BIOMARKER ASSAYS FOR THE DETECTION OF SUB-LETHAL TOXICITY IN THE AQUATIC ENVIRONMENT-A PRELIMINARY INVESTIGATION
Report No.: 952/1/04

EXECUTIVE SUMMARY

1. Background and motivation

Many countries, including South Africa, have adopted an integrated management approach, that includes chemical and biological methods, to address toxic and non-conventional pollutants in the aquatic environment (Slabbert et al., 1998a,b). Over the past 20 years, a range of biological tests measuring the effects of toxicants under controlled laboratory conditions have been established for South African use (Slabbert et al., 1998a). Most of the tests are rapid/short-term tests, aimed at the detection of acute toxicity. Very little attention has so far been given to the use of sub-lethal toxicity tests that can be used alongside acute tests to obtain information on long-term effects of low levels of chemicals on aquatic organisms. Sub-lethal toxicity tests are important management tools when the toxicity of effluents is low and are indispensable to manage ambient water quality.

Traditionally, full life-cycle tests or shorter tests of about 30 days, known as early life-stage tests, are used to establish sub-lethal toxicity in the aquatic environment. Because such tests are labour intensive and expensive, most countries have shortened the duration of the tests to 7 to 10 days, focusing on the most sensitive life-cycle stages. These tests are known as short-term chronic tests (Slabbert et al., 1998a; US EPA, 1991), and the primary endpoints are growth and reproduction. Examples of short-term chronic tests applied in the USA (Slabbert et al.,1998a) for fresh water testing include: a 7-day fish (Pimephales promelas) larval growth/survival test; a 7-day invertebrate (Ceriodaphnia dubia) reproduction/survival test; and a 96-h algal (Selenastrum capricornutum Printz) growth test.

In recent years, considerable attention has been given to the establishment of even more rapid measures of sub-lethal toxicity. A fascinating new approach is the use of biomarker assays that provides measures of the biochemical/molecular mechanisms underlying toxicity. These assays have been derived from mammalian medicine and allow for the rapid assessment of organism health (usually hours to a few days).

A biomarker may be biological material ranging from bio-molecules such as nucleic acids and proteins (e.g. enzymes) to organelles, cells, tissues, organs and whole organisms (Appendix A). Biomarkers indicating damage caused at the biochemical/molecular level in a cell have rapid response times and are highly sensitive. These biomarkers are currently the most applied in aquatic biomarker studies. Biomarkers of growth and reproductive responses are of high biological and ecological significance, but exhibit relatively low sensitivity to stress. Manifestations of stress at the tissue level represent an intermediate effect between the biochemical and reproductive levels. Histological changes occur earlier than reproductive changes and have been shown to be more sensitive than growth or reproductive measures, and as an integrative parameter, provides a better evaluation of organism health than a single biochemical marker.

Biomarkers can be used as indicators of exposure and effect, and can be categorised as non-specific and specific (Mayer et al., 1992). Some of the non-specific biomarkers, e.g. ribonucleic acid/deoxyribonucleic acid, radiolabelled amino acid or nucleotide incorporation, and adenylate energy charge, give direct information on the growth rate or potential of an organism. These biomarkers cannot be used to identify the specific toxicant causing the effect. There are two types of specific indicators of stress, namely organ- and toxicant specific. Organ-specific biomarkers include organ specific enzymes and iso-enzymes, and often depend on the detection of enzymes at increased concentrations in organs. These enzymes appear in blood when organs are damaged and indicate the presence and extent of damage. Examples include lactate dehydrogenase (LDH), transaminases, creatine phosphokinase, lysosomal enzymes, alkaline phosphatases and mixed function oxidase (Mayer et al., 1992; Gagné and Blaise, 1993). Toxicant-specific biomarkers indicate the exposure and effects due to a single chemical or group of chemicals, e.g. inhibition of acetylcholinesterase (AChE) by organophosphates and inhibition of cytochrome P monooxygenase, methallothionein and metal binding proteins by metals.

Since biomarkers measure events along the entire metabolic pathway during exposure to a chemical, and since the toxic effect on organisms is secondary to change inside cells, biomarkers hold promise for higher sensitivity and earlier detection of toxicants than the traditional assays measuring acute toxicity. Biomarkers can act as early warning signals of imminent, irreversible, permanent damage to an organism (Depledge, 1993).

2. Objectives

This project was carried out to establish biomarker assays, as rapid alternative measures to the more tedious growth and reproduction measures used in chronic toxicity tests, to detect sub-lethal toxicity in the South African aquatic environment.

The objectives of the project were:

1. To establish biological techniques to detect sub-lethal toxicity in the aquatic environment, and
2. To produce an operational manual on the established methodologies.

The focus of the project was on using biomarkers as sub-lethal endpoints in a laboratory fish test employing tilapia (Oreochromis mossambicus), to complement the acute (lethality) tilapia test developed for local use (Slabbert et al., 1999). Since juveniles are much more sensitive than adult organisms, the aim was to use the youngest possible juvenile fish in the laboratory test. Once the biomarker assays were established for laboratory use, it was planned to apply the assays to tilapia and other organisms collected from selected field sites to evaluate their applicability for field use. The study was not aimed at extrapolating the laboratory results to the field data, but rather to establish two options (laboratory and field) for biomarker application.

3. Methodology

Assays for the following biomarkers were established and optimised: protein; AChE; ethoxyresorufin-O-deethylase (EROD); glucose; glycogen; delta-aminovulinic acid dehydratase (ALA-D); LDH; glucose-6-phosphate-dehydrogenase (G-6-P-DH); pyruvate kinase (PK); heat shock protein (Hsp 70) and osmotic ion analyses. Published protocols were followed. Where possible, techniques were miniaturised (microplate assays) or test kits were used.

The biomarker assays were applied in laboratory and field studies. The indigenous fish species O. mossambicus was selected as test organism. A preliminary evaluation using unexposed (control) fish indicated that 1 to 1.5 cm (2 to 4 week old) fish were the most suitable for use in laboratory tests with reference to detectability of biomarker activity, smaller variation, ease of use, small sample volumes and expected higher sensitivity. Laboratory fish were maintained and bred according to well-established protocols. Laboratory studies were carried out following static acute fish test protocols. Biomarker activity was evaluated in the laboratory by exposing 30 fish per container (single exposures) or 60 fish divided over 3 containers (triplicate exposures) to various concentrations of selected chemicals [cadmium, zinc, pentachlorophenol (PCP) and cyanide] using exposure periods of 8, 24 and 96 h. Selected biomarker assays were carried out on the supernatant of the homogenised fish. Assays were carried out in triplicate, which allowed the calculation of means and coefficients of variation (CVs). All the biomarker results, except Hsp 70, were expressed in terms of fish weight. Hsp 70 was expressed in terms of protein. The percentage fish lethality and biomarker induction/inhibition was calculated by comparing test and control results. The Rust de Winter Dam was selected as reference site, and the Loskop and Hartbeespoort Dams as test sites for field studies. Two field surveys were carried out, one in summer and the other in winter. Where possible, 20 fish (including males and females) were caught. Biomarker assays (single) were carried out on the blood (whole blood, plasma and erythrocytes), brain and liver of the fish. Biomarker results were expressed in terms of protein.

An analysis of variance (Student's t-test) was applied to laboratory biomarker data to establish whether or not results differed significantly at the P<0.05 level. This allowed the derivation of the toxicity endpoint referred to as a LOEC (lowest of consecutive test concentrations at which biomarker activity was significantly different from that of the control). Where possible, regression analysis (linear regression) was performed on exposure data (concentration-response curves) to establish EC20s (the concentration causing a 20% induction or inhibition in biomarker activity). A two-way analysis of variance (ANOVA) and Dunnett's test were applied to field data to determine significant differences between localities and seasons. The significance level was set at P<0.05. Student' t-test was applied (p<0.05) to compare fish weights and lengths.

4. Summary of major results

4.1 Laboratory studies

The control results showed that, although values were low, all the biomarkers were present at measurable levels in the fish homogenate. The results obtained during single exposures were as follows: protein: 0.0166 to 0.0613 mg/mg fish; AChE (microplate assay): 0.0003 to 0.0041 abs/min/mg fish; AChE (test kit) (0.0003 to 0.0140 U/l/mg fish; glucose: 0.0017 to 0.0044 mg/mg fish; EROD: 0.0699 to 0.3605 nM/min/mg fish; PK: 0.0201 to 0.6691 mU/ml/mg fish: LDH: 0.0011 to 0.0120 U/l/mg fish; G-6-P-DH: 0.0562 to 4.6738 mU/l/mg fish: Hsp 70 (WITS): 7.80 to 122.01 mm2/20 µg protein; and Hsp 70 (RAU): 2 137 to 10 252 relative intensity/15 µg protein. The control values obtained during triplicate exposures were within the above ranges, except in the case of Hsp 70 (RAU), where values were considerably higher (6 238 to 15 685 relative intensity/15 µg protein.

A considerable fluctuation was noticed in the control biomarker values (mean of triplicate assays) of different single exposures. This was possibly because offspring of different weights and sizes (18 to 28 days old) were used in experiments while a fixed volume of buffer was used for homogenization. Glucose, protein, Hsp70 and EROD results exhibited the smallest fluctuation, and PK, AChE (test kit) and G-6-P-DH the largest fluctuation. When triplicate exposures were carried out, the fluctuation between the results of triplicate assays were much smaller, probably because the age of the fish used in these studies ranged between 21 and 24 days, reducing differences in weight and size.

Large inter-assay variations (triplicate assays) were noticed in the control results of single exposures in the case of: G-6-P-DH (14 to 72%), protein (0 to 92%), PK (5 to 98%), AChE (test kit) (2 to 109%) and LDH (2 to 110%). The best repeatability was obtained with the Hsp 70 (RAU) (CV: 5 to 28%) and glucose (CV: 0 to 32%) assays. With the exception of EROD, with CVs ranging from 8 to 102%, the inter-assay variation of triplicate exposures was much smaller (CVs: <50%) than the inter-assay variation of single exposures. The inter-experimental variation of triplicate exposures (triplicate data sets) of EROD and PK was >50% in some instances. With a few exceptions (mostly test kit data), most of the control CVs were in agreement with published values for fish homogenates.

In general, control lethalities were below 10% (detection limit for acute toxicity). A large lethality occurred in the control fish of the 96-h zinc exposure study (53%), which could have influenced biomarker activity. Lethality also occurred at some of the upper test concentrations during cadmium and zinc exposure studies. Oxygen concentrations in test and control containers during cyanide exposure studies were below the limit required for the sustenance of aquatic life (2.4 to 2.7 mg/l), which could have influenced biomarker results.

The exposure studies showed that biomarkers were affected by the test chemicals. Biomarkers were induced or inhibited. Although large effects were observed in some instances, effects were not always significant because of the large variation between replicate results. The results of replicate experiments (PCP and cyanide) were often inconsistent. Test concentrations showing significant biomarker induction in one experiment did not affect the biomarkers in the other experiment or vice versa. This could be attributed to variability in the sensitivity of test fish or to large variations between replicate results, rendering test results not significantly different from control results.

Protein was significantly affected by PCP (inhibition) and cyanide (induction) during 24 h exposure, and by cadmium (inhibition) during 8 h exposure. AChE was significantly affected (induction or inhibition) by all the chemicals during the shorter exposure periods (8 and 24 h), while only PCP affected (induction) AChE during 96 h exposure. Glucose was significantly affected (induction and inhibition) by cadmium during 8 h exposure, by cyanide (induction) during 24 h exposure, and by zinc (inhibition) and PCP (induction) during 96 h exposure. EROD was significantly affected by cadmium (induction) during 8 and 96 h exposure, by zinc (induction and inhibition) and cyanide (induction) during 24 h exposure and by PCP (induction) during 96 h exposure. LDH was significantly inhibited by zinc during 96 h exposure and significantly induced by PCP during 24 and 96 h exposure. Hsp 70 displayed significant responses to cadmium (induction and inhibition) during 96 h exposure, to zinc (inhibition) during 96 h exposure, and to cyanide (induction) during 24 h exposure. PK was significantly induced by cyanide during 24 h exposure, while G-6-P-DH did not exhibit significant responses. In general, the effects exhibited during single and triplicate cadmium exposures were similar.

A comparison with data in literature, showed that some results were in agreement with published data, while other results differed from published data.

Results showed that concentrations of cadmium and zinc that exhibited significant biomarker responses during 8 and 24 h exposure, caused lethality during the 96 h exposure. The biomarkers thus acted as early warning signals for acute toxicity. The significant responses detected during 96 h exposure, generally occurred at lethal concentrations. PCP caused significant biomarker responses at sub-lethal concentrations during 24 and 96 h exposure, resulting in a 10 to 60 times enhancement of sensitivity upon longer exposure in some instances. As in the case of metals and PCP, cyanide exhibited significant biomarker responses at sub-lethal concentrations during 24 h exposure.

In general, no significant concentration-response functions were observed after 24 and 96 h exposure. Curves either showed an upward-downward or a downward-upward trend at increasing concentrations, or reached plateaus parallel to the x-axis. Significant linear fits were, however, obtained for all the biomarkers during 8 h exposure to cadmium. AChE, glucose, EROD and Hsp 70 showed elevated activity at higher concentrations, while protein levels decreased.

The following toxicity endpoints were derived for the different chemicals:

The 8-h LOECs of protein, AChE, EROD and Hsp 70 were between two to four times higher than the calculated EC20s, while the LOEC of glucose was approximately four times lower than the EC20. The endpoints different because a fixed variation of 20% was used for the EC20 while the actual variations of test and control data used for the analysis of variation (LOECs) were higher or lower. In some instances shorter exposure periods resulted in lower LOECs while in other instances LOECs were lower during longer exposure.

4.2 Field studies

In general, the lengths and weights of fish collected from the different dams differed significantly. Fish collected from the Hartbeespoort Dam in summer were small (mean weight: 0.36 kg; mean length: 24.9 cm) compared to the fish collected in winter, and from the other dams (mean weight: 0.90 to 1.71; mean length: 35.8 to 41.4 cm). Most of the fish collected in summer were female, while numbers were more equally distributed during winter sampling.

Protein levels were the highest in erythrocytes, followed in order of magnitude by whole blood, blood plasma and liver. The erythrocyte protein in replicate fish showed the least variation (CVs: 15 to 30%). Large variations occurred between the results obtained for whole blood, blood plasma and liver protein (CV: whole blood - 18 to 52%; erythrocytes - 5 to 85%; liver - 25 to 52%). Protein in fish homogenates were between 10 to 50 times lower than the protein levels in whole blood, between 5 to 25 times lower than the protein levels in blood plasma and between 3 to 20 times lower than the protein in liver.

Large variations occurred between some of the replicate biomarker results of fish collected at the reference (control) site (Rust de Winter Dam). The CVs for the different biomarkers were as follows: ALA-D: 47 to 52%; AChE (erythrocytes): 54 to 121%; AChE (brain): 22 to 51%; glucose: 34 to 54%; LDH (plasma): 97 to 119%; LDH (liver): 95 to 98%; calcium: 14 to 27%; potassium: 12 to 25%; sodium: 4 to 5%; EROD: 33 to 54%; G-6-P-DH: 170 to 219%; glycogen: 32 to 112% and PK: 138 to 164%. A considerable number of fish collected from the dams showed an absence of LDH, G-6-P-DH and PK activity.

All the biomarkers, except the AChE in erythrocytes, showed positive results on one or more of the sampling occasions. The effects of test sites on G-6-P-DH could not be determined because of the large variability in results. Induction as well as inhibition occurred. Results were generally in agreement with those found in literature. Since chemical analyses were not carried out, effects could not be attributed to specific toxicants. Seasonal changes (temperature, pH, oxygen concentration) and other stressors, e.g. availability of food and handling, could also have affected biomarker activity.

5. Conclusions

• In broad terms, the objectives of the project, namely 1) to establish biological techniques to detect sub-lethal toxicity in the aquatic environment, and 2) to produce an operational manual on the established methodologies, were met. However, certain issues, discussed below, were not adequately addressed.
• Some of the biomarker assays were successfully applied to the homogenates of 2 to 4 week old tilapia (O. mossambicus) used in laboratory toxicity tests, as well as to the organs of tilapia collected from the field.
• Although whole fish homogenates contained low levels of the studied biomarkers as a result of the dilution of enzyme levels in organs, the levels were sufficiently high to detect induction and inhibition in exposed laboratory fish.
• The biomarker assay protocols were relatively easily established in the participating laboratories, because well-established, published methodologies were used. The miniaturisation of some of the methods improved assays in terms of rapidity (small sample volumes and large numbers of samples), simplicity and cost. However, some of the tests, especially the test kits, appeared to be incompatible with whole fish homogenates, showing large variation in some instances.
• The inter-assay and inter-experimental variation in laboratory and field studies were very high for some of the biomarkers. Operational manipulation (sample preparation and assaying) and test method limitation (tissue compatibility) appeared to be the most likely causes of the variation. Sample storage played a minor role.
• Apart from operational manipulation and test method limitation, factors such as handling stress and the development stage of the fish could have contributed to the variation in biomarker values of field studies. It is therefore important to minimise stress and select fish of similar size/age when conducting field surveys.
• Although there was some degree of inconsistency in the results of laboratory exposure studies, sufficient evidence was obtained that biomarkers could be applied as sub-lethal measures of toxic activity in laboratory toxicity tests.
• The limited number of offspring obtained for tilapia maintained in the laboratory and the inconsistency in reproduction was a major constraint. As a result
• elaborate experimental designs were not possible;
• experiments were not repeated a sufficient number of times;
• experiments were not carried out in replicate;
• wide concentration ranges were not examined; and
• a range of chemicals were not tested.
Consequently, the sub-lethal test using biomarkers as endpoints was not sufficiently validated to enable informed decisions on its potential as a routine laboratory test.
• · At this stage the sub-lethal test protocols, outlined in the Operational Manual (Venter et al., 2003) is only applicable for qualitative studies. Protocols were optimised, but extensive testing/evaluation is required before the methods can be standardised for specific applications.
• Although the biomarkers were not applied to other test organisms in the field (e.g. worms, snails) as was initially envisaged, the established biomarker assays should be applicable to other organisms, including other fish, with little or no method modification.
• The concept of using a multi-organisational team seemed like a good idea to build capacity and transfer knowledge. However, this resulted in large delays in the execution of experiments. The use of students that came and went, instead of well-trained technicians, resulted in large inter-assay and inter-experimental variation which could not be remedied.

6. Recommendations

• The use of replicate sets of fish and replicate assays in laboratory exposure studies is recommended, to reduce experimental and assay variability. Replicate assays should also be carried out on fish of field studies to improve variability.
• It is also recommended to use well-trained operators to conduct biomarker assays and to follow quality assurance protocols and practices to ensure low variability.
• Because of the large differences in the control variation of the different biomarkers studied, and because no clear concentration-response functions were identified for most of the laboratory results, the use of a fixed endpoint such as an EC10 or an EC20 is not recommended. The LOEC, derived by means of an analysis of variance, proved to be a more appropriate sub-lethal toxicity endpoint.
• Based on the data obtained for different exposure periods, it is recommended that the sub-lethal fish toxicity test is carried out over 96 h. Any effects on biomarkers due to handling and stress manifested during the shorter exposure periods should have subsided after 96 h.
• The results of 8 and 24 h tests indicated that certain biomarkers can serve as early warning signals for acute toxicity. It is recommended that further studies are carried out (organism/cellular/sub-cellular level) to exploit the potential of biomarkers as rapid screening tests.
• Based on a combination of features such as sensitivity, selectivity, simplicity, repeatability and rapidity, the following biomarker assays are recommended for the detection of sub-lethal responses in the 96 h laboratory toxicity test: Protein (micro Bradford method), AChE (microplate assay), glucose, EROD, PK, LDH and Hsp 70. ALA-D (whole blood), AChE (microplate assay) (erythrocytes and brain), glucose (plasma), glycogen (liver), LDH (plasma and liver), EROD (liver), osmotic ions (plasma) and Hsp 70 are recommended for field use. The Hsp 70 assay is the most complex and time consuming of the assays. All the assays require relatively expensive equipment (e.g. spectrophotometers). The biomarkers should be used in battery form to allow for optimal detection of harmful pollutants. The test protocols are outlined in the Operational Manual (Venter et al., 2003).
• The AChE microplate assay is preferred to the AChE test kit because of higher reproducibility, less fluctuation and more reliable data calculation. Likewise, the micro Bradford method (micro method) is preferred to the macro method because of rapidity and less variability.
• The G-6-P-DH assay is not recommended for use because it requires large sample volumes and produces unreliable and highly variable results. PK is not recommended for field use because of high variability and because its potential as an in situ marker is questioned by other researchers.
• It is recommended that fish weight is used to express biomarker activity in the laboratory toxicity test, instead of protein, because of smaller variation and fluctuation in results, and because protein acted as biomarker.
• A specific volume of buffer was added to a certain number of fish for homogenation. Because weight increases with age, this approach resulted in considerable fluctuation in laboratory control (unexposed fish) values. It is, therefore, recommended to base the volume of buffer on weight rather than on the number of fish.
• The established biomarkers should be used to complement fish lethality data and not to substitute chronic measurements such as growth and reproduction. Likewise, the biomarker approach is not a replacement for conventional field assessments, but a supplementary tool with great ecological relevance.
• Limited fish numbers is a major concern regarding the success of this fish as test organism in acute and sub-lethal toxicity tests. National breeding facilities should be established to ensure a consistent and healthy supply of the fish.

In addition, other fish species, with the potential to regularly provide large numbers of offspring, should be investigated as test organisms.

• Tilapia were present in large numbers in the dams studied. This might not be the case in other environments. Before studying biomarkers in organisms in the field, their abundance should be established to ensure sufficient replicates.
• Since biomarkers have been proven to be useful measures of sub-lethal stresses, it is recommended that additional biomarkers, including metallothioneins, are investigated for application in laboratory and field studies.
• The present laboratory study was only conducted on individual chemicals. It is recommended that the study be extended to environmental samples (drinking water, effluents, sediments, etc) to evaluate the applicability of the biomarkers with reference to complex chemical mixtures.
• As the biomarker techniques are becoming more established for aquatic studies, more countries are applying such measures in continuous monitoring programmes. This approach should also be valuable for our National Toxicants Monitoring Programme that is currently being developed. Molluscs have been found excellent test organisms in such studies. It is, therefore, recommended that funds are invested in the establishment of an in situ biomarker test using a local mollusc.