Comparative Biochemistry and Physiology Part C 123 (1999) 153–163

Relationship between metallothioneins and metals in a naturalpopulation of the clam Ruditapes decussatus from Sfax coast: a

non-linear model using Box-Cox transformation

A. Hamza-Chaffai a,*, J.C. Amiard b, R.P. Cosson b

a Uni6ersite de Sfax, IPEIS, BP 805, Sfax 3018, Tunisiab EP61-CNRS-ISOMer, 1 rue Gaston Veil, 44035 Nantes, France

Received 15 May 1998; received in revised form 23 February 1999; accepted 3 March 1999

Abstract

Cadmium, copper and zinc were determined concomitantly with metallothionein-like proteins (MTLPs) in the subcellularfractions of Ruditapes decussatus digestive gland. This study covered 4 months and aimed to evaluate the effect of metal pollutionand other factors such as sex, size and reproductive state on MTLP levels. Copper concentrations did not vary with month,however Cd and Zn concentrations showed high levels during August. Organisms showing low cadmium concentrations presentedthe highest cadmium percentages in the soluble fraction (SF) containing MTLPs. However for high cadmium concentrations, theinsoluble fraction (IF) was implicated in cadmium association. MTLP levels varied according to the month, the sex and the sizeof the organisms. A non-linear model based on the Box-Cox transformation, was proposed to describe a positive and a significantrelationship between MTLPs and the studied metals. A model including sex and size showed that these two factors affected MTLPlevels, but were less important than metals. Males of R. decussatus showed higher significant correlations between MTLP levelsand cadmium than females. Moreover, the effect of size and reproductive state on MTLP levels was less perceptible in males thanin females. As a result, MTLPs in males of R. decussatus could be proposed as suitable biomarker for detecting metalcontamination. © 1999 Elsevier Science Inc. All rights reserved.

Keywords: Biomarker; Biomonitoring; Box-Cox transformation; Cadmium; Copper; Metallothionein-like proteins; Pollution;Ruditapes decussatus ; Zinc

1. Introduction

Biomonitoring based on biomarkers is a recent con-cept relying on the interaction between a toxic chemicaland a biological receptor in a living organism [23,26].This new approach allows the detection of the pollutanteffect at the cellular level. To be efficient, a biomarkershould have an early warning capacity, a certain specifi-city and other criteria as mentioned by some authors[15,27].

Among the biomarkers proposed in monitoring pro-

grams, there are metallothioneins (MTs), which are lowmolecular weight, heat stable proteins of extremely highsulphur and metal content [25]. These proteins arefound in various tissues of vertebrates and inverte-brates, and have been identified in most living organ-isms [2,14,18]. Marine invertebrates synthesis MTshaving biochemical characteristics and functional prop-erties similar to those of vertebrates [42].

Bivalves exposed to pollutants accumulate variouscontaminants by mechanisms related to their filter-feed-ing habits. They are able to accumulate toxic metals intheir organs to relatively high levels without any appar-ent detrimental effects. Many studies established a rela-tionship between MT synthesis and metal pollution in

* Corresponding author. Tel.: +216-4-614109; fax: +216-4-614109.

0742-8413/99/$ - see front matter © 1999 Elsevier Science Inc. All rights reserved.PII: S 0 7 4 2 -8413 (99 )00023 -7

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163154

bivalves [8,16,36,41]. Some species such as Mytilusgallopro6incialis and Mytilus edulis have been inten-sively used as bioindicators in the mussel watch [30].

Ecological and biological characteristics of bivalves,as well as the demonstrated relation between MT syn-thesis and metal contamination have permitted theproposition of MTs in these organisms as an efficientbiomarker [16]. Nevertheless, many considerationsshould be taken into account: (i) basal levels of MTsexist in organisms free of metal contamination; (ii)inherent and endogenous variations in MT levels aresusceptible to interfere with variations because ofmetal contamination and can therefore lead to erro-neous interpretations; (iii) in some bivalves, MTs arenot the only cellular compound binding metals, infact, metals can be found in lysosomes and/or in in-soluble precipitates [11,28,42]. Moreover, high molecu-lar weight proteins can be also involved in metalbinding.

Among the marine bivalves, Ruditapes decussatus iswidely distributed in the Mediterranean sea. Owing toits considerable economic importance, it is heavilycaught along the Tunisian coast, especially in thesouth eastern region (gulf of Gabes) where the ecolog-ical conditions (temperatures, nutrients, salinity,...) arefavourable for its growth and reproduction [39]. Inthis area, previous studies dealing with fish have evi-denced the industrial activity impact on metal andMT concentrations [19–21].

Most of published data on metals and MTs in R.decussatus resulted from laboratory studies and haveshown that toxic metals found in the tissues dependedon the surrounding water contamination and that thedigestive gland is an important organ for metal andMTLP determination [3,7,22,36]. Cadmium bioaccu-mulation kinetics indicated that R. decussatus couldbe classified as a high affinity species. Unfortunately,laboratory contamination could not reflect the realityof field conditions owing to the high and unrealisticconcentrations used as well as the difference in theway of metal assimilation [3]. Consequently, whenstudying metal and MTs relationship for a biomoni-toring purpose, field studies are needed to considernatural and inherent fluctuations.

The present work aimed to validate in field condi-tions the efficiency of MTs as biomarkers of metalcontamination in a natural population of the clam R.decussatus from the south eastern coast of Tunisia(Sfax). Since physico-chemical forms of stored metalsare very important to evaluate their toxicological sig-nificance, Cd, Cu and Zn were analysed in both solu-ble (SF) and insoluble fractions (IF) of the digestivegland; whereas MTLPs were analysed in the cytosolafter heat-denaturation. Sampling covered 4 months

(June to September) showing an important reproduc-tive activity [39].

Sex of the organism, as well as the size were con-sidered. Size of specimens were chosen within a smallrange in order to limit the variability of metal [3,1]and MTLP levels [29]. Despite the fact that sampledbivalves were calibrated, the residual influence of sizewas examined.

Our aim was to evaluate MTLP variability linkedto metal contamination and the other factors. Thelinear model MTLPs= f (Cd, Cu, Zn, Sex, size),based on ordinary least squares (OLS) method, aswell as a non-linear model using Box-Cox transforma-tion, were tested and proposed.

2. Material and methods

2.1. Sampling and fraction preparation

R. decussatus were collected from Gargour, a sitelocated 17 km to the south of Sfax. Sample collectionwas made from June to September, an important pe-riod in the reproductive cycle [39]. Samples consistedof specimens with an acceptable size for commerciali-sation (37–42 mm) and available all over the sam-pling period. Among 40 organisms, 10 males and 10females were selected. After sex determination, the di-gestive gland was dissected and stored at −80°C un-til preparation and analysis.

The digestive glands were homogenised with a tis-sue grinder in ice-cold 50 mM Tris solution and b−mercaptoethanol buffer pH 8.6. The ratio of the Trissolution to the fresh tissue weight was 4 ml/g. Thehomogenates were ultra-centrifuged (100 000×g, 60min, at 4°C). The supernatants SF were separatedfrom the pellet IF), heat denatured (80°C/15 min) toprecipitate the heat sensitive compounds and subse-quently centrifuged (30 000×g, 30 min, at 4°C). Theresulting supernatants (heat stable fraction HSF) wereused for MTLP quantification. During sample prepa-ration, the temperature was maintained at 4°C tominimise any risk of protein degradation.

2.2. Metal analysis

For metal analysis, samples were treated accordingto the method described in [20] and analysed byatomic absorption spectrophotometry (HITACHI Z8100) using the Zeeman effect [4]. Metal quantifica-tion was based on the method of standard additionsin an iso-medium. Our analyses were quality assuredduring a recent intercalibration exercise involving theFrench laboratory [13] (Table 1).

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163 155

2.3. Quantification of metallothionein-like proteins

MTLPs were analysed in the heat stable fraction (thesupernatant of the second centrifugation) using thedifferential pulse polarographic assay (D.P.P) for -SHcompounds, based on the Brdicka reaction [10] anddescribed by Thompson and Cosson [38]. Polaro-graphic measurements were made with a PAR Model174A analyser, a PAR/EG and G Model 303 staticmercury drop electrode (SMDE) and a flat-bed X-Yrecorder (RE 0089). During MTLP analysis the temper-ature was maintained at 4°C. MTLP amounts in R.decussatus were calculated using rabbit liver metalloth-ionein MT-I (Sigma Chemical, St Louis, MO) as refer-ence for the standard addition calibration curve.Olafson and Olsson [31] have confirmed the validityand the efficiency of this method. Results were ex-pressed as mg of MTLP per gram of homogenisedtissue.

2.4. Box-Cox transformation

The relationship between MTLP levels and the stud-ied factors (Cd, Cu, Zn, size and sex) can be expressedin some cases by a linear model (I), [20].

MTLP=b+a1Cd+a2Cu+a3Zn+a4Sex+a5Size(I)

In many cases this relationship is not linear. TheBox-Cox transformation can determine the non-linearfunctional relation between MTLPs and the otherparameters. The general functional form of the Box-Cox transformation [9] can be written:

Y(l)=a+ %i=n

i=1

biXi(l) (II)

a=0 when we take the standardised coefficients.

X(l)=Xl−1

l, (Z=X or Y)

In our case, model (II) can be written

MTLP(l)=+b1Cd(l)+b2Cu(l)+b3Zn(l)+b4Sex(l)

+b5Size(l)

MTLP= [b1(Cdl-1)+b2(Cul-1)+b3(Znl-1)

+b4(Sexl-1)+b5(Sizel-1)]1/l

where b1, b2, b3, b4 and b5(bi) are regression parame-ters, l is a scalar parameter defining a particular trans-formation. For l=1, the equation reduces to the linearform and as l�0, it becomes linear in the logarithmsof MTLP and Cd, Cu, Zn. We estimated a, bi, thestandard error of estimate s and l in two stages. In afirst stage, we obtained the least squares estimates of a,bi and s for fixed value of l. In a second stage wesearched for the value of l, say l0, which maximises thelog Likelihood function, say Lmax(l), and thus producesthe optimal transformation [9,40]. Data were standard-ised to avoid the problem of unit. Sex was introducedas a dummy variable. Lmax(l) values were determinedusing Shazam program (version 6.1).

3. Results

3.1. Variations of metal and MTLP le6els according tothe month

The results showing the variations of metal andMTLP according to month are expressed either inconcentrations (ng metal/g or mg MTLP/g) (Fig. 1) orin quantities (ng metal or mg MTLP) (Fig. 2). DuringAugust we observed significant variation of the diges-tive gland weight. Thus if the factor month is taken intoaccount, data treatments and interpretation were basedon metal and MTLP amounts rather thanconcentrations.

Monthly changes were observed for the studiedparameters and a certain similarity between metal andMTLP levels in both sexes was particularly evidencedwhen amounts were considered (Fig. 2). Statistics aboutdifferences between months based on metals andMTLP amounts are shown in Table 2. In which we cansee that month combinations including August showedsignificant differences for all the studied parameters.

3.2. Distribution of metals between soluble andinsoluble fractions

Metal percentages were used in order to show therelative importance of each metal in the subcellularfractions: the IF and the SF. In general, the percentages

Table 1A result of the intercalibration exercise, mean and (SD) 13, ourlaboratory code is 12

Cd Cu Zn(mg/g)

Fish homogenateMA-MEDPOL-1/TM

0.021Our values 0.66 17.3(0.0070) (0.10) (0.6)

Certified values 16.80.620.015(0.012) (0.12) (0.48)

Marine sedimentSE-MEDPOL-1/TMOur values 0.70 23.9 189

(0.06) (1.4) (2.3)Certified values 0.59 25.1 191

(17)(0.10) (3.8)

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163156

Fig. 1. Month (X axis)-related variations of metal concentrations in ng/g and MTLP concentrations in mg/g (Y axis). C1F: insoluble fraction offemales; C1M: insoluble fraction of males; S1F: soluble fraction of females; S1M: soluble fraction of males; M: MTLP in males, F: MTLP infemales

of metals within the fractions were not significantlyrelated to the sex (t-test at the level of 95%). Conse-quently, the results presented in this part will concernboth males and females (Table 3).

Copper was mainly associated to the SF with meanpercentages remaining remarkably stable all over theperiod of our study (Table 3). On the contrary, changeswere observed for soluble Cd and Zn, the highestpercentages of which were observed in June and coin-cided with the lowest amount of these metals in thedigestive gland (Table 3).

If we compare July and September, we can noticethat the total cadmium quantities were equivalent aswell as cadmium ratios in the studied fractions (Tables3 and 4). From June to July the increase in totalcadmium quantity (×1.8) led to an important increasein the IF (×3.2). Nevertheless, cadmium in the SFremains unchanged. From August to June, total cad-mium quantities became 10 times higher, a fact leadingto an important increase of Cd in the IF (×11.3)followed by the SF (×9.7) (Table 4). These resultsmean that the SF including MTLP is not the onlycompartment involved in cadmium association, the IF

links the excess of cadmium. Moreover, a moderatedincreasing in total cadmium quantities affects princi-pally the IF. On the other hand, when Cd increasebecame very important, it affects both IF and SF. Thelatter reflects the bioavailable cadmium and can lead totoxicological problems if MTLP capacity in binding Cdis overlapped.

In the case of zinc, a moderated increase in zincquantities from June to August (×3) affected princi-pally the IF.

3.3. Variations of metal and MTLP amounts accordingto the sex

As mentioned above, it is better to establish a com-parison between metal and MTLP amounts than be-tween their respective concentrations. MTLP and metalevolution according to the month were similar formales and females (Fig. 2). Nevertheless, in July MTLPamounts were significantly higher (99%) in females thanin males. Statistics about differences in MTLP, Cd, Cuand Zn quantities according to the sex are summarisedin Table 5. When considering metal quantities (but also

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163 157

concentrations; not shown) no significant differencewas established between males and females.

3.4. Simple correlations between MTLP and metals

For simple correlations, we drew a distinction be-tween males and females. When we plotted means ofmetal and MTLP amounts versus the months, we ob-served an ascending evolution for June, July and Au-gust (Fig. 2). A linear relationship between MTLP andmetal was established. The results are presented inTable 6. Generally, cadmium, copper and zinc werehighly correlated to MTLP for both sexes. The highestcorrelation coefficients were obtained for males. Forcadmium, the correlation coefficient (r=0.54) and theslope value (1.31) were higher for males, whereas forfemales r and slope values were low. In the case ofcopper and zinc, correlations were highly significantwith highest values of r in the case of males. Slope (a)values for males and females were of the same order(females/males: Cu 0.23/0.24; Zn: 0.06 /0.08).

3.5. The effect of size on MTLPs

Despite the fact that sampled bivalves were cali-brated (37–42mm), the residual influence of size wasexamined. Using data of June, July, August and Sep-tember, the effect of size on MTLP quantities wasstudied in males and females. For females a positiveand significant (at 99% level) relationship was estab-lished between MTLPs and size [model (III)].

Females MTLP= −4.3+0.17(t=3.86)

Size (III)

(n=39, r=0.53, P\99%)

However for males the relationship MTLPs/size wasnot significant [model (IV)] indicating that in malesMTLP variation can not be explained by size. Theseobservations indicate that MTLPs in males are lesssensitive to size variation than in females.

Males MTLP= −1.53+0.87(t= −1.77)

Size (IV)

(n=40, r=0.28, n.s)

Fig. 2. Month (X axis)-related variations of metal quantities (ng) and MTLP quantities in mg (Y axis). C1F: insoluble fraction of females; C1M:insoluble fraction of males; S1F: soluble fraction of females; S1M: soluble fraction of males; M: MTLP in males, F: MTLP in females.

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163158

Table 2MTLP, Cd, Cu and Zn: statistical significance of differences based on the montha

MTLP Cd Cd CuMTLP Cu Zn ZnM F M FF M F M

n.s n.s ** ** n.s n.s **June/July ****** *** *** ****** ***June/August *** ***

n.sJune/September n.s n.s ** *** ** n.s **n.sJuly/August ** *** ** *** *** *** ***

** n.s n.s **** *July/September n.s **** *** ** ** ** *** **August/September ***

a F: females, M: males, (***) significant at 99% level, (**) significant at 95% level, (*) significant at 90% level, (n.s) not significant.

3.6. Multiple correlation analysis and Box-Cox models

The main advantage of multiple regression analysis isto show the combined effect of each metal on MTLPs.Each positive coefficient (bi) indicated that, when oneof the parameters increases by 1 unit, MTLP increasesof bi unit whatever the sign of l. The linear modelcannot describe in all the cases the functional relation-ship between MTLP and the other independent vari-ables such as metals, size and sex. The Box-Coxtransformation gives us l value, the model coefficientsimproves t-test values and therefore increases the sig-nificance of each coefficient.

In the present paper, using data of the studiedmonths, MTLP amounts (mg) were correlated to metalquantities (mg), length and sex. The proposed modelwill concern only June, July and August because aphytoplanctonic bloom, the influence of which will bediscussed later, affected individuals caught during Sep-tember. We present here a comparison between thelinear and non-linear models. Firstly, we will presentresults of MTLPs= f (Cd, Cu, Zn) taking into accountsex of the animals (Table 7). Then we will give modelsincluding size and sex MTLPs= f (Cd, Cu, Zn, sex,size).

As shown in Table 7, the linear model based on themean square estimation of the coefficients presentedlow R2 and Student’s t-test values. The Box-Cox trans-formation improved the significance of each variableand gave higher R2. MTLPs were highly correlated tocadmium and zinc when we considered only males andboth males and females. For females, significant coeffi-cients were obtained only for Cu and Zn (Zn\Cu).

As it was shown in simple correlations, other measur-able factors such as sex and size can affect MTLPlevels. Thus, it is interesting to visualise together withmetals the effect of these factors. In this case, data werestandardised to avoid the problem of unit and sex wasintroduced as an independent dummy variable.

When we estimated the relationship between MTLPand Cd, Cu, Zn, sex and size using the linear model,only some coefficients were significant. However, usingBox-Cox transformation, the non-linear model resulted

in significant coefficients for most of the variables (seeresults in Table 8). Using the presented model, wenoticed that MTLPs were affected by Cd, Zn, size inthe following order Zn\Cd\size. The t value of sexcoefficient indicates that sex of individuals affects sig-nificantly MTLP levels. Comparing non-linear modelsof Tables 7 and 8, we observed that there is a positivecorrelation between MTLP and Cd and Zn consideringthe three metals (Table 7). This correlation remainedsignificant even if sex and size were introduced in themodel (Table 8). Nevertheless, compared to metals, sizeand sex affected less MTLP levels (lower coefficients)but should be taken into account when studyingMTLPs variations.

4. Discussion

Before considering MTLPs as a biomarker of metalcontamination, we should: (i) verify that the relation-ship between metals and MTLPs is relevant; (ii) studyindividual and combined effect of some factors (sex,size, month...) affecting MTLP levels; and (iii) check ifvariations linked to some endogenous factors arelargely lower than variations related to metal contami-nation as suggested by Cairns [12]. This work reflects,in natural conditions (in situ), the effect of metal con-tamination, sex, month and size on MTLPs in thedigestive gland of R. decussatus. Each factor was firstconsidered independently from the others. Then, thecombined effect of these factors was studied using bothlinear and non-linear models.

From a human health point of view, cadmium con-centrations in the whole soft tissue (in toto) were rela-tively high [0.30 mg/g wet weight (WW)] but remainedalways below the acceptable levels [24]. For the diges-tive gland our results were equivalent for Zn (31.3 mg/gWW) and generally higher for Cu (5.76 mg/g WW) andCd (0.40 mg/g WW) than the literature results thanthose obtained for R. decussatus from the French Med-iterranean coasts (Zn: 32.4, Cu: 2.22, Cd: 0.1 mg/g WW)[36]. For MTLP, mean concentration (3.40 mg/g) wasof the same order than those published by Bebianno et

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163 159

al. [7]. However, results for R. decussatus published byRomeo and Gnassia-Barelli [36] were about 24 timeslower than ours.

Our results as well as those published by Bebianno etal. [7] were obtained using the differential pulse polar-ography. A fact explaining similarities in results. Never-theless, Romeo and Gnassia-Barelli [36] used thecolorimetric method as described by Viarengo et al.[43]. The important difference observed in our case canbe explained, at least partly, by the different techniquesused in MTLP quantification. In fact, the comparisonof these two techniques has evidenced an important lossof MTLPs in the second one, owing principally to theuse of ethanol/chloroform precipitation [17,37,44].Moreover, the mentioned results concerned R. decussa-tus from an aquaculture farm [36] where water qualityis not the same as in field conditions (our case).

The factor sex did not affect metal concentrations inthe digestive gland of R. decussatus. Nevertheless, im-portant differences in MTLP levels linked to sex wereobserved during July (females\\males, significant atP\95% level) (Fig. 2). This indicates that factors otherthan metal contamination affected MTLP synthesis. Itis well known that the reproductive state of the organ-ism and the hormonal induction of MTLPs duringspawning period in females affect MTLP levels in vari-ous organisms [33,34]. Our explanation for the ob-served difference between males and females duringJuly, is the large gap in the condition index (CI). Infact, a study about the reproductive cycle of R. decussa-tus from our site (Gargour) has shown that during Julyfemales are in a gametic emission phase; they loseweight and have a low CI (CI=82). In contrast, malesare in a gonadic restoration phase and have a highCI=103 [39]. These observations explain the importantand highly significant difference in MTLP concentra-tions and amounts between males and females (Figs. 1and 2). For the other months, CI of males and femaleswere similar and the reproductive state was nearly thesame, leading to less variability in MTLP levels accord-ing to the sex.

The sampling month has to be taken into accountwhen studying MTLPs and metal relationship. MTLPfluctuations related to the month are more evident infemales. During the sampling period, the CI variationsin females (15%) were more important than in the caseof males (12%). Moreover, from June to Septembermales were in a gametic emission phase and the CIdecreased progressively from 116 to 85. Yet, during thesame period females underwent a gametic emissionfollowed by a gonadic restoration. A fact influencinghormonal stimulation of MTLP synthesis in females.Similar observations were made in the case of Salmogairdneri [32].

During the studied months, metal amounts in thedigestive gland varied from 0.09 to 0.9 mg for Cd, from2.1 to 5.3 mg for Cu and from 7 to 21 mg for Zn. Theabundance of Cu and Zn is as a result of the biologicalrole of these two essential metals.

Copper percentages exhibited the same pattern ofdistribution SF�IF whatever the month. In the case ofzinc, SF\IF only during June, for the other monthsthe SF was equivalent with the IF. In the SF, these twometals were mainly associated to the heat stable frac-tion containing MTLPs, this is as a result of the knownrole of MT in providing cations which are essential fornewly synthesised apoenzymes [35,42].

Cadmium distribution between the studied fractionswas not the same every month and changed with to thetotal Cd quantities. In June (0.09 mg of Cd), the SF wasthe major compartment for cadmium storage (71%).However, in July and September (0.16 and 0.15 mg ofCd, respectively) the IF increased and became equiva-lent to the SF. We demonstrated a relationship betweenthe increase of total cadmium amounts and its associa-tion with insoluble ligands (Table 4). These findings arein agreement with the implication of insoluble com-pounds in toxic metal neutralisation [42]. The MTLPcapacity to bind cadmium is limited and the excess ofcadmium is found in both insoluble and soluble heatdenatured fractions. During August Cd quantitiesshowed an important variation coefficient (71%) andincreased 10 times compared to June. In this case, Cd

Table 3Metal distribution in the studied fractions with respect to total amountsa

%Cu-SFCu(mg)Month (n) %Cd-SF (mg)Cd (mg) %Zn-SFZn(mg)

0.09 71 2.1June 78 7 66(0.03) (9) (0.7)(n=19) (7) (3) (8)0.16 55 2.9July 77 10 57

(18)(0.08) (8)(n=20) (4)(7)(0.4)5521735.3630.90August

(0.64) (18) (1.1)(n=20) (8) (4) (8)0.15 44September 3.7 76 13 47

(0.07) (9) (0.7) (4) (3) (8)(n=20)

a S.D. are between ( )

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163160

Table 4Cadmium and zinc ratios in the different fractions with reference to the month

Cd-IF Cd-SF Total ZnTotal Cd Zn-IF Zn-SF

1July: September 1.21.1 0.8 0.6 0.73.2 1.2 1.41.8 2July: June 1.1

August: June 11.310 9.7 3 4 2.2

Table 5MTLP, Cd, Cu and Zn amounts according to the sex, statisticalsignificancea

July AugustJune September

MTLP ***n.s n.s *n.s n.sn.s n.sCd

n.sCu n.s n.s n.sn.s n.s n.sZn n.s

a (*)significant at 85% level, (***) significant at 99% level, (n.s) notsignificant.

established between MTLPs/Cd and Zn for both malesand females. The highest correlation was obtained formales and described the relationship between MTLPs/Cd and Zn (R2=0.71, significant at 99%). The relation-ship MTLPs/Cd or Zn remained significant when sexand size were introduced in the model. In this case, justslight differences in t values were observed. MTLPlevels were affected by the studied variables accordingto this order Zn\Cd\size\sex. A very importantresult, demonstrating that the variation linked to sexand size was lower than variations linked to metals.Nevertheless, these factors should not be neglected in amonitoring program.

In this work, one of the studied months (September)was marked by an ecological phenomenon, an algalbloom. As a result, the normal physiological processesof metal regulation in organisms were disturbed. At themiddle of July, the phytoplankton was rich and multi-specific, a fact explaining the increase of the digestivegland weight during August. Gymnodinium sp, a toxicalga, appeared in the gulf of Gabes and reached oursampling station (September). Analysis performed atthe beginning of September showed 60×103 cell/ml [6].In vitro 2460 cell/ml of this species (Gymnodinium sp)induced 85% of embryological abnormalities in theoyster Crassostrea gigas during 2 days of exposure(Arzul, pers. comm.). This alga is supposed to cause ananoxia and then the bivalve closes its shell and stopseating. The result will be death or weight loss depend-ing on the species and the stress duration.

was mainly found in the soluble form (63%) and couldcause toxicological problems. In fact, in the SF cad-mium is bioavailable, and can either bind the heatstable compounds mainly MTs or link heat denaturedcompounds including enzymes, preferential targets fortoxic metals. Cd presence in the IF including metalgranules could limit toxicological problems.

For both males and females, MTLPs are significantlycorrelated to metals. The positive correlation betweenMTLPs and cadmium is more relevant in the case ofmales than in females. Compared to females, MTLPs inmales were less affected by the factors month and size.For multiple regression analysis, the linear model is notthe best way to explain the relation between MTLPsand metals. Using the Box-Cox transformation, weestablished significant correlations between MTLP/Cuand Zn (R2=0.48, at 95% level) for females. Correla-tions (R2=0.58 significant at 99% level) were also

Table 6Results of the linear relationship between MTLPs and metalsa

Slope r (t value)Intercept

0.54 (3.95)***MTLP= f(Cd) M (n=30) 1.31 1.36(2.46)**0.372.011.01F (n=29)

1.69 0.43M+F (n=59) 1.15 (4.2)***

0.30 0.57MTLP= f(Cu) M (n=30) (4.28)***0.240.48 (3.30)***0.890.23F (n=29)

0.24 0.60 0.50 (5.10)***M+F (n=59)

0.630.33 (5.55)***0.08MTLP= f(Zn) M (n=30)0.95 0.55F (n=29) 0.06 (4.01)***0.654 0.60M+F (n=59) 0.07 (6.61)***

a M: males, F: females, M+F: males and females. ( ): t value.*** Significant at P\99%.** Significant at P\95%.

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163 161

Table 7Functional relationship between MTLP and metals by using both linear and non-linear models, MTLPs= f(Cd, Cu, Zn)a

Linear model (OLS) Non-linear model (Box-Cox)R2Coefficient (ai)Variables Lambda (l)R2Coefficient (bi)

Cd n.sb 0.49Males (n=30)(2.58)***

−0.32Cu 0.71(n.s)0.53n.s0.48 0.37

Zn (1.71)** (2.03)***Females (n=29) Cd n.s (n.s)

Cu n.s 0.39 0.34 0.48 0.19(1.5)**

Zn 0.61 0.40(2.06)** (1.7)**

0.32Males +Females (n=59) n.sCd(2.28)***

Cu n.s 0.46 (n.s) 0.58 −0.120.61 0.48Zn

(3.37)***(3.5)***

a ( ) t ratios 56 df.** Significant at 95%.*** Significant at 99%.n.s, not significant.

an important reproductive activity, basal levels ofMTLP should be known in a reference site (unpol-luted). Any random ecological phenomena (i.e. blooms)should be noticed in order to avoid any erroneousinterpretations.

From a technical point of view, different methodshave been proposed to determine MTLP levels [14]. Abiomonitoring program needs a reliable, efficient andlow cost technique. The differential pulse polarographywas used for MTLP determination. It was establishedthat this technique constitutes a direct method for MTdetermination without need of chromatographic sepa-ration. In fact, centrifugation and heat treatment ofcytosolic extracts of tissues are sufficient to remove themajority of potentially interfering high molecularweight cysteine-containing proteins [7,31]. In thepresent work, the cytosol was separated by ultra-cen-trifugation, but centrifugation can be alternatively used,as it was demonstrated by Geret et al. [17]. Facilities insample analysis as well as the cited advantages permitto choose this technique in a monitoring program.However, owing to the different methods used forMTLP analysis [16,43] intersite comparisons are possi-ble only if the same method is used. In order togeneralise one technique for monitoring purposes, inter-calibration exercises for MTLP are needed as it was thecase for metal [5].

Acknowledgements

Two research programs FICU 95/PAS/14 andSERST/MENRS 223/96 supported this work. Dr Ami-

5. Conclusions

R. decussatus exhibits many features [30] enabling usto choose it as a bioindicator of metal contamination.We should avoid ‘in toto’ analysis and focus on oneparticular organ important in metal accumulation andgiving MTLP response such as the digestive gland.Calibrated organisms should be used in order to avoidas far as possible size effects. Moreover, it is better tochoose males because of the high correlations and thelow MTLP variability linked to the size and the spawn-ing period. For each month, especially those presenting

Table 8Functional relationship between MTLP and the studied variablesusing both linear and non-linear models, MTLPs= f(Cd, Cu, Zn,Sex, Size)a

Non-linear model (Box-Cox)Linear model (OLS) R2=0.53R2=0.65

Coefficients (bi) Lambda (l)Variables Coefficients (ai)

(n.s) 0.22Cd(1.75)**

(n.s) (n.s)Cu−0.15Zn

0.50 0.39(2.72)*** (2.82)***

0.130.23Size(3.20)***(2.84)***

−0.12Sex (n.s)(−1.70)**

a ( ) t ratios 54 df, (OLS): ordinary least squares.*** Significant at 99%.** Significant at 95%.

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163162

ard-Triquet C. (CNRS) and Arzul G. (Ifremer) aregratefully acknowledged for their interesting adviceduring paper preparation. Pr Chaffai M. is acknowl-edged for his collaboration during statistical treatments.The authors thank Souissi H. and Chermi H. for theirhelp during sample collection, Makhloufi M. for histechnical assistance and Kammoun K. for linguisticcomments.

References

[1] Amiard JC, Baud JP. Influence de l’utilisation des eaux saleessouterraines dans le nourrissage des coquillages: niveau d’ele-ments traces chez les bivalves adultes. In: Amiard-Triquet C,Hamon T, editors. Qualite de l’eau. France: Universite de Nan-tes, 1996:57–60.

[2] Amiard JC, Cosson RP. Les metallothioneines. In: Lagadic L,Caquet T, Amiard JC, Ramade F, editors. Biomarqueurs enEcotoxicologie Aspects Fondamentaux. Paris: Masson, 1997:53–66.

[3] Amiard JC, Metayer C, Baud JP, Ribeyre F. Influence de diversfacteurs ecologiques sur la bioaccumulation d’elements metal-liques (Cd, Cu, Pb, Zn) chez de jeunes palourdes (Ruditapesphilippinarum) au cours du pregrossissement en nourricerie. Re-vue des Sciences de l’eau 1991;4:441–52.

[4] Amiard JC, Pineau A, Boiteau H, Metayer C, Amiard-TriquetC. Application de la spectrometrie d’absorption atomique Zee-man aux dosages de huit elements traces (Ag, Cd, Cr, Cu, Mn,Ni, Pb et Se) dans des matrices biologiques solides. Water Res1987;21:693–7.

[5] Amiard-Triquet C, Hamza-Chaffai A. L’utilisation des proteinestype metallothioneine comme biomarqueurs d’exposition auxmetaux. In: Amiard-Triquet C, Hamon T, editors. Qualite del’eau. France: Universite de Nantes, 1996:7–12.

[6] Arzul G, Turki S, Hamza A, Daniel P, Merceron M. Fish killsinduced by phycotoxins. Toxicology 1995;339:1119.

[7] Bebianno MJ, Nott JA, Langston WJ. Cadmium metabolism inthe clam (Ruditapes decussata): the role of metallothioneins.Aquat Toxicol 1993;27:315–34.

[8] Bebianno MJ, Langston W. Cadmium induction of metalloth-ionein synthesis in (Mytillus gallopro6incialis). Comp BiochemPhysiol 1992;103C:79–85.

[9] Box GEP, Cox DR. An analysis of transformations. J RoyalStat Soc 1964;Ser B:211–252.

[10] Brdicka AR. Polarographic studies with the dropping mercurymethod: a new test for proteins in the presence of cobalt salts inammoniacal solution of ammonium chloride. Collect CzechChem Commun 1933;5:112–28.

[11] Brown BE. The form and function of metal-containing ‘granules’in invertebrate tissues. Biol Rev 1982;57:621–67.

[12] Cairns JJr. The threshold problem in ecotoxicology. Ecotoxicol-ogy 1992;1:3–16.

[13] Coquery M, Horvat M. The analytical performance study forMEDPOl area: determination of trace elements in marine sedi-ment SD-MEDPOL-1/TM and in fish homogenate MA-MED-POL-1/TM. Report IAEA, Monaco 1996:85

[14] Cosson RP, Amiard JC. Utilisation des metallothioneinescomme biomarqueurs d’exposition aux metaux. In: Lagadic L,Caquet T, Amiard JC, Ramade F, editors. Utilisation desbiomarqueurs pour la surveillance de la qualite de l’environ-nement. Paris: Lavoisier, 1998:77–109.

[15] Depledge MH, Amaral-Mendes JJ, Halbrook RS, Kloepper-Sams P, Moore MNN, Peakall DB. The conceptual basis of thebiomarker approach. In: Peakall DB, Shugart LR, editors.

NATO Advanced Science Institutes Series. Berlin: Springer Ver-lag, 1993:15–29.

[16] George SG, Olsson PE. Metallothionein as indicator of tracemetal pollution. In: Kramer KJM, editor. Biological Monitoringof Coastal waters and Estuaries. Boca Raton: CRC Press,1994:151–78.

[17] Geret F, Ringlet F, Cosson RP. Comparaison de deux proto-coles d’extraction pour la purification des metallothioneines de lamoule Mytilus edulis. J Rech Oceanographique 1997;22:151–6.

[18] Hamer DH. Metallothionein. Ann Rev Biochem 1986;55:913–51.

[19] Hamilton SJ, Mehrle PM. Metallothioneins in fish: review of itsimportance in assessing stress from metal contaminants. TransAm Fish Soc 1986;115:596–609.

[20] Hamza-Chaffai A, Amiard-Triquet C, El Abed A. Metalloth-ionein-like protein: is it an efficient biomarker of metal contami-nation? A case study based on fish from the Tunisian coast. ArchEnviron Contam Toxicol 1997;33:53–62.

[21] Hamza-Chaffai A, Cosson RP, Amiard-Triquet C, El Abed A.Physico-chemical forms of storage of metals (Cd, Cu and Zn)and metallothionein-like proteins in fish from the Tunisian coast:ecotoxicological consequences. Comp Biochem Physiol1995;111C:329–41.

[22] Hamza-Chaffai A, Romeo M, El Abed A. Heavy metals indifferent fishes from the middle eastern coast of Tunisia. BullEnviron Contam Toxicol 1996;56:766–73.

[23] Henry M, Huang W, Cornet C, Bellau M, Durbec JP. Contam-ination accidentelle par le cadmium d’un mollusque (Ruditapesdecussatus): bioaccumulation et toxicite (CL50, 96). Oceanolog-ica Acta 1984;73:329–35.

[24] Huggett RJ, Kimerle RA, Mehrle PM, Bergman HL. Biomark-ers-Biochemical, Physiological and Histological Markers of An-thropogenic Stress. Boca raton: Lewis Publishers, 1992:347.

[25] JORF. Arrete du 25 Juillet 1995 relatif au classement desalubrite et a la surveillance des zones de production et des zonesde reparcage des coquillages vivants. 1 Sept 1995.

[26] Kagi JHR, Kojima Y. Chemistry and biochemistry of metalloth-ionein. In: Kagi JHR, Kojima Y, editors. Metallothionein II.Basel: Birkhauser Verlag, 1987:25–61.

[27] Kramer KJM. Biomonitoring of coastal waters and estuaries.Boca Raton: CRC Press, 1994:327.

[28] Lagadic L, Caquet T, Amiard JC. Biomarqueurs en ecotoxicolo-gie: principes et definitions. In: Lagadic L, Caquet T, AmiardJC, Ramade F, editors. Biomarqueurs en ecotoxicologie. Paris:Masson, 1997:1–9.

[29] Mason AZ, Jenkins KD. Metal detoxification in aquatic organ-isms. In: Tessier A, Turner DR, editors. Metal speciation andbioavailability in aquatic systems. Chichester: John Wiley andSons, 1995:479–607.

[30] Mouneyrac C, Amiard JC, Amiard-Triquet C. Effect of naturalfactors (salinity and body weight) on cadmium, copper, zinc andmetallothionein-like protein levels in resident populations ofoysters (Crassostrea gigas) From a polluted estuary. Mar EcolProg Ser 1998;162:125–35.

[31] NAS. The International Mussel Watch. US Nat Acad Sci Wash-ington DC, 1980:248.

[32] Olafson RW, Olsson PE. Electrochemical detection of metalloth-ionein. Methods Enzymol 1991;205:205–13.

[33] Olsson PE, Haux C, Forlin L. Variation in hepatic metalloth-ionein, zinc and copper levels during an annual reproductivecycle in Rainbow trout (Salmo gairdneri ). Fish Physiol Biochem1987;31:39–47.

[34] Olsson PE, Zafarullah M, Gedman L. A role of metallothioneinin zinc regulation after oestradiol induction of vitellogenin rain-bow trout (Salmo gairdneri ). Biochem J 1989;257:555–60.

[35] Overnell J, Mc intosh R, Flechter TC. The levels of livermetallothionein and zinc in plaice (Pleuronectes platessa) during

A. Hamza-Chaffai et al. / Comparati6e Biochemistry and Physiology, Part C 123 (1999) 153–163 163

the breeding season, and the effect of oestradiol injection. J FishBiol 1987;30:539–46.

[36] Roesijadi G. Metallothionein induction as a measure of responseto metal exposure in aquatic animals. Environ Health Prospect1994;102(12):91–6.

[37] Romeo M, Gnassia-Barelli M. Metal distribution in differenttissues and in subcellular fractions of the Mediterranean clam(Ruditapes decussatus) treated with cadmium, copper or zinc.Comp Biochem Physiol 1995;111C(3):457–63.

[38] Romeo M, Cosson RP, Gnassia-Barelli M, Risso C, Stein X,Lafaurie M. Metallothionein determination in the liver of the seabass (Dicentrachtus labrax) treated with copper and B(a)P Mar.Environ Res 1997;443:275–84.

[39] Thompson JAJ, Cosson RP. An improved electrochemicalmethod for the quantification of metallothionein in marine or-ganisms. Mar Environ Res 1984;11:137–52.

[40] Trigui-Mnif N. La palourde (Ruditapes decussatus) des cotesTunisiennes: biometrie, reproduction et impact de l’environ-nement sur la bioaccumulation en metaux traces. These de 3emeC, Universite de Tunis II, Tunisia, 1995:269.

[41] UNEP. Statistical treatments and interpretation of marine com-munity data. IOC/UNEP: Athens, 1991:196.

[42] Viarengo A. Heavy metals in marine invertebrates: mechanismsof regulation and toxicity at the cellular level. CRC Crit RevAquatic Sci 1989;1:295–317.

[43] Viarengo A, Nott JA. Mechanisms of heavy metal cationhomeostasis in marine invertebrates. Comp Biochem Physiol1993;104C:355–72.

[44] Viarengo A, Ponzano E, Dondero F, Fabbri R. A simplespectrophotometric method for metallothionein evaluation inmarine organisms: an application to Mediterranean and Antarc-tic Molluscs. Mar Environ Res 1997;441:69–84.

.