- Email: [email protected]
A multivariate study of protective effects of Zn and Cu against nephrotoxicity induced by cadmium metallothionein in rats
TOXICOLOGY
AND
APPLIED
PHARMACOLOGY
114,239-245
(1992)
A Multivariate Study of Protective Effects of Zn and Cu against Nephrotoxicity Induced by Cadmium Metallothionein in Rats X. Y. LIU,* T. Y. JIN,*’
G. F. NORDBERG,*
S. R.&NNAR,~ M. SJOSTROM,~
AND
Y. ZHOU~
*Departmeni of Environmental Medicine and TResearch Group for Chemometrics, Department of Organic Chemistry. Lime6 University, S-901 87 UmeL Sweden Received July 16, 199 1; accepted January 3 1, 1992
A Multivariate Study of Protective Effects of Zn and Cu against Nephrotoxicity Induced by Cadmium Metallothionein in Rats. LIU, X. Y., JIN, T. Y., NORDBERG, G. F., RANNAR,S., SJOSTROM, M., ANDZHOU,Y. (1992). Tuxicol. Appf. Pharmacof. 114,239-245.
Factorial experimental designwasusedto study the protective effects of Zn and Cu on cadmium-metallothionein(CdMT)-inducednephrotoxicity in male Wistar rats. In the factorial design two levelsof Zn (0 and 25 mg/kg body weight), two levelsof Cu (0 and 12.5 mg/kg), and two levels of CdMT (0.1 and 0.4 mg of Cd/kg) were usedas varied factors. The factorial designwas complementedwith a center point with all three variables at an intermediate setting, i.e., Zn at 12.5 mg/kg, Cu at 6.25 mg/kg, and CdMT at 0.25 mg Cd/kg. Each of the nine combinations of settingswasadministeredto one of nine groups with six rats in each. Zn and Cu were injected SC24 hr prior to the injection of CdMT. The concentrations of protein and Ca in urine and Ca in renal cortex wereusedaseffects. The relationship between the experimental designsettings and the effects were modeled with multiple regression.The multiple regressionanalysisrevealed that for the high doseof CdMT (i) the enhancedvaluesof protein in urine causedby CdMT injection could be more efficiently reducedby Zn than by Cu, and (ii) excessiveCa in urine and renal cortex could be more efficiently reduced by Cu than by Zn. No significant synergismor antagonism between Cu and Zn was
found. These models can be used to estimate the dose levels of Zn and Cu which will reduce the toxic effects of CdMT. The treatment of 20.4 mg/kg Zn, for example, will reducethe effects of 0.4 mg Cd/kg as CdMT on protein in urine, and 2.8 mg/kg Cu will reducethe Ca in urine to the levels of those causedby 0.25 mg Cd/kg (no Zn and Cu). Similarly, the effect of 0.4 mg Cd/kg on Ca level in renal cortex can be reducedto that of 0.28 mg Cd/kg as CdMT by 7.98 mg Cu/kg, which is three times as efficient as Zn. The obtained resultsmight be of importance in understandingthe mechanismof cadmium toxicity and the potential risk to the health of the population exposedto cadmium occupationally or environmentally. Q 1992 Academic press, k
The interaction of cadmium with essential nutrients, e.g., zinc, copper. manganese, iron, and calcium etc., has I To whom correspondence should be addressed.
been reviewed (Sandstead, 1977; Nordberg et al., 1978) and it has been recognized that several trace elements might act together to protect against cadmium toxicity. In particular, the zinc status of the body is considered to be related to the development of cadmium toxicity and the induction of metallothionein (MT) synthesis (Nordberg et al., 1978; Sato and Nagai, 1989). In a similar way, copper can induce synthesis of MT in rat liver (Bremner and Davies, 1976; Bremner and Young, 1976) and provide tolerance to the toxicity of cadmium within a given range of doses. The animal models of acute nephrotoxic effects of cadmium metallothionein (CdMT) (Nordberg et al., 1975a) enable us to investigate the possible mechanisms of cadmium nephrotoxicity after chronic cadmium exposure. Since kidney damage in rats, indicated by proteinuria, calciuria, and increased calcium in kidney cortex caused by CdMT has been previously reported (Fowler et al., 1987; Jin et al., 1987) these indices of cadmium nephrotoxicity can be considered as established. There are some studies dealing with the effects of zinc or copper on cadmium toxicity (Nordberg et al., 1978; Fowler et al., 1987) but there appears to be little data on the quantitative description of the protective effects of zinc and/or copper against the toxicity of CdMT. The present study aims at an investigation of the effects of zinc, copper. and CdMT as well as the effects of interaction between these metals. This study is based on a factorial design at two levels. Multiple regression analysis and three-dimensional effect surface were used to evaluate the effects of the individual metals and their possible interaction effects. The advantage of a factorial experimental design over a design where one variable at a time is changed is that the interaction effect between the metals can be estimated. A series of experiments designed according to a factorial design will also give a higher degree of precision of the estimated effects with the same number of animals. Factorial design is described in numerous textbooks (Box et al., 1978; Box and Draper, 1987) and is implemented in many statistical packages.
239 Copyright 0 1992 All rights of reproduction
0041-008x/92 $5.00 by Academic Press. Inc. in any form reserved.
LIU ET AL.
240 MATERIALS
AND METHODS
Animals. Male Wistar rats, 100-120 g, were obtained from ALAB, (Stockholm, Sweden) and food (Ewos-ALAB rat/mouse) and tap water were provided ad libitum. After conditioning for 1 week in a temperature-light controlled room, they were divided into 10 groups (six rats in each). The rats in each group were administered different dose levels of Zn. Cu, and/ or CdMT with 0.9% NaCl as vehicle by subcutaneous injection. This route of exposure provides a suitable supply of CdMT to the systemic circulation, thus allowing renal accumulation of Cd and the development of renal tubular dysfunction in a controlled way (Nordberg et al.. 1975a). Subcutaneous injection of Zn (Chmielnicka et al., 1983) or Cu (Durnam and Palmiter. 198 I) is an established method ofadministration for MT induction. In order to observe the effects of pretreatments which induce MT synthesis, Zn as ZnClz and Cu as CuS04 were administered 24 hr before the CdMT challenge. The rats were housed individually in metabolic cages immediately after CdMT injection. Twenty-four-hour urine samples were collected and stored at -70°C. The animals were euthanized after the urine samples had been collected. Kidneys were removed and stored at -70°C until they were analyzed. During the study one animal was lost in Group 1. Experimental design. A full factorial design at two levels (Box e( al.. 1978) complemented with a center point was used (Table 1). This is the experimental domain where effects occur which differ from background (control) observations. The control group injected with normal saline (No. IO in Table I ) was thus not included in the factorial design. The nine groups of rats received different doses of Zn, Cu, and/or CdMT (independent variables). The doses of Zn and Cu were chosen in order to obtain a high level of MT induction based on a small pilot study. A molar ratio of 2: I for Zn: Cu was found to be suitable based on the relative subcutaneous toxicity of these metals. The method for the CdMT preparation was described in 1975 (Nordberg et al.. 1,975b). Concentrations of protein and Ca in urine and Ca in renal cortex were used as effect variables. Chemical analysis. The biuret method was used for assay of urinary protein (Piscator, 1962). Urinary creatinine (crea) determination was performed as described by Hare ( 1950). MT was measured by ELISA (Garvey, 1984). Urine samples for analysis of metals were digested with an equal volume of concentrated nitric acid (65%). Renal cortex tissue (50-100 mg) was dried at 105°C for 16 hr and ashed at 450°C for 24 hr. Ashed samples were dissolved in 1 M nitric acid. Ca. Zn, Cu, and Cd concentrations in urine and renal cortex were measured by atomic absorption in the flame mode (Varian-AAS 875). Bovine liver from the National Bureau of Standards
(SRM 1577a) was analyzed. Ca, Zn. Cu. and Cd were 96.6, 102, 93, and 100%. respectively, of those values specified for the SRM 1577a. Dafa analysis. The measurements for the individual animals were used in the multiple regression analysis. The data analysis and the effect domain modeling were performed by the statistical package MODDE (Umetri AB, UmeP). Due to the skewed distribution of the effect variable measurements, recommended In transformation was performed (Box et al., 1978). This transformation will make the distributions of the effect measurements resemble the normal distribution more closely. In the analysis of the factorial design, the relationship between the measurements (r) of an effect and the corresponding variable settings (X) is modeled as:
Here the b. is the model intercept (i.e., when all settings Xz,, Xc,, and Xc,,, equal 0 in coded form) and bZn. b,,, and bcdMT are regression coefficients which express the main effectsof Cu. Zn, and CdMT. The coefficients bzncu. bZnCdMT.and hCuCdMTexpress the conjugate interaction between Cu, Zn. and is the coefficient for the three-factor interaction. By CdMT. The bZnCuCdMT means of a I test and the standard error of a coefficient, the p value for the regression coefficient can be estimated. The p value measures the probability of observing the value of the coefficient or a more extreme value given that the null hypothesis (that the coefficient is zero) is true. In the caseof significant interaction terms, these coefficients should always be interpreted with the corresponding main effects. Just as those of individual effects can be estimated, the p value for the sum of each individual coefficient can be estimated. The regression coefficients are estimated with linear multiple regression. In the model the levels of the independent variable settings Xz., Xc”, and .X&,,, are in their coded form. i.e., x4 = (A - M)/f(H
- L).
where X, is a dose in code, .4 is the actual administered dose mg/kg of a metal, 111is the mean value of the high and low level of a metal. and Hand L are the high and low doses of an administered metal. For example, Xz, = (Zn - 1X)/12.5. where 12.5 is the mean value ofZn doses administered. Similarly in the equations. Xc, = (Cu - 6.25)/6.25 and X,,,, = (CdMT 0.25)/O. 15. For a more detailed description of the analysis of factorial designs. see Box et ul. (1978) and Box and Draper ( 1987).
TABLE 1 Factorial Design (FD) and Corresponding Variable Settings Variable settings
Coded FD settings Grow
CdMT
1 2
-1
3 4 5 6 7 8 9
-1
10"
1
Zn
cu
CdMT mgikg
Zn w/kg
Cu m&kg
-I
-1 -I -I -1
0.1 0.4 0.1 0.4
0 0 25 25 0 0 25 25 12.25
0 0 0 0 12.5 12.5 12.5 12.5 6.25
-I 1 1
1 -1 1
-1 -1
-1
I 1 0
1 0
-1.7
-I
n The control group is not included in the factorial design.
1 1 1 1 0 -1
0.1 0.4
0.1 0.4 0.25 0
0
0
PROTECTION
AGAINST
Cd NEPHROTOXICITY
241
BY Zn OR Cu
TABLE 2 Measured Mean Values for Protein (PrUR) and Calcium (CaUR) in Urine and Calcium in Renal Cortex (CaK) Group
ln(PrUR)”
2 3 4 5b 5’ 6 7h I’ 8 9
1.22 2.39 1.21 1.67 3.21 1.05 1.93 2.88 1.39 1.46 1.35
10d
1.24
SD
In(CaUR)”
SD
ln(CaK)”
SD
0.21 0.11 0.37 0.30
0.24 0.61 0.32 0.50
0.35 0.40 0.31
2.34 4.69 2.34 3.16 3.34 2.46 3.22 2.96 2.27 2.96 2.52
0.08 0.18 0.25 0.22
0.35 0.38 0.17
2.89 3.29 2.79 2.80 3.51 2.59 2.44 4.07 2.58 2.19 2.40
0.12
2.25
0.59
2.34
0.16
0.60 0.59
0.44 0.53
0.26 0.29 0.16 0.55 0.31
’ Natural logarithmic transformation of PrUR in mg/mg creatinine, CaUR in rg/mg creatinine, and CaK in fig/g wet wt. b Mean value for the two deviating animals. ’ Group mean with the two animals deleted. d The control group is not included in the factorial design and the model development.
RESULTS The mean values of protein and Ca in 24-hr urine samples and Ca concentration in renal cortex after CdMT injection were determined (Table 2). The values of analysis in urine samples were calculated in relation to the creatinine excretion and thus a correction for changes in urine volume was achieved. In this section and in the following discussion, we use the term “effect” to describe the results observed instead of the term “response” used in the methods of experimental designs (Box and Draper, 1987). This terminology is the one which is widely used in metal toxicology (Pfitzer and Vouk, 1986). The measurements for the three effects (proteinuria, calciuria, and Ca in the kidney cortex) of CdMT injection for the individual animals were then subjected to multiple regression modeling and three models were calculated. Inspection of the model residuals for each of the effects with established residual diagnostic tests as normal probability plots (Box et al., 1978; Box and Draper, 1987) revealed en-
hanced values in all three effects for four animals, one of whom had visible blood in the urine. These deviating animals were from Groups 5 and 7, with two animals in each group. The variable settings were at a high dose (12.5 mg/kg) for Cu and at a low dose (0.1 mg/kg) for Cd in Groups 5 and 7 and with Zn administration in Group 5. The multiple regression analysis for the three effects was repeated with these four animals deleted. The coefficients in the multiple regression models with and without these four animals were calculated (Table 3). However, as indicated in the following discussion, only the Models 2b, 3b, and 4b in the absence of the four deviating animals will be considered. The effects of the pretreatments with Zn and/or Cu on the toxicity of CdMT were examined (Figs. 1,2a and 2b, 3a and 3b). In these projections the X-axis and Y-axis are used for independent variables, and the Z-axis for an effect. Some of the cadmium and MT data will be briefly mentioned at the end of this section in order to aid the understanding of this paper.
TABLE 3 Regression Coefficients” for the Three Effects
PrUR PrUR CaUR CaUR CaK CaK
Eq.
bo
2a 2b 3a 3b 4a 4b
1.654* 1.517* 2.754* 2.659* 2.999* 2.94*
b Z”
-0.133 -0.106*** -0.085 -0.107 -0.19* -0.182*
b C” 0.07 1 -0.085 -0.144*** -0.252* -0.197 -0.268**
b CdMT
bzncu
b Z”CdMT
0.17*** 0.324* -0.119 -0.012 0.598* 0.669*
0.045 0.074 0.063 0.042 0.067 0.076
-0.162*** -0.189* -0.1 -0.078 -0.105*** -0.114**
’ The p values for the regression coefficients are *p < 0.0 I. **O.Ol < p < 0.05, and ***0.05 < p < 0. I ’ Standard error of the regression coefficients.
bC”CdMT -0.238* -0.082 -0.223* -0.115*** -0.371* -0.3*
b Z”C”CdMT
Sh
0.017 -0.012 -0.002 0.019 0.102*** 0.093***
0.08 0.057 0.078 0.068 0.057 0.05 I
LIU ET AL.
FIG. 1. Effect surface for protein in urine with injected Zn and CdMT without pretreatment of copper.
Proteinuria. Among the 10 groups, Group 2 (0.4 mg Cd/ kg) demonstrated the highest mean level of urinary protein excretion, while rats in Group 1 (0.1 mg Cd/kg) were not significantly different from those in the control group (Table 2). According to the multiple regression model 2b (Table 3) urinary protein (PrUR) can be modeled as
ln(PrUR)
= 1.517 - 0.106Xz, - O.OSSXc,
+ 0.324XcdMT + 0.074&&, - 0.082&,&,,,
- 0.189Xz,Xc,,, - 0.0 12X,&-,&~,,
(2b).
In this model and the following ones, the injected doses of Zn, Cu, and CdMT are in their coded form with the corresponding effect surface presented (Fig. 1). In Eq. 2b, the influences of CdMT (p < 0.01) and the combination of Zn
and CdMT (p < 0.05) are significant (Table 3). The regression coefficient for the interaction term between Cu and CdMT or Zn and CdMT in Model 2b shows that the effect of CdMT on protein in urine can be reduced more effectively by Zn than by Cu. The model predicts that PrUR = 5.9 mg/mg crea if 0.25 mg Cd/kg as CdMT and no Zn or Cu is administered (i.e., Xc,,, = 0, Xc, and X,, = -1). If a single dose of 20.4 mg/kg of Zn is administered (i.e., X,, = 0.68 and Xc, = -1) the PrUR value in rats treated with 0.4 mg Cd/ kg as CdMT will be at the same level as that for rats treated with only 0.25 mg Cd/kg of CdMT. Calciuria. Rats administered with 0.4 mg Cd/kg as CdMT (Group 2) demonstrated the highest mean level of Ca in urine among the 10 groups (Table 2). Also, rats in Group I, with 0.1 mg Cd/kg, displayed a value of Ca concentration in urine significantly higher than that of the controls (Table 2). The Regression Model 3b for Ca level in urine (CaUR) is ln(CaUR) - 0.0 12&MT
= 2.659 - 0.107Xz, - 0.252X,, + O.O42Xz,Xc,
- 0.115&“&,,
- 0.078X&&,,
+ 0.0 19X,,&“&,,,
(3b).
In this model the p values for the coefficients of Xc,, Xc,,, , and X,, are co.00 1, 0.1, and 0.13, respectively. Thus the model predicts that at a high dose of CdMT (X,,,, = 1) the effect of CdMT can be reduced by injection of Zn or Cu or both. According to the regression coefficients (6,, = -0.107, bZnCdMT= -0.078, bcU= -0.252 and bCuCdMT = -0.115) the administration of Cu is more efficient than that of Zn at intermediate or higher levels of CdMT (see Fig. 2a and 2b). If no Zn is administered, 2.8 mg/kg Cu could reduce the effect of 0.4 mg Cd/kg as CdMT on CaUR to the same level as for 0.25 mg Cd/kg.
FIG. 2. Effect surfaces for Ca in urine (a) with injected Cu and CdMT without pretreatment of Zn and (b) with injected Zn and CdMT without pretreatment of Cu.
PROTECTION
AGAINST
Cd NEPHROTOXICITY
If a single high dose of 0.4 mg Cd/kg as CdMT is given (Xc, = - I), the p value for the sum of all coefficients monitoring the protective effect of Zn on calciuria in Model 3b would be calculated to be 0.05. Calcium in the renal cortex. In order to assessthe damage of kidney, the Ca concentration in renal cortex (CaK) was analyzed. A similar result in Group 2 to those for the other two indices mentioned above was observed; this group reached the highest level of Ca in renal cortex among the 10 groups (Table 2) while the rats in Group 1 (0.1 mg Cd/kg) displayed no difference from the control group (Group 10). Both Zn and Cu can protect the kidney damage caused by the injection of CdMT (Fig. 3a). In Model 4b the coefficients are all statistically significant (p < 0.05) except for the coefficient of the interaction term between Cu and Zn (bz,,-,) and the three-variable interaction (see Table 3): ln(CaK)
= 2.94 - 0.1 82Xz, - 0.268&,
+ 0.669&~~
+ O.O76Xz,X-, -
0.3’h&dMT
o.o93X&&u&,,MT
Cab).
The negative signs of the coefficients for the terms X,, , Xc,,, show that the effect of high doses and &&&dMT of CdMT can effectively be reduced by either Cu or Zn or both. The presence of interaction terms, i.e., Cu or Zn with CdMT, predicts that a certain dose of Cu and Zn is more effective at higher doses of CdMT than at lower ones. Similar to that of the calciuria, the Ca concentration in the renal cortex could be more efficiently decreased by Cu pretreatment than by Zn. Although slightly outside the experimental domain, the model suggests that the effect of 0.4 mg Cd/kg as CdMT on Ca concentration in the renal cortex can be reduced to that in the dose of 0.28 mg Cd/kg if 26 mg/kg Zn or 7.98 mg/kg Cu is administered 24 hr before the injection of CdMT. &n&dMTT
243
Cadmium concentration in the renal cortex. Concentrations of 13.25 + 1.62, 26.19 -t 2.61, and 19.10 + 5.25 pg/g wet weight (means fr SD) were found in Groups 1, 2, and 6, respectively. MT concentration in renal cortex. Concentrations of 0.535 + 0.248, 2.129 + 0.910, and 4.143 + 1.363 mg/g wet weight (means + SD) were found in groups 1, 2, and 6, respectively. During the modeling no quadratic terms, i.e., X; are included. Such terms may exist for Ca in kidney and are indicated by comparing the model intercept (bO) with the experimental values with all the variable settings at their intermediate values. Since it is not possible from the present design to resolve the individual quadratic terms of Cu, Zn, and CdMT, the quadratic terms are not included in the model. To resolve this term an at least three-level design is necessary (in this case with 15 groups). DISCUSSION
- O.l14Xz,Xc~~~ +
BY Zn OR Cu
Models for the protective effects of Zn and Cu against the renal toxic effects of CdMT have been determined and illustrated using multiple regression and effect surface plotting. A full factorial design with a center point was used for the experimental settings for the nine groups of rats studied. With this experimental design we have quantitatively estimated the effects of CdMT, Zn, and Cu on kidneys of rats as well as the interaction between al1 three variables. We found that the pretreatments with a certain dose level of Zn and/or Cu can protect rat kidney against the nephrotoxicity of CdMT. The important effects of Cu or Zn pretreatments on calciuria induced by CdMT, some of which have not previously been reported, were identified. The Eqs. 2b, 3b, and 4b can be used to estimate the values of the toxic effects of CdMT at different dose levels of Zn, Cu, and Cd when they are expressed in coded form (see under Materials and Methods) provided that the levels of
FIG. 3. Effect surfaces for Ca in renal cortex with injected Cu and Zn and (a) with 0.25 mg Cd/kg and (b) with 0.4 mg Cd/kg administered asCdMT.
244
LIU
ET
AL.
doses are within or quite close to the experimental domain. the events underlying the appearance of calciuria are different Furthermore, a model can be used to find an equation ex- from those for proteinuria. These results are also of interest pressing the level of Cu or Zn which can reduce the toxic in relation to our previous findings that the calciuria appeared effect of interest to a certain value after a specific Cd dose. earlier than proteinuria (Jin et al., 1987). Possibly, the deThis can be done by introducing a preset coded dose level creased calciuria was fully picked up during the 24-hr urine of Cd and a certain value (In transformed) of a toxic effect. collection period of the present experiment, while that might Then the model will be transformed to an expression con- not be the case for proteinuria. taining the coded Cu and Zn levels as unknowns. Thus, by By multiple regression analysis of results obtained in this setting a level for one of the two metals, the level for the study, it has been revealed that the Ca concentrations in other will be estimated, as exemplified under Results. urine and renal cortex could be more efficiently reduced by It has been recognized that a single CdMT injection in Cu than by Zn pretreatment (see Eqs. 2b and 3b). This finding rabbits and rats can cause acute kidney damage, indicated may be related to the interactive effect between Cu and Cd by abnormal urinary indices and calcification of kidney cells because of their different affinity to ligands in the plasma or (Fowler and Nordberg, 1978; Fowler et al., 1987). As com- kidney. The following two explanations might be possible: pared with Group 10 (control), the minimum dose of injected (i) Less CdMT was delivered to the kidney. It is known CdMT which gives an increase in urinary protein in rats is that Cu binds more firmly to MT than to Cd (Bremner, estimated to be 0.19 mg Cd/kg. This value is obtained by 1987); the Cd in the CdMT which is administered 24 hr after inserting Xz, = - 1, Xc, = - 1 and can be considered as the Cu injection might be partly replaced by Cu available from critical dose for proteinuria induced by CdMT injection. other proteins present in plasma. Thus, the concentration Similarly in Eq. 4b, the minimum dose of injected CdMT of Cd bound to MT in plasma might be decreased and thus in rats which gives an increase of Ca concentration in the less CdMT is delivered to the kidney. This was supported renal cortex is estimated to be 0.16 mg Cd/kg. by the fact that in this study a lower kidney concentration The mean value of Cd in the renal cortex at 0.1 mg Cd/ of Cd was observed in rats with Cu pretreatment (see under kg dose of CdMT was 13.25 & 1.62 kg/g wet weight. It sup- Results, Groups 2 and 6). ports the conclusion suggested in several previous investi(ii) MT synthesis in the kidney was promoted. In rats gations (e.g., Nordberg and Nordberg, 1987) that the critical exposed to Cd, kidney MT was reported to be rich in Cu concentration of Cd in renal cortex when administered as (Webb, 1982). Thus, when the kidney contains a certain CdMT is within lo-20 pg/g wet weight. amount of Cu because of the Cu administration, the MT The dose-dependent relationship observed in this study synthesis in kidney cells is induced and protects against the was similar to that observed in a study of CdMT nephroCd nephrotoxicity. A high concentration of MT in the kidtoxicity in rabbits (Fowler and Nordberg, 1978). By a single neys of Cu-pretreated rats was also observed in this study intravenous injection of 0.1 mg Cd/kg as CdMT, these an- (see under Results, Groups 2 and 6). One or both of the imals disclosed a normal cellular architecture in kidney by possible explanations, mentioned above, are likely to account light microscopic evaluation, while those rabbits adminisfor the observation that the Ca concentrations in urine and tered with 0.4 mg Cd/kg showed extensive tubular necrosis. renal cortex were most efficiently reduced by the Cu preIn the present study, it has been shown that both proteintreatment. uria and calciuria caused by CdMT injection could be effecAs shown in Figs. 3a and 3b, when the rats were treated tively reduced by pretreatment with Zn (see under Results, with a higher dose of CdMT (0.25 or 0.4 mg Cd/kg) and a Figs. 1 and 2b). However, the magnitudes of the decreased relatively low dose of Zn or Cu (or both), the Ca level in the protein and Ca concentrations in urine were different. For renal cortex was reduced more effectively. With increasing example, if a single dose of 20 mg Zn/kg is administered to doses of Cu and/or Zn, the additional positive (protective) the rats, the protein concentration in urine after CdMT in- effect will be less pronounced. This may be reasonable when jection of 0.4 mg Cd/kg would be as low as 6.0 mg crea and one considers the competitive binding between Cd and Cu the Ca concentration in urine would be 17.6 pg/mg crea. to the MT in kidney. Since Cu is bound more firmly to MT The former is close to the value when only a single 0.25 mg than Cd, the capacity of MT binding to Cd might be deCd/kg as CdMT is administered and the latter is close to the creased when the Cu concentration in kidney is high. It could value when 0.1 mg Cd/kg as CdMT is administered. One lead to an increased proportion of non-MT-bound Cd in previous study also reported similar findings (Fowler et al., kidney, which has been considered to be responsible for the 1987); when rats are treated with 20 mg Zn/kg prior to the nephrotoxicity of Cd (Sato and Nagai, 1980). The binding injection of 0.4 mg Cd/kg CdMT, the concentration of Ca of Cd to MT might also release Cu ions, which in turn act in urine is the same as that in controls while the protein on the kidney. Several investigations have suggested the imconcentration in urine remains much higher than that of portance of Zn and Cu as modifying factors in the develcontrols. This phenomenon, both from our study and from opment of Cd and mercury nephrotoxicity (Petering et al., that conducted by Fowler et al. (1987) might indicate that 1984; Bogden et al., 1980).
PROTECTION
AGAINST
Cd NEPHROTOXICITY
It is also of interest to discuss the four rats with deviating effect variables reported under Results. The rats were from Groups 5 and 7, i.e., groups with high Cu and low CdMT treatments. They exhibited serious kidney dysfunction, as indicated by their values in all three measured effects (Table 2) when compared to those of the other rats in the same group. It is not known with certainty that Cu caused these effects but a possibility is that a subpopulation of the Wistar rats for which a relatively low dose of Cu might possibly be nephrotoxic exists. In conclusion, the results of this study extend previous reports on the CdMT-injection model as a tool for evaluating Cd-induced injury to renal proximal tubule cells. By the experimental design used in this study, we have obtained empirical models which relate the administered dose levels of Zn, Cu, and CdMT to the three effects monitoring nephrotoxicity. Data from this study have also shown the doseeffect relationships of the interactive effects among CdMT, Zn, and Cu on the kidney. The results showed that CdMTinduced proteinuria and calciuria may be reduced or prevented by adequate Zn and Cu administration in rats and calciuria can be reduced more efficiently by a certain dose of Cu pretreatment than by Zn. In a future paper we will further discuss and analyze other biological effects from the same experimental design. ACKNOWLEDGMENTS X. Y. Liu and Y. Zhou were supported by WHO Grants during part of this study. A grant from the Swedish Environmental Protection Agency (802-832-91-Fx) is gratefully acknowledged. Thanks are also given to M. Nordberg for the gift of CdMT.
REFERENCES Bogden, J. D., Kemp, F. W., Troiano, R. A., Jortner, B. S., TimPone, C., and Giuliani, D. (1980). Effect of mercuric chloride and methylmercury chloride exposure on tissue concentrations of six essential minerals. Environ. Res. 21, 350-359. Box, G. E. P., Hunter, W. G., and Hunter, J. S. (I 978). Statistics for Experimenters. Wiley and Sons, New York. Box, G. E. P., and Draper, N. R. (1987). Empirical Model-Building and Response Surface. Wiley and Sons. New York. Bremner. I., and Davies, N. T. (1976). Studies on the appearance of a hepatic copper-binding protein in normal and zinc-deficient rats. Br. Nutr. 36, 101-l 12. Bremner, I., and Young, B. W. (1976). Isolation of (copper, zinc)-thioneins from the livers of copper-injected rats. Biochem. J. 157, 5 17-520. Bremner, I. (1987). Interactions between metallothionein and trace elements. Prog. Food Nutr. Sci. 11, l-37.
BY Zn OR Cu
245
Chmielnicka, J., Komsta Szumska. E., and Zareba, G. (1983). Effect of interaction between 65Zn, mercury and selenium in rats (retention, metallothionein, endogenous copper). Arch. Toxicol. 53, 165-175. Durnam, D. M., and Palmiter, R. D. (1981). Transcriptional regulation of the mouse metallothionein-I gene by heavy metals. J. Biol. Chem. 256, 5712-5716. Fowler, B. A., and Nordberg, G. F. (1978). The renal toxicity of cadmium metallothionein: Morphometric and X-ray microanalytical studies. Toxicol. Appl. Pharmacol. 46, 609-623. Fowler, B. A., Goering, P. L., and Squibb, K. S. (1987). Mechanism of cadmium-metallothionein-induced nephrotoxicity: Relationship to altered renal calcium metabolism. Experientia Suppl. 52, 66 l-668. Garvey, J. S. (1984). Metallothionein: Structure/antigenicity and detection/ quantitation in normal physiological fluids. Environ. Health Perspect. 54, 117-127. Hare, R. S. (1950). Endogenous creatinine in serum and urine. Proc. Sot. Exp. Biol. 74, 148- 15 1. Jin, T. Y., Leffler, P., and Nordberg, G. F. (1987). Cadmium-metallothionein nephrotoxicity in the rat: Transient calcuria and proteinuria. Toxicology 45,307-3 17. Nordberg, G. F., Goyer, R., and Nordberg, M. (1975a). Comparative toxicity of cadmium-metallothionein and cadmium chloride on mouse kidney. Arch. Pathol. 99, 192-197. Nordberg, M., Nordberg, G. F., and Piscator, M. (1975b). Isolation and characterization of a hepatic metallothionein from mice. Environ. Physiol. Biochem. 125, 396-403. Nordberg, G. F., Fowler, B. A., F&erg, L., Jemelov, A., Nelson, N., Piscator, M., Sandstead, H. H., Vostal, J., and Vouk, V. B. (1978). Factors influencing metabolism and toxicity of metals: A consensus report. Environ. Health Perspect. 25, 3-4 1. Nordberg, M., and Nordberg, G. F. (1987). On the role of metallothionein in cadmium induced renal toxicity. Experientia Suppl. 52, 669-675. Petering, D. H., Loftsgaarden, J., Schneider, J., and Fowler, B. A. (1984). Metabolism of cadmium. zinc and copper in the rat kidney: The role of metallothionein and other binding sites. Environ. Health Perspect. 54, 73-81. Pfitzer, E. A., and Vouk. V. (1986). Genera.1 considerations of dose-effect and dose-response relationships. In Handbook on the Toxicology of Metals (L. Friberg et al., Eds.). Vol. 1, pp. 149-l 74. Elsevier, B.V., Amsterdam. Piscator, M. (1962). Proteinuria in chronic cadmium poisoning. Arch. Environ. Health 5, 325-332. Sandstead, H. H. (1977). Nutrient interactions with toxic elements. In Toxicology of Trace elements (R. A. Goyer and M. A. Mehlman. Eds.), pp. 241-256. Hemisphere, Washington, DC. Sato, M., and Nagai, Y. (1980). Mode of existence of cadmium in rat liver and kidney after prolonged subcutaneous administration. Toxicol. Appl. Pharmacol. 54, 90-99. Sato, M., and Nagai, Y. (1989). Effect of zinc deficiency on the accumulation of metahothionein and cadmium in the rat liver and kidney. Arch. Environ. Contam. Toxicol. 18, 587-593. Webb, M. (1982). Role of metallothioneins and other binding proteins in the renal handling and toxicity of metals. In Nephrotoxicity: Assessment and Pathogenesis (P. H. Bach et al., Eds.), pp. 296-309. St. Edmundsbury Press. Great Britain.
AND
APPLIED
PHARMACOLOGY
114,239-245
(1992)
A Multivariate Study of Protective Effects of Zn and Cu against Nephrotoxicity Induced by Cadmium Metallothionein in Rats X. Y. LIU,* T. Y. JIN,*’
G. F. NORDBERG,*
S. R.&NNAR,~ M. SJOSTROM,~
AND
Y. ZHOU~
*Departmeni of Environmental Medicine and TResearch Group for Chemometrics, Department of Organic Chemistry. Lime6 University, S-901 87 UmeL Sweden Received July 16, 199 1; accepted January 3 1, 1992
A Multivariate Study of Protective Effects of Zn and Cu against Nephrotoxicity Induced by Cadmium Metallothionein in Rats. LIU, X. Y., JIN, T. Y., NORDBERG, G. F., RANNAR,S., SJOSTROM, M., ANDZHOU,Y. (1992). Tuxicol. Appf. Pharmacof. 114,239-245.
Factorial experimental designwasusedto study the protective effects of Zn and Cu on cadmium-metallothionein(CdMT)-inducednephrotoxicity in male Wistar rats. In the factorial design two levelsof Zn (0 and 25 mg/kg body weight), two levelsof Cu (0 and 12.5 mg/kg), and two levels of CdMT (0.1 and 0.4 mg of Cd/kg) were usedas varied factors. The factorial designwas complementedwith a center point with all three variables at an intermediate setting, i.e., Zn at 12.5 mg/kg, Cu at 6.25 mg/kg, and CdMT at 0.25 mg Cd/kg. Each of the nine combinations of settingswasadministeredto one of nine groups with six rats in each. Zn and Cu were injected SC24 hr prior to the injection of CdMT. The concentrations of protein and Ca in urine and Ca in renal cortex wereusedaseffects. The relationship between the experimental designsettings and the effects were modeled with multiple regression.The multiple regressionanalysisrevealed that for the high doseof CdMT (i) the enhancedvaluesof protein in urine causedby CdMT injection could be more efficiently reducedby Zn than by Cu, and (ii) excessiveCa in urine and renal cortex could be more efficiently reduced by Cu than by Zn. No significant synergismor antagonism between Cu and Zn was
found. These models can be used to estimate the dose levels of Zn and Cu which will reduce the toxic effects of CdMT. The treatment of 20.4 mg/kg Zn, for example, will reducethe effects of 0.4 mg Cd/kg as CdMT on protein in urine, and 2.8 mg/kg Cu will reducethe Ca in urine to the levels of those causedby 0.25 mg Cd/kg (no Zn and Cu). Similarly, the effect of 0.4 mg Cd/kg on Ca level in renal cortex can be reducedto that of 0.28 mg Cd/kg as CdMT by 7.98 mg Cu/kg, which is three times as efficient as Zn. The obtained resultsmight be of importance in understandingthe mechanismof cadmium toxicity and the potential risk to the health of the population exposedto cadmium occupationally or environmentally. Q 1992 Academic press, k
The interaction of cadmium with essential nutrients, e.g., zinc, copper. manganese, iron, and calcium etc., has I To whom correspondence should be addressed.
been reviewed (Sandstead, 1977; Nordberg et al., 1978) and it has been recognized that several trace elements might act together to protect against cadmium toxicity. In particular, the zinc status of the body is considered to be related to the development of cadmium toxicity and the induction of metallothionein (MT) synthesis (Nordberg et al., 1978; Sato and Nagai, 1989). In a similar way, copper can induce synthesis of MT in rat liver (Bremner and Davies, 1976; Bremner and Young, 1976) and provide tolerance to the toxicity of cadmium within a given range of doses. The animal models of acute nephrotoxic effects of cadmium metallothionein (CdMT) (Nordberg et al., 1975a) enable us to investigate the possible mechanisms of cadmium nephrotoxicity after chronic cadmium exposure. Since kidney damage in rats, indicated by proteinuria, calciuria, and increased calcium in kidney cortex caused by CdMT has been previously reported (Fowler et al., 1987; Jin et al., 1987) these indices of cadmium nephrotoxicity can be considered as established. There are some studies dealing with the effects of zinc or copper on cadmium toxicity (Nordberg et al., 1978; Fowler et al., 1987) but there appears to be little data on the quantitative description of the protective effects of zinc and/or copper against the toxicity of CdMT. The present study aims at an investigation of the effects of zinc, copper. and CdMT as well as the effects of interaction between these metals. This study is based on a factorial design at two levels. Multiple regression analysis and three-dimensional effect surface were used to evaluate the effects of the individual metals and their possible interaction effects. The advantage of a factorial experimental design over a design where one variable at a time is changed is that the interaction effect between the metals can be estimated. A series of experiments designed according to a factorial design will also give a higher degree of precision of the estimated effects with the same number of animals. Factorial design is described in numerous textbooks (Box et al., 1978; Box and Draper, 1987) and is implemented in many statistical packages.
239 Copyright 0 1992 All rights of reproduction
0041-008x/92 $5.00 by Academic Press. Inc. in any form reserved.
LIU ET AL.
240 MATERIALS
AND METHODS
Animals. Male Wistar rats, 100-120 g, were obtained from ALAB, (Stockholm, Sweden) and food (Ewos-ALAB rat/mouse) and tap water were provided ad libitum. After conditioning for 1 week in a temperature-light controlled room, they were divided into 10 groups (six rats in each). The rats in each group were administered different dose levels of Zn. Cu, and/ or CdMT with 0.9% NaCl as vehicle by subcutaneous injection. This route of exposure provides a suitable supply of CdMT to the systemic circulation, thus allowing renal accumulation of Cd and the development of renal tubular dysfunction in a controlled way (Nordberg et al.. 1975a). Subcutaneous injection of Zn (Chmielnicka et al., 1983) or Cu (Durnam and Palmiter. 198 I) is an established method ofadministration for MT induction. In order to observe the effects of pretreatments which induce MT synthesis, Zn as ZnClz and Cu as CuS04 were administered 24 hr before the CdMT challenge. The rats were housed individually in metabolic cages immediately after CdMT injection. Twenty-four-hour urine samples were collected and stored at -70°C. The animals were euthanized after the urine samples had been collected. Kidneys were removed and stored at -70°C until they were analyzed. During the study one animal was lost in Group 1. Experimental design. A full factorial design at two levels (Box e( al.. 1978) complemented with a center point was used (Table 1). This is the experimental domain where effects occur which differ from background (control) observations. The control group injected with normal saline (No. IO in Table I ) was thus not included in the factorial design. The nine groups of rats received different doses of Zn, Cu, and/or CdMT (independent variables). The doses of Zn and Cu were chosen in order to obtain a high level of MT induction based on a small pilot study. A molar ratio of 2: I for Zn: Cu was found to be suitable based on the relative subcutaneous toxicity of these metals. The method for the CdMT preparation was described in 1975 (Nordberg et al.. 1,975b). Concentrations of protein and Ca in urine and Ca in renal cortex were used as effect variables. Chemical analysis. The biuret method was used for assay of urinary protein (Piscator, 1962). Urinary creatinine (crea) determination was performed as described by Hare ( 1950). MT was measured by ELISA (Garvey, 1984). Urine samples for analysis of metals were digested with an equal volume of concentrated nitric acid (65%). Renal cortex tissue (50-100 mg) was dried at 105°C for 16 hr and ashed at 450°C for 24 hr. Ashed samples were dissolved in 1 M nitric acid. Ca. Zn, Cu, and Cd concentrations in urine and renal cortex were measured by atomic absorption in the flame mode (Varian-AAS 875). Bovine liver from the National Bureau of Standards
(SRM 1577a) was analyzed. Ca, Zn. Cu. and Cd were 96.6, 102, 93, and 100%. respectively, of those values specified for the SRM 1577a. Dafa analysis. The measurements for the individual animals were used in the multiple regression analysis. The data analysis and the effect domain modeling were performed by the statistical package MODDE (Umetri AB, UmeP). Due to the skewed distribution of the effect variable measurements, recommended In transformation was performed (Box et al., 1978). This transformation will make the distributions of the effect measurements resemble the normal distribution more closely. In the analysis of the factorial design, the relationship between the measurements (r) of an effect and the corresponding variable settings (X) is modeled as:
Here the b. is the model intercept (i.e., when all settings Xz,, Xc,, and Xc,,, equal 0 in coded form) and bZn. b,,, and bcdMT are regression coefficients which express the main effectsof Cu. Zn, and CdMT. The coefficients bzncu. bZnCdMT.and hCuCdMTexpress the conjugate interaction between Cu, Zn. and is the coefficient for the three-factor interaction. By CdMT. The bZnCuCdMT means of a I test and the standard error of a coefficient, the p value for the regression coefficient can be estimated. The p value measures the probability of observing the value of the coefficient or a more extreme value given that the null hypothesis (that the coefficient is zero) is true. In the caseof significant interaction terms, these coefficients should always be interpreted with the corresponding main effects. Just as those of individual effects can be estimated, the p value for the sum of each individual coefficient can be estimated. The regression coefficients are estimated with linear multiple regression. In the model the levels of the independent variable settings Xz., Xc”, and .X&,,, are in their coded form. i.e., x4 = (A - M)/f(H
- L).
where X, is a dose in code, .4 is the actual administered dose mg/kg of a metal, 111is the mean value of the high and low level of a metal. and Hand L are the high and low doses of an administered metal. For example, Xz, = (Zn - 1X)/12.5. where 12.5 is the mean value ofZn doses administered. Similarly in the equations. Xc, = (Cu - 6.25)/6.25 and X,,,, = (CdMT 0.25)/O. 15. For a more detailed description of the analysis of factorial designs. see Box et ul. (1978) and Box and Draper ( 1987).
TABLE 1 Factorial Design (FD) and Corresponding Variable Settings Variable settings
Coded FD settings Grow
CdMT
1 2
-1
3 4 5 6 7 8 9
-1
10"
1
Zn
cu
CdMT mgikg
Zn w/kg
Cu m&kg
-I
-1 -I -I -1
0.1 0.4 0.1 0.4
0 0 25 25 0 0 25 25 12.25
0 0 0 0 12.5 12.5 12.5 12.5 6.25
-I 1 1
1 -1 1
-1 -1
-1
I 1 0
1 0
-1.7
-I
n The control group is not included in the factorial design.
1 1 1 1 0 -1
0.1 0.4
0.1 0.4 0.25 0
0
0
PROTECTION
AGAINST
Cd NEPHROTOXICITY
241
BY Zn OR Cu
TABLE 2 Measured Mean Values for Protein (PrUR) and Calcium (CaUR) in Urine and Calcium in Renal Cortex (CaK) Group
ln(PrUR)”
2 3 4 5b 5’ 6 7h I’ 8 9
1.22 2.39 1.21 1.67 3.21 1.05 1.93 2.88 1.39 1.46 1.35
10d
1.24
SD
In(CaUR)”
SD
ln(CaK)”
SD
0.21 0.11 0.37 0.30
0.24 0.61 0.32 0.50
0.35 0.40 0.31
2.34 4.69 2.34 3.16 3.34 2.46 3.22 2.96 2.27 2.96 2.52
0.08 0.18 0.25 0.22
0.35 0.38 0.17
2.89 3.29 2.79 2.80 3.51 2.59 2.44 4.07 2.58 2.19 2.40
0.12
2.25
0.59
2.34
0.16
0.60 0.59
0.44 0.53
0.26 0.29 0.16 0.55 0.31
’ Natural logarithmic transformation of PrUR in mg/mg creatinine, CaUR in rg/mg creatinine, and CaK in fig/g wet wt. b Mean value for the two deviating animals. ’ Group mean with the two animals deleted. d The control group is not included in the factorial design and the model development.
RESULTS The mean values of protein and Ca in 24-hr urine samples and Ca concentration in renal cortex after CdMT injection were determined (Table 2). The values of analysis in urine samples were calculated in relation to the creatinine excretion and thus a correction for changes in urine volume was achieved. In this section and in the following discussion, we use the term “effect” to describe the results observed instead of the term “response” used in the methods of experimental designs (Box and Draper, 1987). This terminology is the one which is widely used in metal toxicology (Pfitzer and Vouk, 1986). The measurements for the three effects (proteinuria, calciuria, and Ca in the kidney cortex) of CdMT injection for the individual animals were then subjected to multiple regression modeling and three models were calculated. Inspection of the model residuals for each of the effects with established residual diagnostic tests as normal probability plots (Box et al., 1978; Box and Draper, 1987) revealed en-
hanced values in all three effects for four animals, one of whom had visible blood in the urine. These deviating animals were from Groups 5 and 7, with two animals in each group. The variable settings were at a high dose (12.5 mg/kg) for Cu and at a low dose (0.1 mg/kg) for Cd in Groups 5 and 7 and with Zn administration in Group 5. The multiple regression analysis for the three effects was repeated with these four animals deleted. The coefficients in the multiple regression models with and without these four animals were calculated (Table 3). However, as indicated in the following discussion, only the Models 2b, 3b, and 4b in the absence of the four deviating animals will be considered. The effects of the pretreatments with Zn and/or Cu on the toxicity of CdMT were examined (Figs. 1,2a and 2b, 3a and 3b). In these projections the X-axis and Y-axis are used for independent variables, and the Z-axis for an effect. Some of the cadmium and MT data will be briefly mentioned at the end of this section in order to aid the understanding of this paper.
TABLE 3 Regression Coefficients” for the Three Effects
PrUR PrUR CaUR CaUR CaK CaK
Eq.
bo
2a 2b 3a 3b 4a 4b
1.654* 1.517* 2.754* 2.659* 2.999* 2.94*
b Z”
-0.133 -0.106*** -0.085 -0.107 -0.19* -0.182*
b C” 0.07 1 -0.085 -0.144*** -0.252* -0.197 -0.268**
b CdMT
bzncu
b Z”CdMT
0.17*** 0.324* -0.119 -0.012 0.598* 0.669*
0.045 0.074 0.063 0.042 0.067 0.076
-0.162*** -0.189* -0.1 -0.078 -0.105*** -0.114**
’ The p values for the regression coefficients are *p < 0.0 I. **O.Ol < p < 0.05, and ***0.05 < p < 0. I ’ Standard error of the regression coefficients.
bC”CdMT -0.238* -0.082 -0.223* -0.115*** -0.371* -0.3*
b Z”C”CdMT
Sh
0.017 -0.012 -0.002 0.019 0.102*** 0.093***
0.08 0.057 0.078 0.068 0.057 0.05 I
LIU ET AL.
FIG. 1. Effect surface for protein in urine with injected Zn and CdMT without pretreatment of copper.
Proteinuria. Among the 10 groups, Group 2 (0.4 mg Cd/ kg) demonstrated the highest mean level of urinary protein excretion, while rats in Group 1 (0.1 mg Cd/kg) were not significantly different from those in the control group (Table 2). According to the multiple regression model 2b (Table 3) urinary protein (PrUR) can be modeled as
ln(PrUR)
= 1.517 - 0.106Xz, - O.OSSXc,
+ 0.324XcdMT + 0.074&&, - 0.082&,&,,,
- 0.189Xz,Xc,,, - 0.0 12X,&-,&~,,
(2b).
In this model and the following ones, the injected doses of Zn, Cu, and CdMT are in their coded form with the corresponding effect surface presented (Fig. 1). In Eq. 2b, the influences of CdMT (p < 0.01) and the combination of Zn
and CdMT (p < 0.05) are significant (Table 3). The regression coefficient for the interaction term between Cu and CdMT or Zn and CdMT in Model 2b shows that the effect of CdMT on protein in urine can be reduced more effectively by Zn than by Cu. The model predicts that PrUR = 5.9 mg/mg crea if 0.25 mg Cd/kg as CdMT and no Zn or Cu is administered (i.e., Xc,,, = 0, Xc, and X,, = -1). If a single dose of 20.4 mg/kg of Zn is administered (i.e., X,, = 0.68 and Xc, = -1) the PrUR value in rats treated with 0.4 mg Cd/ kg as CdMT will be at the same level as that for rats treated with only 0.25 mg Cd/kg of CdMT. Calciuria. Rats administered with 0.4 mg Cd/kg as CdMT (Group 2) demonstrated the highest mean level of Ca in urine among the 10 groups (Table 2). Also, rats in Group I, with 0.1 mg Cd/kg, displayed a value of Ca concentration in urine significantly higher than that of the controls (Table 2). The Regression Model 3b for Ca level in urine (CaUR) is ln(CaUR) - 0.0 12&MT
= 2.659 - 0.107Xz, - 0.252X,, + O.O42Xz,Xc,
- 0.115&“&,,
- 0.078X&&,,
+ 0.0 19X,,&“&,,,
(3b).
In this model the p values for the coefficients of Xc,, Xc,,, , and X,, are co.00 1, 0.1, and 0.13, respectively. Thus the model predicts that at a high dose of CdMT (X,,,, = 1) the effect of CdMT can be reduced by injection of Zn or Cu or both. According to the regression coefficients (6,, = -0.107, bZnCdMT= -0.078, bcU= -0.252 and bCuCdMT = -0.115) the administration of Cu is more efficient than that of Zn at intermediate or higher levels of CdMT (see Fig. 2a and 2b). If no Zn is administered, 2.8 mg/kg Cu could reduce the effect of 0.4 mg Cd/kg as CdMT on CaUR to the same level as for 0.25 mg Cd/kg.
FIG. 2. Effect surfaces for Ca in urine (a) with injected Cu and CdMT without pretreatment of Zn and (b) with injected Zn and CdMT without pretreatment of Cu.
PROTECTION
AGAINST
Cd NEPHROTOXICITY
If a single high dose of 0.4 mg Cd/kg as CdMT is given (Xc, = - I), the p value for the sum of all coefficients monitoring the protective effect of Zn on calciuria in Model 3b would be calculated to be 0.05. Calcium in the renal cortex. In order to assessthe damage of kidney, the Ca concentration in renal cortex (CaK) was analyzed. A similar result in Group 2 to those for the other two indices mentioned above was observed; this group reached the highest level of Ca in renal cortex among the 10 groups (Table 2) while the rats in Group 1 (0.1 mg Cd/kg) displayed no difference from the control group (Group 10). Both Zn and Cu can protect the kidney damage caused by the injection of CdMT (Fig. 3a). In Model 4b the coefficients are all statistically significant (p < 0.05) except for the coefficient of the interaction term between Cu and Zn (bz,,-,) and the three-variable interaction (see Table 3): ln(CaK)
= 2.94 - 0.1 82Xz, - 0.268&,
+ 0.669&~~
+ O.O76Xz,X-, -
0.3’h&dMT
o.o93X&&u&,,MT
Cab).
The negative signs of the coefficients for the terms X,, , Xc,,, show that the effect of high doses and &&&dMT of CdMT can effectively be reduced by either Cu or Zn or both. The presence of interaction terms, i.e., Cu or Zn with CdMT, predicts that a certain dose of Cu and Zn is more effective at higher doses of CdMT than at lower ones. Similar to that of the calciuria, the Ca concentration in the renal cortex could be more efficiently decreased by Cu pretreatment than by Zn. Although slightly outside the experimental domain, the model suggests that the effect of 0.4 mg Cd/kg as CdMT on Ca concentration in the renal cortex can be reduced to that in the dose of 0.28 mg Cd/kg if 26 mg/kg Zn or 7.98 mg/kg Cu is administered 24 hr before the injection of CdMT. &n&dMTT
243
Cadmium concentration in the renal cortex. Concentrations of 13.25 + 1.62, 26.19 -t 2.61, and 19.10 + 5.25 pg/g wet weight (means fr SD) were found in Groups 1, 2, and 6, respectively. MT concentration in renal cortex. Concentrations of 0.535 + 0.248, 2.129 + 0.910, and 4.143 + 1.363 mg/g wet weight (means + SD) were found in groups 1, 2, and 6, respectively. During the modeling no quadratic terms, i.e., X; are included. Such terms may exist for Ca in kidney and are indicated by comparing the model intercept (bO) with the experimental values with all the variable settings at their intermediate values. Since it is not possible from the present design to resolve the individual quadratic terms of Cu, Zn, and CdMT, the quadratic terms are not included in the model. To resolve this term an at least three-level design is necessary (in this case with 15 groups). DISCUSSION
- O.l14Xz,Xc~~~ +
BY Zn OR Cu
Models for the protective effects of Zn and Cu against the renal toxic effects of CdMT have been determined and illustrated using multiple regression and effect surface plotting. A full factorial design with a center point was used for the experimental settings for the nine groups of rats studied. With this experimental design we have quantitatively estimated the effects of CdMT, Zn, and Cu on kidneys of rats as well as the interaction between al1 three variables. We found that the pretreatments with a certain dose level of Zn and/or Cu can protect rat kidney against the nephrotoxicity of CdMT. The important effects of Cu or Zn pretreatments on calciuria induced by CdMT, some of which have not previously been reported, were identified. The Eqs. 2b, 3b, and 4b can be used to estimate the values of the toxic effects of CdMT at different dose levels of Zn, Cu, and Cd when they are expressed in coded form (see under Materials and Methods) provided that the levels of
FIG. 3. Effect surfaces for Ca in renal cortex with injected Cu and Zn and (a) with 0.25 mg Cd/kg and (b) with 0.4 mg Cd/kg administered asCdMT.
244
LIU
ET
AL.
doses are within or quite close to the experimental domain. the events underlying the appearance of calciuria are different Furthermore, a model can be used to find an equation ex- from those for proteinuria. These results are also of interest pressing the level of Cu or Zn which can reduce the toxic in relation to our previous findings that the calciuria appeared effect of interest to a certain value after a specific Cd dose. earlier than proteinuria (Jin et al., 1987). Possibly, the deThis can be done by introducing a preset coded dose level creased calciuria was fully picked up during the 24-hr urine of Cd and a certain value (In transformed) of a toxic effect. collection period of the present experiment, while that might Then the model will be transformed to an expression con- not be the case for proteinuria. taining the coded Cu and Zn levels as unknowns. Thus, by By multiple regression analysis of results obtained in this setting a level for one of the two metals, the level for the study, it has been revealed that the Ca concentrations in other will be estimated, as exemplified under Results. urine and renal cortex could be more efficiently reduced by It has been recognized that a single CdMT injection in Cu than by Zn pretreatment (see Eqs. 2b and 3b). This finding rabbits and rats can cause acute kidney damage, indicated may be related to the interactive effect between Cu and Cd by abnormal urinary indices and calcification of kidney cells because of their different affinity to ligands in the plasma or (Fowler and Nordberg, 1978; Fowler et al., 1987). As com- kidney. The following two explanations might be possible: pared with Group 10 (control), the minimum dose of injected (i) Less CdMT was delivered to the kidney. It is known CdMT which gives an increase in urinary protein in rats is that Cu binds more firmly to MT than to Cd (Bremner, estimated to be 0.19 mg Cd/kg. This value is obtained by 1987); the Cd in the CdMT which is administered 24 hr after inserting Xz, = - 1, Xc, = - 1 and can be considered as the Cu injection might be partly replaced by Cu available from critical dose for proteinuria induced by CdMT injection. other proteins present in plasma. Thus, the concentration Similarly in Eq. 4b, the minimum dose of injected CdMT of Cd bound to MT in plasma might be decreased and thus in rats which gives an increase of Ca concentration in the less CdMT is delivered to the kidney. This was supported renal cortex is estimated to be 0.16 mg Cd/kg. by the fact that in this study a lower kidney concentration The mean value of Cd in the renal cortex at 0.1 mg Cd/ of Cd was observed in rats with Cu pretreatment (see under kg dose of CdMT was 13.25 & 1.62 kg/g wet weight. It sup- Results, Groups 2 and 6). ports the conclusion suggested in several previous investi(ii) MT synthesis in the kidney was promoted. In rats gations (e.g., Nordberg and Nordberg, 1987) that the critical exposed to Cd, kidney MT was reported to be rich in Cu concentration of Cd in renal cortex when administered as (Webb, 1982). Thus, when the kidney contains a certain CdMT is within lo-20 pg/g wet weight. amount of Cu because of the Cu administration, the MT The dose-dependent relationship observed in this study synthesis in kidney cells is induced and protects against the was similar to that observed in a study of CdMT nephroCd nephrotoxicity. A high concentration of MT in the kidtoxicity in rabbits (Fowler and Nordberg, 1978). By a single neys of Cu-pretreated rats was also observed in this study intravenous injection of 0.1 mg Cd/kg as CdMT, these an- (see under Results, Groups 2 and 6). One or both of the imals disclosed a normal cellular architecture in kidney by possible explanations, mentioned above, are likely to account light microscopic evaluation, while those rabbits adminisfor the observation that the Ca concentrations in urine and tered with 0.4 mg Cd/kg showed extensive tubular necrosis. renal cortex were most efficiently reduced by the Cu preIn the present study, it has been shown that both proteintreatment. uria and calciuria caused by CdMT injection could be effecAs shown in Figs. 3a and 3b, when the rats were treated tively reduced by pretreatment with Zn (see under Results, with a higher dose of CdMT (0.25 or 0.4 mg Cd/kg) and a Figs. 1 and 2b). However, the magnitudes of the decreased relatively low dose of Zn or Cu (or both), the Ca level in the protein and Ca concentrations in urine were different. For renal cortex was reduced more effectively. With increasing example, if a single dose of 20 mg Zn/kg is administered to doses of Cu and/or Zn, the additional positive (protective) the rats, the protein concentration in urine after CdMT in- effect will be less pronounced. This may be reasonable when jection of 0.4 mg Cd/kg would be as low as 6.0 mg crea and one considers the competitive binding between Cd and Cu the Ca concentration in urine would be 17.6 pg/mg crea. to the MT in kidney. Since Cu is bound more firmly to MT The former is close to the value when only a single 0.25 mg than Cd, the capacity of MT binding to Cd might be deCd/kg as CdMT is administered and the latter is close to the creased when the Cu concentration in kidney is high. It could value when 0.1 mg Cd/kg as CdMT is administered. One lead to an increased proportion of non-MT-bound Cd in previous study also reported similar findings (Fowler et al., kidney, which has been considered to be responsible for the 1987); when rats are treated with 20 mg Zn/kg prior to the nephrotoxicity of Cd (Sato and Nagai, 1980). The binding injection of 0.4 mg Cd/kg CdMT, the concentration of Ca of Cd to MT might also release Cu ions, which in turn act in urine is the same as that in controls while the protein on the kidney. Several investigations have suggested the imconcentration in urine remains much higher than that of portance of Zn and Cu as modifying factors in the develcontrols. This phenomenon, both from our study and from opment of Cd and mercury nephrotoxicity (Petering et al., that conducted by Fowler et al. (1987) might indicate that 1984; Bogden et al., 1980).
PROTECTION
AGAINST
Cd NEPHROTOXICITY
It is also of interest to discuss the four rats with deviating effect variables reported under Results. The rats were from Groups 5 and 7, i.e., groups with high Cu and low CdMT treatments. They exhibited serious kidney dysfunction, as indicated by their values in all three measured effects (Table 2) when compared to those of the other rats in the same group. It is not known with certainty that Cu caused these effects but a possibility is that a subpopulation of the Wistar rats for which a relatively low dose of Cu might possibly be nephrotoxic exists. In conclusion, the results of this study extend previous reports on the CdMT-injection model as a tool for evaluating Cd-induced injury to renal proximal tubule cells. By the experimental design used in this study, we have obtained empirical models which relate the administered dose levels of Zn, Cu, and CdMT to the three effects monitoring nephrotoxicity. Data from this study have also shown the doseeffect relationships of the interactive effects among CdMT, Zn, and Cu on the kidney. The results showed that CdMTinduced proteinuria and calciuria may be reduced or prevented by adequate Zn and Cu administration in rats and calciuria can be reduced more efficiently by a certain dose of Cu pretreatment than by Zn. In a future paper we will further discuss and analyze other biological effects from the same experimental design. ACKNOWLEDGMENTS X. Y. Liu and Y. Zhou were supported by WHO Grants during part of this study. A grant from the Swedish Environmental Protection Agency (802-832-91-Fx) is gratefully acknowledged. Thanks are also given to M. Nordberg for the gift of CdMT.
REFERENCES Bogden, J. D., Kemp, F. W., Troiano, R. A., Jortner, B. S., TimPone, C., and Giuliani, D. (1980). Effect of mercuric chloride and methylmercury chloride exposure on tissue concentrations of six essential minerals. Environ. Res. 21, 350-359. Box, G. E. P., Hunter, W. G., and Hunter, J. S. (I 978). Statistics for Experimenters. Wiley and Sons, New York. Box, G. E. P., and Draper, N. R. (1987). Empirical Model-Building and Response Surface. Wiley and Sons. New York. Bremner. I., and Davies, N. T. (1976). Studies on the appearance of a hepatic copper-binding protein in normal and zinc-deficient rats. Br. Nutr. 36, 101-l 12. Bremner, I., and Young, B. W. (1976). Isolation of (copper, zinc)-thioneins from the livers of copper-injected rats. Biochem. J. 157, 5 17-520. Bremner, I. (1987). Interactions between metallothionein and trace elements. Prog. Food Nutr. Sci. 11, l-37.
BY Zn OR Cu
245
Chmielnicka, J., Komsta Szumska. E., and Zareba, G. (1983). Effect of interaction between 65Zn, mercury and selenium in rats (retention, metallothionein, endogenous copper). Arch. Toxicol. 53, 165-175. Durnam, D. M., and Palmiter, R. D. (1981). Transcriptional regulation of the mouse metallothionein-I gene by heavy metals. J. Biol. Chem. 256, 5712-5716. Fowler, B. A., and Nordberg, G. F. (1978). The renal toxicity of cadmium metallothionein: Morphometric and X-ray microanalytical studies. Toxicol. Appl. Pharmacol. 46, 609-623. Fowler, B. A., Goering, P. L., and Squibb, K. S. (1987). Mechanism of cadmium-metallothionein-induced nephrotoxicity: Relationship to altered renal calcium metabolism. Experientia Suppl. 52, 66 l-668. Garvey, J. S. (1984). Metallothionein: Structure/antigenicity and detection/ quantitation in normal physiological fluids. Environ. Health Perspect. 54, 117-127. Hare, R. S. (1950). Endogenous creatinine in serum and urine. Proc. Sot. Exp. Biol. 74, 148- 15 1. Jin, T. Y., Leffler, P., and Nordberg, G. F. (1987). Cadmium-metallothionein nephrotoxicity in the rat: Transient calcuria and proteinuria. Toxicology 45,307-3 17. Nordberg, G. F., Goyer, R., and Nordberg, M. (1975a). Comparative toxicity of cadmium-metallothionein and cadmium chloride on mouse kidney. Arch. Pathol. 99, 192-197. Nordberg, M., Nordberg, G. F., and Piscator, M. (1975b). Isolation and characterization of a hepatic metallothionein from mice. Environ. Physiol. Biochem. 125, 396-403. Nordberg, G. F., Fowler, B. A., F&erg, L., Jemelov, A., Nelson, N., Piscator, M., Sandstead, H. H., Vostal, J., and Vouk, V. B. (1978). Factors influencing metabolism and toxicity of metals: A consensus report. Environ. Health Perspect. 25, 3-4 1. Nordberg, M., and Nordberg, G. F. (1987). On the role of metallothionein in cadmium induced renal toxicity. Experientia Suppl. 52, 669-675. Petering, D. H., Loftsgaarden, J., Schneider, J., and Fowler, B. A. (1984). Metabolism of cadmium. zinc and copper in the rat kidney: The role of metallothionein and other binding sites. Environ. Health Perspect. 54, 73-81. Pfitzer, E. A., and Vouk. V. (1986). Genera.1 considerations of dose-effect and dose-response relationships. In Handbook on the Toxicology of Metals (L. Friberg et al., Eds.). Vol. 1, pp. 149-l 74. Elsevier, B.V., Amsterdam. Piscator, M. (1962). Proteinuria in chronic cadmium poisoning. Arch. Environ. Health 5, 325-332. Sandstead, H. H. (1977). Nutrient interactions with toxic elements. In Toxicology of Trace elements (R. A. Goyer and M. A. Mehlman. Eds.), pp. 241-256. Hemisphere, Washington, DC. Sato, M., and Nagai, Y. (1980). Mode of existence of cadmium in rat liver and kidney after prolonged subcutaneous administration. Toxicol. Appl. Pharmacol. 54, 90-99. Sato, M., and Nagai, Y. (1989). Effect of zinc deficiency on the accumulation of metahothionein and cadmium in the rat liver and kidney. Arch. Environ. Contam. Toxicol. 18, 587-593. Webb, M. (1982). Role of metallothioneins and other binding proteins in the renal handling and toxicity of metals. In Nephrotoxicity: Assessment and Pathogenesis (P. H. Bach et al., Eds.), pp. 296-309. St. Edmundsbury Press. Great Britain.