Animal and human diseases due to inadequate or excess amounts of Se in the diet have led to extensive investigations of the metabolism, biochemistry and nutritional requirements of the mineral (Combs and Combs 1984, Shamberger 1984). The metabolic fate of Se in the body has been studied by administering labeled Se and determining the amount of tracer present in blood, tissues and excreta at different time intervals (Burk et al. 1972, Butler and Peterson 1962, Janghorbani et al. 1990a and 1990b, Langlands et al. 1986, Lopez et al. 1969, Rosenfeld and Eppson 1964, Wright 1965, Wright and Bell 1964).

Both in vitro and in vivo studies have shown that Se is readily absorbed from the intestine and extracted rapidly by the liver and erythrocytes (Symonds et al. 1981a and 1981b). Though feces and urine are the major excretory pathways of Se in all species, urinary excretion has been shown to exercise homeostatic control in monogastric animals under conditions of deficiency (Hansard 1987). On the other hand, there has been contradictory evidence on the relative importance of the fecal and urinary routes of excretion in the regulation of Se metabolism in ruminants (Butler and Peterson 1962, Langlands et al. 1986, Lopez et al. 1969, Peterson and Spedding 1963). RUMEN microbes reduce Se compounds in the feed to insoluble selenides, which are excreted in the feces. The presence of higher levels of Se in the feces than in the urine has led to the suggestion (Hansard 1987) that, unlike in monogastric animals, the fecal route may be more important in homeostatic regulation in ruminants. Nevertheless, the kinetic analyses employed in most studies have not been sensitive enough to establish unequivocally the homeostatic mechanism(s) involved. An integrated knowledge of the kinetic parameters of Se absorption, distribution and excretion in the animal as a whole is necessary to elucidate the regulatory steps involved.

Recently, compartmental models of Se metabolism in sheep (Krishnamurti et al. 1989) and humans (Patterson et al. 1989) have been reported using the Simulation Analysis And Modeling (SAAM) computer program (Berman and Weiss 1978). During the course of fitting tracer data from ewes fed Se-deficient rations to the model proposed earlier for Se-replete ewes (Krishnamurti et al. 1989), it became evident that the configuration of the model was inadequate to resolve the urinary and tissue subsystems satisfactorily. It was necessary to make revisions in the earlier model so that it would better describe the kinetics in ewes fed adequate or low levels of Se. The present study was undertaken to determine the kinetic parameters and steady-state transport rates of Se using the revised model and to identify the homeostatic regulatory mechanisms involved.

MATERIALS AND METHODS

Four Dorset X Suffolk nonpregnant ewes (ewes no. 999, 998, 123 and 325) weighing 52 ± 15 kg obtained from the University of British Columbia animal research units were used in this study. They were given free access to hay (95% canary and 5% timothy) containing 9.7% crude protein and 17.6 kJ/g of gross energy. The hay was obtained from a local farm identified to be Se deficient by the British Columbia Ministry of Agriculture and Feed nutritional analysis laboratory, (Kelowna, British Columbia), on the basis of routine feed analysis and incidence of Se deficiency problems.

Random samples of hay were collected on five different occasions during the experiment, mixed thoroughly, and aliquots taken for Se analysis, which confirmed that the hay contained less than 0.126 nmol/g, low enough to be classified as Se deficient for ruminants (National Research Council 1985, Stevens et al. 1985). The ewes had free access to water and iodized block salt. On the basis of the amount of hay available, the ewes were fed Se-deficient hay for a preliminary period of approximately 8 wk prior to the balance and tracer studies to deplete their Se reserves as much as possible. The assessment of the actual Se status of the ewes was based on plasma and tissue Se concentrations.

Selenium concentration in the samples was determined by dry ashing followed by atomic absorption spectrometry (Tam and Lacroix 1982). Radioactivity in blood, plasma and tissues was counted in an automatic gamma counter (Tricarb 4530, Packard Instruments, Downers Grove, IL).

To identify the metabolic sites at which homeostatic regulation of Se takes place, tracer data from one ewe in each treatment group with the maximum experimental duration of 14 d (ewes 100 and 325) were fitted simultaneously in the model. By keeping some kinetic parameters fixed and letting others change individually or in various combinations, the parameters in which changes were both necessary and sufficient to explain the observed differences in the kinetics (Berman 1963) were identified and were considered to be the regulatory steps involved in Se homeostasis.

(1)

Tracer data from ewes fed normal hay [ewes no. 65, 99 and 100, reported earlier (Krishnamurti et al. 1989), and 622 (unreported)] were fitted to the revised model, and statistical differences in the mean kinetic parameters due to dietary Se deficiency were evaluated by the Student's t test using SAS software (SAS 1985).

L (I,J) = Non-linear (primary) parameter representing the fractional rate of flow of material from compartment J to I. (fraction/d). L(0,I) refers to irreversible loss from the system via compartment I. FSD = Mean fractional standard deviation of estimated parameters generated by CONSAM during fitting of the data in individual ewes.  W_k [(QO_k  QC_k) ÷ QC_k]/W_k, where W_k = weight assigned to datum k; QO_k = observed value for datum k and QC_k = calculated value for datum k. S (I,J) = Linear (secondary) parameter which is the summing coefficient referring to the fraction of compartment J seen in I. R(I,J) = Steady state transport rate of material (µmol/d) from compartment J to I. R (I,J) = M(I) · L (I,J). M(I) = Steady state compartment mass (µmole). M(15) = P(10) · (1/S(10,15)), where P(10) is the experimentally determined plasma Se concentration (µmol/L) and 1/(S(10,15)) is the plasma volume of distribution (L). U(1) = Steady state input of new material into compartment 1 from outside the system (µmol/d). The system is assumed to be in steady state during the experimental period when the total rate of appearance in the compartments is equal to the outflow. V_f (Endogenous fecal Se) = R(8,15) (µmol/d). IC(I) = Initial amount of tracer in compartment (I) rendered adjustable by defining it as a P(I) parameter in the input file.

V_i = Se intake (µmol/d); V_F = Total fecal Se (µmol/d) (unabsorbed Se plus endogenously excreted Se). V_a = V_i  (V_F  V_f) (µmol/d).  = Fraction of intake absorbed (U(1)/Vi) or (V_a/V_i).

RESULTS

Table 1. Physiological condition and selenium status of ewes fed hay containing normal or inadequate levels of selenium1 [View Table]

The pattern of rapid disappearance of the tracer from plasma immediately after the injection, followed by a transient increase in radioactivity that gradually decreased with time (Fig. 1), was similar to the one observed earlier in Se-replete ewes (Krishnamurti et al. 1989). Cumulative tracer concentrations in urine and feces for 5 d of the balance trial are also shown in Figure 1 and radioactively in tissues is shown in Fig. 2. The radioactivity in plasma is assumed to be composed of kinetically and chemically different forms of Se and is represented in the model by compartments 1 and 15 (Fig. 3). Compartment 15 is virtually all bound Se (TCA precipitable), whereas compartment 1 is an aggregate of free and bound forms of Se. This is depicted in the model by introducing two summing components, 10 and 11, each representing a linear combination of compartments 1 and 15 such that the coefficients S(10,15) and S(11,15) are identical and S(10,1) and S(11,1) are different.

The liver subsystem is shown (Fig. 3) to be made up of two compartments, one of which (compartment 3) has a small pool size and exchanges rapidly with compartments 1 and 15; the other (compartment 17) is a slowly turning over pool with a big mass. Material from compartment 15 exchanges with a large peripheral compartment 16, whose physiological identity has not been established experimentally. It is possible that compartment 16 represents Se in bone and other tissues that have not been analyzed. Longer experimental periods and analyses of tracer concentration in bone and other tissues that may serve as a sink would be necessary to understand long-term regulatory mechanisms.

In the blood cell subsystem, the total turnover of compartments 12 and 13 was constrained to 100 d based on an average life span of erythrocytes. Compartment 12 represents Se that readily exchanges with plasma (compartment 1) and also with the much larger erythrocyte compartment 13, which seems to serve as a sink (Fig. 3). Most of the cell mass in compartment 13 is deemed to come from compartment 15 via the newly introduced compartments, 26, 27 and 28, with a delay of approximately 3 wk for the incorporation of Se into blood cells.

In the urinary subsystem, two transient compartments 14 and 24 were introduced with Se input from plasma compartments 1 and 15, respectively (Fig. 3). Kidney is considered to be made up of a rapidly exchanging compartment (2) and a slowly turning over one (52). Urinary Se (compartment 4) is assumed to be derived partly from plasma compartment 1 via transient compartment 14, L(4,14), and partly from plasma compartment 15 via transient compartment 24, L(4,24). The source of fecal Se is exclusively from plasma compartment 15 (Fig. 3).

The flow of Se to and from other tissues is considered to occur from plasma compartment 1 (Fig. 3) and not from compartment 16, as envisaged in the earlier model (Krishnamurti et al. 1989). Appropriate equations were included in the input file so that the transport of tracer was constrained to conform to the experimentally determined unlabeled Se in blood, tissues and excretory products. The composite model that incorporates all the subsystems studied is given in Figure 3.

The kinetic parameters of Se metabolism in ewes fed Se-deficient hay are given in Table 2. Mean parameters that differed significantly from Se-replete ewes are the plasma summing coefficients, S(10,1) and S(10,15), incorporation into blood cells, L(13,12) and L(26,15), extraction from plasma by the liver L(3,1), kidney L(2,1) and lung L(6,1), urinary excretion L(4,14), and uptake by the slowly turning over kidney compartment, L(52,2). Intercompartmental steady-state transport rates of unlabeled Se (R(I,J)) in Se-deficient and Se-replete ewes are given in Table 3. In ewes fed Se-deficient hay, significant reductions were found in blood cell uptake, R(13,12), R(12,13) and R(26,15), hepatic extraction, R(3,1), urinary excretion, R(14,1), R(24,15) and R(4,14), and by the unidentified peripheral compartment R(16,15). The plasma entry rate of Se into plasma compartment 1 from outside the system, U(1), represents material absorbed from the gastrointestinal tract. U(1) was significantly (P < 0.01) lower in ewes fed Se-deficient hay than in the Se-replete ewes. The mass (M) of compartments except kidney, (compartment 2), skeletal muscle (compartment 5), heart (compartment 9) and the peripheral compartment 16 was lower in the Se-deficient group than in Se-replete ewes. (Table 3).

Table 2. Kinetic parameters of selenium metabolism in ewes fed hay containing normal or inadequate levels of Se1,2 [View Table]

Table 3. Intercompartmental transport rates, R (I, J), and compartment masses (M) estimated in the model depicting Se metabolism in ewes fed hay containing normal or inadequate levels of Se1 [View Table]

The net or true Se absorption (µmol/d) estimated by the balance method (V_a, Eq. 1) or by using model parameter U(1), was significantly lower in ewes fed Se-deficient hay than in Se-replete ewes (Table 4). However, the fraction of Se intake absorbed () did not show significant differences between the treatments when the balance method (V_a/V_i) was used. On the other hand, when the entry rate (U(1)) of unlabeled Se into plasma compartment 1 from outside the system estimated in the model was taken to represent net or true absorption from the gut, the fraction () of Se intake absorbed (U(1)/V_i)) was significantly higher in ewes fed Se-deficient hay than in Se-replete ewes (Table 4). Using the levels of Se intakes observed in this study, there was a a linear relationship between Se absorption (V_a) and intake (V_i), V_a = 0.26 + 0.48 V_i (r² = 0.9, P < 0.001) and between the plasma entry rate, U(1), and Se intake (V_i), U(1) = 0.24 + 0.13 V_i (r² = 0.6, P < 0.02). However, additional data points of Se intake would be needed to elucidate the relationship accurately. The rate of endogenous fecal Se excretion was higher in normal than in ewes fed Se-deficient hay, but the difference was not significant (P > 0.05). The ratio of endogenous fecal Se to intake (V_f/V_i) was significantly higher in Se-deficient than in normal ewes (Table 4).

Table 4. Comparison of selenium absorption in ewes fed hay containing normal or inadequate levels of Se1 [View Table]

When tracer data from one ewe from each group (ewes no. 325 and 100) were fitted to the revised model simultaneously, it was observed that the kinetic parameters in which changes were necessary and sufficient to account for the differences were hepatic extraction, L(3,1), urinary excretion, L(14,1) and L(24,15), and fecal excretion, L(8,15). Homeostasis of Se metabolism in the ewes under deficiency conditions thus seems to be regulated by increased capacity for absorption from the gut, enhanced hepatic extraction and reduced excretion through urine and feces.

DISCUSSION

Though the model proposed earlier (Krishnamurti et al. 1989) fitted the plasma data in ewes fed Se-deficient hay as well, individual subsystems pertaining to the excretion and tissue exchange were ill defined. For example, urinary excretion was originally suggested to be derived from plasma compartments 1 and 15 without any provision for reabsorption in the renal tubule or exchange with renal tissue. Thus the model was inadequate to elucidate the role of the kidney in homeostatic regulation under deficiency or toxicity conditions. Sec- ondly, using the earlier model, it was not possible to define the kinetics of uptake and retention of Se by blood cells, which have been shown (Krishnamurti et al. 1989) to contain more than 70% of whole blood Se. Thirdly, tissue uptake of Se was proposed to occur by exchange with a peripheral compartment 16, whose physiological or chemical identity was not established, so the kinetics of tissue uptake was poorly resolved. These limitations of the earlier model became more obvious when the model was used to fit data from Se-deficient ewes. It was therefore necessary to revise the model as indicated to enhance its utility under different physiological conditions.

The noteworthy features of Se deficiency were the reduction in fecal and urinary excretion and absorption. In general, fecal Se excretion has been consistently reported to be higher than urinary excretion in ruminants (Burk et al. 1972, Krishnamurti et al. 1989, Langlands et al. 1986, Lopez et al. 1969, Wright 1965, Wright and Bell 1964). This was true even in Se-deficient ewes, the amount of Se in the feces being two to three times higher than in urine (Table 1). These observations, coupled with the fact that orally administered radioactive Se could be readily recovered in the feces, have led to the suggestion (Hansard 1987, Lopez et al. 1969) that in ruminants the fecal route is the major excretory pathway and may therefore have a role in homeostasis. When daily total fecal Se excretion is expressed as a fraction of Se intake (V_F/V_i), there was no significant difference between the two groups of ewes. But the ratio of endogenous fecal Se to intake (V_f/V_i) was significantly higher in deficient than in replete ewes (Table 4). It is therefore appropriate to discuss the significance of fecal Se excretion with reference to Se intake, absorption and contribution from endogenous sources.

The increased fecal Se excretion with higher intake is in general agreement with other reports (Langlands et al. 1986, Lopez et al. 1969). It has also been reported (Lopez et al. 1969) that fecal Se was more sensitive to total organic matter intake than to Se intake per se. This was attributed by the authors to the reduction of dietary Se by enhanced rumen microbial activity and subsequent excretion of the mineral.

True or net absorption of minerals (V_a) is usually calculated from daily dietary intake (V_i) and fecal excretion (V_F) obtained in nutritional balance trials after correcting the latter for endogenous loss (V_f) determined by concurrent tracer techniques using Eq. 1. In the present study, this method of calculation resulted in a rate of absorption that was two to three times higher than the actual rate of entry of unlabeled Se into the plasma, U(1) estimated in the model. This discrepancy may be ascribed to the cumulation of technical and analytical errors inherent in both the balance and tracer methods. Differences in the absorption of different chemical forms of Se and the significantly higher fraction of Se intake excreted endogenously (V_f/V_i) in Se-deficient ewes (Table 4) may also lead to overestimation of V_a by the combined balance and tracer methods. The relative short duration of both the balance and tracer techniques in the present study further confounds the problem. It was therefore suggested earlier (Krishnamurti et al. 1989) that U(1) would be a better estimate of absorption of dietary Se than V_a calculated by Eq. 1. It may be justified to assume that gastrointestinal absorption occurs only through plasma compartment 1 and accounts solely for the entry of new material entering into the system from outside the model. The linear relationship between the plasma entry rate, U(1), and the dietary intake supports this contention. Using U(1) as an estimate of true or net absorption, the fraction of Se intake absorbed (), (U(1)/V_i) is found to be 1.5 times higher in Se-deficient than in normal ewes (Table 4), indicating the homeostatic capacity of the gut in absorbing a larger fraction of dietary Se under deficiency conditions. In this context it is relevant to consider the human selenite model (Patterson et al. 1989), wherein the authors have proposed four plasma components and estimated that approximately 84% of the administered dose was absorbed. In our study, such a high level of absorption (Table 4) was observed only when it was calculated by the tracer-balance method. In the other ewes, the level of absorption ranged from 18 to 53%. Though these differences could be partly ascribed to the influence of microbial activity and a lower rate of passage of digesta along the gastrointestinal tract in ruminants, the important comments made by Patterson et al. (1989) of the difficulties arising from the presence of both unabsorbed as well as absorbed resecreted label in the feces need to be reemphasized. Therefore the use of the plasma entry rate (U(I)) estimated in a composite model as a measure of absorption would overcome the problem of identifying and quantifying the different components of fecal Se experimentally. This comment is applicable even to other minerals.

The metabolic fate of inorganic and organic Se compounds administered orally, intraruminally or intravenously in lactating goats has been reported recently (Apsila 1991). The mean net absorption in these studies was 64%, comparable to the value of 68.5% observed in the present studies (Table 4); however, the mean net absorption was reduced to 32% in our studies when the entry rate, U(1), of Se into plasma was used to represent net absorption, indicating the critical role of methodology in the calculations. The advantage of kinetic modeling of whole-animal metabolism is that physiological processes occurring in different subsystems of the body are also taken into account in constructing the model. Thus the use of plasma entry rate as a reliable estimate of net absorption seems justified.

Regardless of the method used for estimating absorption, the accuracy eventually depends on the accuracy of the estimation of endogenous excretion, V_f. Recently, elegant techniques for the quantitative determination of endogenous secretion and excretion of zinc have been reported (Wastney and Henkin 1989) involving the introduction of additional compartments in the human model. The importance of estimating V_f accurately can be appreciated by the fact that in deficient ewes almost 75% of fecal Se was of endogenous origin (Table 4).

Initially it was thought that the high fecal endogenous loss in Se-deficient ewes was due to errors in the collection of excreta, analysis and/or computation. The fact that other workers (Janghorbani et al. 1990b, Peterson and Spedding 1963), using different calculation methods, have also arrived at similar conclusions shows that this was not the case. The ratio of the specific radioactivity in the feces to plasma has generally been used to estimate endogenous excretion of minerals. However, this is true only if the chemical moiety of the tracer in the plasma analyzed is the immediate precursor of fecal Se. Because in most studies, including the present one, chemical characterization of radioactive selenocompounds in the plasma or in the feces has not been performed, the method is subject to errors.

The possible sources of endogenous fecal Se include biliary and pancreatic secretions, saliva, blood cells and tissue mobilization. The direct measurement of Se excretion into bile has been reported in cattle (Symonds et al. 1981b) and sheep (Langlands et al. 1986) using surgical techniques in which the flow of bile was diverted into a duodenal pouch accessible through a reentrant cannula. Selenium excretion via saliva was measured by cannulating the parotid duct in sheep (Langlands et al. 1986). The salivary Se concentration was reported to be 0.0032 mg/mL, compared with 0.0086 mg/mL in bile. Whether the precursors of these secretions are derived from the general circulation or from Se absorbed from the gut and resecreted into the duodenum is not known. The small amount of Se absorbed in deficient ewes, the high rate of hepatic extraction and the rapidity of the appearance of the tracer in feces after intravenous injection suggest that the precursors of biliary and pancreatic Se are derived from the general circulation. Evidence for an increase in total endogenous loss under deficient conditions comes from recent reports (Janghorbani et al. 1990a and 1990b) in which, using stable Se isotopes, the authors have used calculations based on the concept of selenite exchangeable metabolic pools to demonstrate an increased whole-body endogenous Se loss in adult rats fed Se-restricted Torula yeast diets.

On the basis of tissue levels and the ratio of specific radioactivity in depleted rats to that in controls, it has been reported (Behne et al. 1988, Behne and Hofer-Bosse 1984) that there was a preferential transport of Se to certain target tissues such as the endocrine organs and brain in Se-depleted rats. The specific radioactivity of Se in the tissues in the present study (not reported) was three to seven times higher in Se-deficient than in Se-replete ewes and supports the concept of selective tissue retention and redistribution during deficiency.

During the course of building the present model, the possibility of reabsorption of Se into kidney from the transient urinary compartments 14 and 24 was tested by introducing the parameters L(2,14) and L(52,24). However, even in deficient ewes, these parameters were not found to be necessary to fit the data, indicating that the reduction in urinary excretion during deficiency was due to difference in renal clearance and not to reabsorption.

Although urinary excretion has been shown to be important in Se homeostasis in monogastric animals (Oster and Prellwitz 1990), the excretion of large amounts of Se in the feces in ruminants has led to the suggestion (Hansard 1987) that a different mechanism may operate. To clarify this discrepancy, data from one deficient ewe (no. 325) and one replete ewe (no. 100) were fitted simultaneously to the revised model. On the basis of the minimal change postulate (Berman 1963), it is possible to identify the parameter(s) in which changes were necessary and sufficient to explain the kinetic differences between treatments. Accordingly, changes were required in hepatic extraction, L(3,1), urinary excretion, L(14,1) and L(24,15), and fecal excretion, L(8,15).

It may be concluded that during Se deficiency the fraction of intake absorbed () increases to cope with the meager dietary supply. Although feces is the major route of excretion of Se in ruminants as observed in this and other studies, homeostasis is achieved through a combination of hepatic, fecal and renal regulatory mechanisms. Further improvements in experimental design would increase the accuracy of estimation of endogenous production and absorption. Refinements to the model may be accomplished by the introduction of additional compartments incorporating tracer activity in different sections of the gastrointestinal tract. To elucidate the long-term regulatory mechanisms of Se metabolism, particularly the extent of tissue mobilization and/or distribution under Se deficiency conditions, it would be necessary to extend the experimental duration and undertake chemical characterization of radioactive selenium compounds. The determination of the changes in the specific radioactivity of the target tissues (red blood cells and bone) at frequent intervals is also necessary.