of GST and GSH to a chlorothalonil challenge.

Chlorothalonil was purchased from Parker Hart, Surrey, as the commercial formulation Daconil Turf (of which it is the active ingredient). Although there are many potentially toxic additional constituents of a commercial formulation, this was considered acceptable since, (1) it is the toxicity of the product as it appears in the environment which is of the greatest interest, and (2) in previous studies no significant difference has been found between the 96 hr LC₅₀ of technical (97.5%) chlorothalonil and the commercial formulation Bravo 500 (Ernst et al., 1991).

For the experiment, tanks were filled with 30 l of groundwater at the appropriate concentration and 36 mussels were placed in each tank. The length of the procedures involved in dissection and preparation of tissue meant each treatment group was started at intervals through the day (in random order). Each day the water was replaced by freshly contaminated groundwater to minimise loss due to metabolism, tissue binding and abiotic degradation. This was felt to be sufficient to maintain constant levels since mussels have low xenobiotic metabolism (Goldberg, E.D., 1986), do not bioconcentrate chlorothalonil when exposed to low concentrations (Ernst, W. et al., 1991), and the loss of chlorothalonil in sterile water is negligible in 24 hours (Walker, W.W. et al., 1988). On days 4 and 7 after the initiation of the experiment 12 mussels were removed from each tank and treated as described below. The length of the procedures was such that it was invariably impossible to complete them within the time alloted before the next group was due to be prepared. The error was minimal as a proportion of total treament time on days 4 and 7, but on day 1 it was felt to be of sufficient importance that only 10 mussels were analysed from each group. In addition 12 mussels taken direct from the stock tank were analysed on day 0.

3.1. Determination of reduced glutathione.

From a survey of the literature it was found that essentially four methods have been used to determine GSH previously. Firstly, a method has been described which follows the reduction of DTNB (5,5'-dithiobis-(2-nitrobenzoic acid), Ellmans reagent) by GSH giving a colour change at 412 nm (Moron, M.S., Depierre, J.W., Mannervik, B., 1979). The second approach is an enzymatic modification of the first, with the GSSG produced being recycled to GSH by glutathione reductase (Owens, C.W.I., Belcher, R.V., 1965). The third technique is also enzymatic, but measures the formation of s-lactoyl-GSH by glyoxalase I (Akerboom, T.P.M., Sies, H., 1981). Finally, the fourth method involves the formation of a fluorescent complex with o-phthaldialdehyde (OPT) (Akerboom, T.P.M., Sies, H., 1981).

The fourth, fluorimetric, method was rejected because the assay procedure was too complicated for routine use. The third method was attractive as it is the only method able to provide an assay specific for GSH (the others will also respond to other soluble thiols, such as cysteine). It was rejected however, along with the second method, due to the difficulty in storing enzymes, and their cost. The second method has the additional disadvantage in that it responds to dithiols such as GSSG.

The first method provided a good compromise between ease of use and sensitivity. It was decided to use a similar approach but with CDNB as the substrate. Although the reaction with CDNB is much slower to develop, this substrate was chosen for two reasons;

• DTNB is known to react with a variety of reducing agents (Ellman, G., 1959). For example fresh phosphate buffer sometimes gives considerable colour (Owens, C.W.I., Belcher, R.V., 1965). The colour change with CDNB depends on direct conjugation of the thiol (Habig et al., 1974), and so is less susceptible to interference.

• The response of the CDNB-GSH conjugation had been well investigated in the previous studies, and so an assay could be developed with the minimum of preliminary work.

• It was intended that the assay would be used in conjunction with an assay for GST. Utilisation of a common substrate would streamline the procedure.

• CDNB is far cheaper than DTNB.

To improve the speed of the assay, a high pH was used (the GS^- anion has a higher reactivity than GSH). This has a side effect of producing a white precipitate of acid-soluble proteins (spectrophotometric analysis of redissolved precipitate gives a broad absorbance peak at 250-270 nm), the thiol groups of which could potentially interfere with the assay. For example metallothionein is an acid-soluble protein which contains a high level of sulfhydryl groups and can be induced by pollutants (Thomas, P., Wofford, H.W., 1984).

For analysis, samples were allowed to defrost for 25 minutes at room temperature. A 0.2 ml aliqout was added to a polycarbonate tube containing 0.77 ml 0.1 M K₂PO₄, 0.1 M KOH and 0.03 ml 50 mM CDNB (final assay conditions 20% extract, 1.5 mM CDNB, pH 9.45 (electronic pH measurement of pooled samples)). These were then centrifuged for 30 min at 3,300 g to remove protein precipitate and the supernatant decanted into plastic microcuvettes and left to stand overnight. Three replicates were made of each sample, and in addition 1 control was made with CDNB replaced by 0.03 ml ethanol. Only 1 control was used since 28 samples could be centrifuged simultaneously, and this allowed for maximum throughput.

Absorbance at 365 nm (the dABS maxima of CDNB-GSH at pH 9.45) was read after 24 hours using the program given in Appendix A. Any replicates which differed by more than 10% from the other two were discarded. The absorbance of the control sample was subtracted, as was the absorbance of 1.5 mM CDNB after 24 hrs at pH 9.45 (ABS=0.376 +/- 0.013, mean +/- sd, n=21).

The assay concentration was converted into tissue concentration by compensating for dilution:

i. Total sample volume = 2.5 ml + tissue water

= 2.5 ml + average tissue weight * water content

= 2.5 ml + average tissue weight * 74%

ii. Volume used in assay = 0.1 ml

iii. Dilution factor = sample volume / assay volume

3.2. Measurement of GST activity.

Statistical treament.

The data obtained each day was treated by one-way analysis of variance, using weight as a covariate. Where a significant difference was obtained, further analysis was conducted using Tukey's multiple range test. A significance level of 5% was chosen. Values for both GSH and GST were adjusted to value per gram of wet tissue with the formula;

Where, Y = adjusted value
a = original value
b = coefficient of variation (from analysis of covariation)
x = sample weight in grams
xbar = average sample weight in grams

No evidence was observed to suggest that data was positively skewed (the median was not consistently greater - or lower - than the mean), and so a logarithmic transformation of the data was not conducted.

1. Tissue water content.

The gills were found to have a water content of 74.0 +/- 1% This is comparable with previous data obtained for the fast and slow adductor of Mytilus edulis of 75% (Lange, R., 1972). The foot muscle contained 82.3 +/- 0.3% water. The most important point is the somewhat greater variability of gill water estimation. This is probably incurred during the blotting procedure, where the delicate nature of the gills gives greater scope for loss. Both the composition and volume of intracellular and extracellular fluid in swan mussels is critically dependent on the ambient water conditions (Lange, R., 1972).

2. Km of gill GST towards GSH.

The Km of gill GST was found to be dependent on mussel size. Of the nine mussels tested, 7 were large enough for 1g of gill tissue to be removed from each gill. From these, the Km was found to be 0.112 +/- 0.005 mM. From the other two, the Km was found to be 0.232 +/- 0.025 mM. This is not significant (two sample T-test, p=0.11), but the discrimination of samples this small is nearly impossible (p is less than 0.001 when the variance is pooled). The difference observed suggests two possibilities. Either the small mussels represent juveniles, in which case expression of GST isoenzymes varies with age, or they represent a genetically distinct strain. This is very possible, given that 2 different populations were probably used in the study, both of which were ill defined. Either way, the presence of such a wide variation in isoenzymic expression has serious implications for the homogeneity of the rest of the study.

Taking the modal Km of gill GST to be 0.112 mM, this is the lowest value detected in any species, and around one fifth of the Km of trout liver GST (Wallace, K.B., 1989). This suggests that GST detoxification is very highly developed in Anodonta.

3. Effect of chlorothalonil on GST and GSH

The data is presented in Appendix B, and is summarised in the tables below, and in Figures 6.1 and 6.2.

GSH (nmol / gram wet weight), mean + S.E.M.

* Significantly different from control (p<0.05)

GST (nmol / gram wet weight / sec), mean + S.E.M.

* Significantly different from control (p<0.05)

The control value obtained in this experiment, of around 625 nm/g, is at the lower end of the range of reported values. However the Figures are comparable with those obtained for channel catfish gills (Gallagher, E.P., Di Giulio, R.T., 1992) and black sea bass gills (Allen, P., Min, S.Y., Keong, W.M., 1988), and on this basis it seems likely that swan mussels should be at least as susceptible as fish to the toxic effects of chlorothalonil.

The activity of GST found in this study is comparable to the levels found previously in rat lung (46 nmol/g/sec), but an order of magnitude lower than that found in rat liver (930 nmol/g/sec) (Moron, M.S., DePierre, J.W., Mannervik, B., 1979). A figure for GST activity (as activity per gram wet weight) has not been reported in fish gills, since activity is usually measured as specific activity. GST activity in fish liver has been investigated in five fish species (Brachidanio rerio, Cyprinus carpio, Salmo gairdneri, Lepomis macrochirus and Poecilia reticulata), and values of 200 to 300 nmol/g/sec were reported - with the exeption of P. reticulata, in which activity was comparable to rat liver (Donnaruma, L., De Angelis, G., Gramenzi, F., Vittozzi, L., 1989).

This study has found a large rise in GSH levels in the low and middle concentration groups on day 1, followed by a swift return to control levels. This is unexpected on the basis of studies with fish, but is in concordance with other studies involving GSH depletion. Wenning et al. (1988) found an increase in ribbed mussel GSH after six hours exposure to the redox cycling agent paraquat, and levels remained elevated after 36 hours. Thomas and Wooford (1984) found that injection of paracetamol caused a minor decrease in AST in mullet liver after 3 hours, followed by rapid recovery and significant elevation by 24 hours. And Haloacetonitriles were found to initially cause a very rapid fall in rat liver GSH (t_min can be as little as 30 minutes after ingestion), followed by a rebound by 18 hrs (Lin, E.L.C., Guion, C.W., 1989). In light of these findings, it seems probable that there is an intial (undetected) decrease in GSH, which rapidly stimulates production to give enhanced levels. The dose dependent increase is characteristic of a negative feedback with positive gain. Homeostasis is achieved after four days, despite the continued presence of chlorothalonil. This suggests that utilisation of GSH in detoxification is rapidly compensated for by enhanced production.

In this study the highest dose group suffered a decline in GSH levels, followed by a slow restitution to give enhanced levels after 7 days. This may be a similar mechanism to that in the lower concentration groups, but with delayed rectification caused by toxicity of chlorothalonil. The slow recovery could be due to the large intial depletion of GSH, which will cause incapacition of enzymes due to oxidant stress and direct binding of chlorothalonil.

The most likely explanation for the drop in activity is the alternative role of GST in detoxification, the irreversible binding of xenobiotics. In vitro incubation of rat liver GST with a variety of chlorophenoxyalkyl acid herbicides (CPAs) results in a dose dependent inhibition of activity (Dierickx, P.J., 1983). Incubation of GST from three aquatic species with CPAs, quinones and o-chloranil also resulted in inhibition (Dierickx, P.J., 1984). Kinetic analysis revealed mixed type inhibition. Similarly haloacetonitriles also inhibit activity in vitro, and in vivo after 18 hours (Lin, E.L.C., Guion, C.W., 1989). Most importantly, Davies (1985) demonstrated that chlorothalonil can bind to GST when GSH concentrations are low. The binding is irreversible, probably covalent, and inhibits enzyme activity. Since the production of enzyme is energetically expensive, it seems logical that this defensive mechanism only comes into play when enzymatic detoxification has been overwhelmed.

On this basis the relationship of GSH and GST in the highest concentration group is perhaps significant. A decrease in GSH on day 1 is followed by an apparent decrease in GST activity on day 4. This delayed effect mimics the depletion of GSH followed by inhibition of GST shown in response to haloacetonitriles in rats (Lin, E.L.C., Guion, C.W., 1989). It is possible that the decrease in GSH represents an overwhelming of the antioxidant defence, leaving GST open to attack. Recovery follows and an increase in GST activity occurs when GSH levels are replenished.

In vivo activity of Glutathione S-transferase.

Since there is a complex interplay between levels of toxicant, GSH and GST activity, it may be advantageous to integrate these to give an indication of the actual effects on the ability of the organism to detoxify pollutants. The activity of an enzyme may be estimated by means of the Michaelis-Menten equation (Stryer, L., 1988);

V = Vmax * (S / (S + Km))

Where, V = turnover of the enzyme
Vmax = maximum turnover of the enzyme
S = concentration of the substrate
Km = Michaelis-Menten constant of the enzyme

To utilise this, the actual in vivo concentration of the substrate (GSH) must be ascertained. Since GSH is almost entirely intracellular, this can be determined by estimating the volume of intracellular fluid (ICF) in a tissue, and taking this to be the volume of distribution of the GSH in that tissue. The inulin space of Anodonta cygnae is around 55% of total tissue weight (Burton, R.F., 1983). Assuming this generalisation to hold for the gills, then the intracellular fluid is, 74% - 55% = 19% of total body weight. The concentration of GSH in the ICF can thus be estimated:

GSH per litre = (GSH per gram) / (ICF per gram) * 1000g

For a GSH content of 625 nM/g, this gives an concentration of around 3.2 mM in the ICF. Substituting this into the Michaelis-Menten equation (with a Vmax of 100% and a Km, as determined earlier, of 0.112 mM), this gives an activity of about 97% maximal. It can be seen that, despite the low levels of GSH observed, the affinity of Anodonta gill GST for GSH is such that it is operating at nearly maximal efficacy.

The activity of each individual mussel can be determined as above, and the average Figures thus obtained are plotted in Figure 6.3. It can be seen that the disturbance of GSH on day one has very little impact on the actual metabolic profile of the GST system, and so any variation in in vivo GST activity is due principally to variation in enzyme capacity.

It must be stressed that this estimation depends on the Km of GST for GSH with CDNB as a co-substrate. It is very possible that the Km will be dependent on co-substrate, especially if the co-substrate is preferentially metabolised by a different isoenzyme to that which attacks CDNB. So although no change in the in vivo activity towards CDNB has been observed, the activity towards other substrates may still be profoundly altered with changes in GSH concentration.

Chlorothalonil has an effect on the GST system of mussels at levels which are known to cause toxicity in fish. This means that these mussels can be used as biomarkers of environmental contamination, despite their apparent physiological insensitivity to toxins. There is an induction of GST activity which is evidently still increasing at 7 days. Future tests may benefit from being conducted over longer time periods to allow for maximal effect.

At high exposure concentrations there is an initial reduction in GST activity which may be related to an observed depletion of GSH. It is possible that this reduction of activity may be an indicative of a pathological effect. If this is so then the possibilities for the use of this test as a biomarker are very interesting - an induction of GST would signify pollutant exposure, whereas a reduction of GST would signify a toxic effect due to pollutant exposure. Further work to investigate a possible relationship between macroscopic endpoints and GST activity could be a fruitful avenue of research.

The effect on GSH found in Anodonta is less profound than in fish and this may be related to the reduced toxicity of chlorothalonil to mussels. Ribera et al. (1991) have shown that, whereas mussels contain more polyunsaturated fatty acids than rat liver, lipid peroxidation is no higher. They suggested as a result that antioxidant defences are very efficient in mussels (Ribera, D., Narbonne, J.F., Michel, X., Livingstone, D.R., O'Hara, S., 1991). The low Km for GST found in this study may reflect this. It seems likely that the tolerance of mussels is due to the rapid recovery of glutathione, and the high affinity of GST for GSH, which means that even a large depletion of GSH does not limit the ability of Anodonta to detoxify chlorothalonil.

It might be expected that the presence of two populations would seriously disrupt the normality of the data, especially if the mean value of these groups differed. For example, Ribera et al. (1989) found that male Mytilus have 50% more GSH in their gills compared with females. There is also a 30% decrease in Mytilus digestive gland GSH in older mussels (Viarengo, A., Pertica, M., Canesi, L., Accomando, R., Mancinelli, G., Orunesu, M., 1989). GST activity in non-polluted conditions is generally higher in animals of poor body condition (Johnson, I. et al., 1992). The distribution of the data is shown in Figures 6.4 and 6.5. There appears to be an anomolous peak in the GST data at 1 standard deviation, and the consequence of this is that the confidence limits, which assume normal distribution, are overly conservative. The significance of this result is uncertain. However it is readily apparent that the distribution is not skewed, as the data was in study 4.