Deletion of Selenoprotein P Alters Distribution of Selenium in the Mouse

Selenoprotein P (Se-P) contains most of the selenium in plasma. Its function is not known. Mice with the Se-P gene deleted (Sepp−/−) were generated. Two phenotypes were observed: 1) Sepp−/−mice lost weight and developed poor motor coordination when fed diets with selenium below 0.1 mg/kg, and 2) maleSepp−/− mice had sharply reduced fertility. Weanling male Sepp+/+,Sepp+/−, and Sepp−/−mice were fed diets for 8 weeks containing <0.02–2 mg selenium/kg.Sepp+/+ and Sepp+/−mice had similar selenium concentrations in all tissues except plasma where a gene-dose effect on Se-P was observed. Liver selenium was unaffected by Se-P deletion except that it increased when dietary selenium was below 0.1 mg/kg. Selenium in other tissues exhibited a continuum of responses to Se-P deletion. Testis selenium was depressed to 19% in mice fed an 0.1 mg selenium/kg diet and did not rise toSepp+/+ levels even with a dietary selenium of 2 mg/kg. Brain selenium was depressed to 43%, but feeding 2 mg selenium/kg diet raised it to Sepp+/+ levels. Kidney was depressed to 76% and reachedSepp+/+ levels on an 0.25 mg selenium/kg diet. Heart selenium was not affected. These results suggest that theSepp−/− phenotypes were caused by low selenium in testis and brain. They strongly suggest that Se-P from liver provides selenium to several tissues, especially testis and brain. Further, they indicate that transport forms of selenium other than Se-P exist because selenium levels of all tissues except testis responded to increases of dietary selenium inSepp−/− mice. Selenoprotein P (Se-P) contains most of the selenium in plasma. Its function is not known. Mice with the Se-P gene deleted (Sepp−/−) were generated. Two phenotypes were observed: 1) Sepp−/−mice lost weight and developed poor motor coordination when fed diets with selenium below 0.1 mg/kg, and 2) maleSepp−/− mice had sharply reduced fertility. Weanling male Sepp+/+,Sepp+/−, and Sepp−/−mice were fed diets for 8 weeks containing <0.02–2 mg selenium/kg.Sepp+/+ and Sepp+/−mice had similar selenium concentrations in all tissues except plasma where a gene-dose effect on Se-P was observed. Liver selenium was unaffected by Se-P deletion except that it increased when dietary selenium was below 0.1 mg/kg. Selenium in other tissues exhibited a continuum of responses to Se-P deletion. Testis selenium was depressed to 19% in mice fed an 0.1 mg selenium/kg diet and did not rise toSepp+/+ levels even with a dietary selenium of 2 mg/kg. Brain selenium was depressed to 43%, but feeding 2 mg selenium/kg diet raised it to Sepp+/+ levels. Kidney was depressed to 76% and reachedSepp+/+ levels on an 0.25 mg selenium/kg diet. Heart selenium was not affected. These results suggest that theSepp−/− phenotypes were caused by low selenium in testis and brain. They strongly suggest that Se-P from liver provides selenium to several tissues, especially testis and brain. Further, they indicate that transport forms of selenium other than Se-P exist because selenium levels of all tissues except testis responded to increases of dietary selenium inSepp−/− mice. Se-P 1The abbreviations used are: Se-Pselenoprotein PSeppgene encoding selenoprotein PGSHPxglutathione peroxidaseESembryonic stem is the most extreme example of the interesting protein class known as selenoproteins. Selenoproteins contain selenocysteine in their primary structures (1Stadtman T.C. Annu. Rev. Biochem. 1996; 65: 83-100Crossref PubMed Scopus (824) Google Scholar). Insertion of this 21st amino acid involves an expansion of the genetic code by redefinition of specific internal UGA stop codons (2Low S.C. Berry M.J. Trends Biochem. Sci. 1996; 21: 203-208Abstract Full Text PDF PubMed Scopus (393) Google Scholar). In the majority of selenoproteins there is a single selenocysteine that is at the active site of an enzyme, and it plays a pivotal role in a redox reaction. By contrast, Se-P has many selenocysteines, ranging from 10 in mice and humans to 17 in zebrafish (3Kryukov G.V. Gladyshev V.N. Genes Cells. 2000; 5: 1049-1060Crossref PubMed Scopus (113) Google Scholar, 4Steinert P. Bächner D. Flohé L. Biol. Chem. 1998; 379: 683-691Crossref PubMed Scopus (42) Google Scholar). The function of Se-P is not known, and there is no direct evidence that its selenocysteine residues play a chemically active role. selenoprotein P gene encoding selenoprotein P glutathione peroxidase embryonic stem Se-P is an extracellular glycoprotein that contains most of the selenium in plasma (5Read R. Bellew T. Yang J.-G. Hill K.E. Palmer I.S. Burk R.F. J. Biol. Chem. 1990; 265: 17899-17905Abstract Full Text PDF PubMed Google Scholar). Its abundance is such that Se-P in rat plasma accounts for 8% of the selenium in the animal. The liver is the primary source of plasma Se-P (6Kato T. Read R. Rozga J. Burk R.F. Am. J. Physiol. 1992; 262: G854-G858Crossref PubMed Google Scholar), although virtually all tissues express it. Substantial amounts of Se-P are synthesized, because the plasma half-life of its selenium is only 3–4 h (7Burk R.F. Hill K.E. Read R. Bellew T. Am. J. Physiol. 1991; 261: E26-E30Crossref PubMed Google Scholar). This indicates that selenium cycles through plasma Se-P at a high rate. Because of its plasma location, Se-P has been postulated to be a selenium transport protein (8Motsenbocker M.A. Tappel A.L. Biochim. Biophys. Acta. 1982; 719: 147-153Crossref PubMed Scopus (166) Google Scholar). The selenium in Se-P is present in its primary structure as selenocysteine residues, and so delivery of its selenium to a cell would require degradation of the Se-P and catabolism of its selenocysteine. That would make a selenium transport role for the protein costly to the animal. Nevertheless, evidence suggesting a transport function has been presented (7Burk R.F. Hill K.E. Read R. Bellew T. Am. J. Physiol. 1991; 261: E26-E30Crossref PubMed Google Scholar). Other studies have shown that Se-P is a preferred source of selenium for Jurkat cells (9Saito Y. Takahashi K. Eur. J. Biochem. 2002; 269: 5746-5751Crossref PubMed Scopus (144) Google Scholar) and embryonic neurons (10Yan J. Barrett J.N. J. Neurosci. 1998; 18: 8682-8691Crossref PubMed Google Scholar). Another hypothesis of Se-P function is that it protects against oxidative injury (11Burk R.F. Hill K.E. Awad J.A. Morrow J.D. Kato T. Cockell K.A. Lyons P.R. Hepatology. 1995; 21: 561-569PubMed Google Scholar). Its presence has been correlated with protection against hepatic sinusoidal endothelial cell injury by diquat in the rat (12Atkinson J.B. Hill K.E. Burk R.F. Lab. Invest. 2001; 81: 193-200Crossref PubMed Scopus (30) Google Scholar). Moreover, it binds to endothelial cells throughout the animal, presumably through its heparin binding properties (13Burk R.F. Hill K.E. Boeglin M.E. Ebner F.F. Chittum H.S. Histochem. Cell Biol. 1997; 108: 11-15Crossref PubMed Scopus (75) Google Scholar, 14Hondal R.J. Ma S. Caprioli R.M. Hill K.E. Burk R.F. J. Biol. Chem. 2001; 276: 15823-15831Abstract Full Text Full Text PDF PubMed Scopus (67) Google Scholar). To facilitate studies on Se-P function we have produced mice that lack Se-P using homologous recombination. We report some of the characteristics of these animals here, along with results that support a selenium transport or distribution function for Se-P. Restriction enzymes and ligases were obtained fromPromega (Madison, WI), New England Biolabs (Beverly, MA), and MBI Fermentas (Amherst, NY). Cloning vectors, pBluescript and pBC, were obtained from Stratagene (La Jolla, CA). The loxP-flanked Neo® gene, pKT1LoxA, (15Greer J.M. Capecchi M.R. Neuron. 2002; 33: 23-34Abstract Full Text Full Text PDF PubMed Scopus (310) Google Scholar), and the TK2 gene (16Deng C. Thomas K.R. Capecchi M.R. Mol. Cell. Biol. 1993; 13: 2134-2140Crossref PubMed Scopus (85) Google Scholar) were generous gifts of Dr. Kirk R. Thomas, University of Utah. Oligonucleotides were synthesized by core laboratory facilities at Vanderbilt University Medical Center and the University of Utah. [32P]dATP and [32P]dCTP were obtained from PerkinElmer Life Sciences. [75Se]Selenite (specific activity, 800 mCi/mg selenium) was obtained from the University of Missouri Research Reactor Facility, Columbia, MO. NADPH was purchased from United States Biochemical Corp. (Cleveland, OH). Glutathione reductase was purchased from Sigma. All other chemicals were of reagent grade. A P1 plasmid-containing mouse genomic DNA for Sepp was purchased from Genome Systems, Inc. (St. Louis, MO). PCR primers MG5 (5′-GATTTGTGCAAACATGGAGAAATC-3′) and MG7 (5′-GAATGTAAGAGTAGGAAGACAAAG-3′) were supplied to Genome Systems and used to screen their murine 129/Sv P1 library. One clone, MG6138, was shown to contain Sepp by Southern analysis. P1-DNA, prepared using a procedure from Qiagen (Chatsworth, CA), was digested withEcoRI, HindIII, or BamHI. Southern analysis of the digested MG6138 DNA, using [32P]dCTP-labeled 16C1 cDNA (17Hill K.E. Lloyd R.S. Yang J.-G. Read R. Burk R.F. J. Biol. Chem. 1991; 266: 10050-10053Abstract Full Text PDF PubMed Google Scholar), showed twoHindIII fragments (approximate sizes, 6 and 2 kb), a single EcoRI fragment (10 kb), and a singleBamHI fragment (10 kb). The MG6138-HindIII digest was subcloned into pBluescript II KS. Subclones were selected by screening with [32P]dCTP-labeled 16C1 cDNA. Double-stranded DNA prepared from these clones was sequenced by the Sanger dideoxy termination method. Intron-exon boundaries were identified by a comparison of genomic DNA sequences with the sequence of a mouse selenoprotein P cDNA (MB23A). MB23A was obtained by screening a mouse brain cDNA library with the HindIII fragment of MG6138. The cDNA library was a gift from Dr. Thomas Quertermous, Vanderbilt University. MG6138P1 DNA was prepared and digested with BamHI. The fragments from the digest were ligated into pBluescript II and transformed into Escherichia coli. Bacterial colonies were screened by PCR using primers SeP4342 (5′-CGCAGAACGCACAAAGAATGTAGATGGC-3′) and SePBsuR (5′-GTACCCTTAGGCCAGAAGAGGGCACTGGG-3′). One clone (P1#10) selected was determined to contain a 12-kb fragment of genomic DNA suitable for construction of a Sepp knockout vector (SePCKNT). The SePCKNT vector was constructed using the following steps (Fig. 1). 1) The Neo® cassette, KT1LoxA, was passed through E. coliDM1 strain to demethylate the XbaI site and then was digested with XbaI. KT1LoxA was then inserted at theAvrII site in P1#10, producing Sep#10Neo®. 2) SePCKN-1 was prepared by ligation of a SalI/KpnI 7.1-kb fragment of SeP#10Neo®, a KpnI/NotI 6.3-kb fragment of P1#10, and a SalI/NotI 3.4-kb fragment of pBC. 3) SePCKNT was prepared by ligation of aBamHI/NotI 2.9-kb fragment of pBluescript, aSalI/BamHI 13.4-kb fragment of SePCKN-1, and aNotI/XhoI 2.3-kb fragment of TK-2. SePCKNT had a size of 18.6 kb. The total homology with Sepp genomic DNA was 12 kb. The loxP sites within the Neo® cassette gave translational stops in all three reading frames. ES cell transformation with SePCKNT and blastocyst injection were carried out in the Mouse Core Facility at the University of Utah using published procedures (18Thomas K.R. Capecchi M.R. Cell. 1987; 51: 503-512Abstract Full Text PDF PubMed Scopus (1951) Google Scholar, 19Thomas K.R. Capecchi M.R. Nature. 1990; 346: 847-850Crossref PubMed Scopus (800) Google Scholar). PCR assays were used to determine the genotype of ES cells and adult mice. DNA was extracted from ES cells and from tail biopsies using established procedures. DNA was resuspended in Tris-EDTA buffer (10 mm Tris·Cl, 1 mm EDTA, pH 7.6). Approximately 1 μg of DNA was dissolved in 50 μl of PCR lysis buffer (50 mm KCl, 1.5 mm MgCl2, 10 mm Tris·Cl (pH 8), 0.01% gelatin, 0.45% Nonidet P-40, 0.45% Tween 20), denatured at 95 °C for 5 min, and quick-chilled on ice. Two μl of the denatured DNA solution was amplified for 35 cycles (94 °C for 30 s, 66 °C for 20 s, 72 °C for 60 s) in a 10-μl reaction mixture. The primers used to screen for ES-positive cells were MoSePS6 (5′-GAAGACTGTAATCGCTATAACCACTGTCCAG-3′) and ACNeoS2 (5′-GGTGTTGGGTCGTTTGTTCGGATCG-3′). The primers used to screen tissue DNA were MoSePS14 (5′-GCCATCAGGGCTCAGTGCAG-3′), MoSePA16 (5′-GTTCAAAGCCCAGGAATGCCACAG-3′), and ACNeoS2. PCR product sizes are: wild type (Sepp+/+), 900 bp from MoSePS14 and MoSePA16; homozygote (Sepp−/−), 500 bp from MoSePS14 and ACNeoS2; heterozygote (Sepp+/−), 900 and 500 bp. An additional PCR product of 2.1 kb is predicted for the primers MoSePS14 and MoSePA16 in Sepp+/− andSepp−/− mice. PCR conditions were not optimized for production of this product. Adult Sepp+/− andSepp−/− mice were transferred to the animal facility at Vanderbilt University. The Vanderbilt University Institutional Animal Care and Use Committee approved animal protocols for studies at Vanderbilt, and the corresponding University of Utah committee approved the protocols used to generate the knockout mouse. The mice were housed in plastic cages with aspen shavings as bedding material. The light/dark cycle was 12 h:12 h. Mice received pelleted rodent chow (selenium content, 0.29 ± 0.01 mg/kg) and water ad libitum except when fed diets containing specific amounts of selenium. Matings were set up betweenSepp−/− males andSepp−/− females, betweenSepp−/− males andSepp+/− females, between Sepp+/− males andSepp−/− females, and betweenSepp+/− males andSepp+/− females. Pups were weaned 21 days after birth and separated by sex. For experiments in which different amounts of selenium were fed, aTorula yeast-based diet was used (20Burk R.F. Methods Enzymol. 1987; 143: 307-313Crossref PubMed Scopus (71) Google Scholar). The basal form of this experimental diet contained <0.02 mg of selenium/kg. Sodium selenate was added to this diet during mixing to give final added selenium concentrations that ranged from 0.05 to 2 mg/kg. Male weanling mice were fed basal or selenium-supplemented diet for 8 weeks. They were weighed weekly, and mice that had lost 20% of their highest body weight were euthanized. At 8 weeks the mice were anesthetized with isoflurane, and blood was removed from the inferior vena cava. The blood was treated with Na2EDTA (1 mg/ml) to prevent coagulation, and plasma was separated by centrifugation. Liver, kidney, heart, testis, and brain were harvested and frozen immediately in liquid nitrogen. Plasma and tissues were stored at −80 °C. Tissue homogenates (10%) were prepared in 0.1 m potassium phosphate, pH 7.5. Supernatants from centrifugation of the tissue homogenates at 13,000 ×g for 30 min were used for measurement of glutathione peroxidase activity. The coupled method was used with 0.25 mm hydrogen peroxide as substrate (21Lawrence R.A. Burk R.F. Biochem. Biophys. Res. Commun. 1976; 71: 952-958Crossref PubMed Scopus (3005) Google Scholar). Plasma Se-P was measured with a radioimmunoassay (22Hill K.E. Xia Y. Åkesson B. Boeglin M.E. Burk R.F. J. Nutr. 1996; 126: 138-145Crossref PubMed Scopus (178) Google Scholar) that utilized the polyclonal antibody preparation 695. Selenium was measured using a modification of the fluorometric assay of Koh and Benson (23Koh T.S. Benson T.H. J. Assoc. Off. Anal. Chem. 1983; 66: 918-926PubMed Google Scholar, 24Sheehan T.M.T. Gao M. Clin. Chem. 1990; 36: 2124-2126Crossref PubMed Scopus (78) Google Scholar). The limit of detection of this assay is 1 ng of selenium. Adult mice of all genotypes that were fed the chow diet were each injected intraperitoneally with 15 μCi of [75Se]selenite (in 0.15 m NaCl). Blood was obtained from the mice 24 h after 75Se administration. Plasma was separated by centrifugation and subjected to SDS-PAGE. After being stained with Coomassie Blue, the gel was dried and exposed to Kodak XAR film. In a separate experiment, some of the 75Se-labeled plasma was subjected to immunoprecipitation by polyclonal antibodies raised in rabbits against rat selenoprotein P (25Hill K.E. Chittum H.S. Lyons P.R. Boeglin M.E. Burk R.F. Biochim. Biophys. Acta. 1996; 1313: 29-34Crossref PubMed Scopus (41) Google Scholar) and by polyclonal antibodies raised in rabbits against human GSHPx-3 (pAb 3495, a generous gift of K. R. Maddipati, Cayman Chemical, Ann Arbor, MI). The immunoprecipitates were separated by SDS-PAGE, and75Se-labeled proteins were identified by autoradiography. Male mice of all three genotypes were fed the experimental diet supplemented with 0.1 mg of selenium/kg for 6–8 weeks from the time of weaning. Then each was administered 15 μCi of [75Se]selenite (in 0.15 m NaCl) by gavage. They were housed individually for 24 h and were then anesthetized and exsanguinated by collection of blood from the inferior vena cava. Liver, kidney, testis, and brain were harvested and weighed. The 75Se content of plasma and tissues was determined in a Commugamma 1282 (Amersham Biosciences). Results were analyzed using Student'st test or by analysis of variance with post hocanalysis for statistical differences using the Scheffe test. Significance was set at p < 0.05. Calculations were done on a Macintosh G4 using Statview, version 5.0.1 (SAS Institute, Cary, NC). Se-P knockout mice were produced by homologous recombination using genomic DNA cloned from an Sv-129 P1 library that had been mutated using the strategy shown in Fig. 1. C57Bl/6 blastocysts were injected with ES cells heterozygous for the Sepp mutation and then implanted into pseudopregnant females. A male chimera was identified among the offspring. This male was mated with two C57Bl/6j female mice, and the heterozygote progeny (Sepp+/−) from these matings were used to establish the Se-P knockout mouse colony. Twenty-four hours after75Se had been injected into animals of each genotype, plasma was obtained and subjected to SDS-PAGE. Fig.2A shows the autoradiograph of the resulting gel. Both theSepp+/+ and Sepp+/−plasma samples yielded bands characteristic for Se-P and GSHPx-3.Sepp−/− plasma exhibited only the band of75Se radioactivity corresponding to glutathione peroxidase. The Se-P band was not present in the Sepp−/−lane. The identity of each of the radioactive proteins was verified by precipitation using specific antibody preparations (Fig.2B). Northern analysis of liver RNA was carried out and showed a gene-dose effect with no signal evident in theSepp−/− lane (Fig. 2C). Fig. 2D shows the results of measuring Se-P by competitive radioimmunoassay of plasma samples from the three genotypes. A gene-dose effect is evident that mirrors the liver mRNA results (Fig. 2C). Thus, Se-P has been eliminated from the plasma of the knockout mice, and its concentration is half that of the wild types in the plasma of the heterozygotes. Offspring ofSepp+/− and Sepp−/−animals fed rodent chow were characterized. The ratio of male to female progeny was ∼50/50 regardless of the genotype of the parents (data not shown). Mating of Sepp+/− males andSepp+/− females resulted in viable pups with the predicted genotype distribution (TableI, top). Those pups survived to weaning (Table 1, bottom). Mating of Sepp+/−males and Sepp−/− females produced fewerSepp−/− pups than predicted, and 31% of these homozygous pups died before weaning. WhenSepp−/− males were mated withSepp+/− and Sepp−/−females over a period of 6 months, only one litter was born from 18 mating pairs (Table II). In contrast,Sepp+/− males produced many pregnancies withSepp+/− and Sepp−/−females. These results indicate that female homozygotes have difficulties producing and raising homozygote pups and that male homozygotes have sharply reduced fertility.Table IGenotype distribution and viability of pups produced by Sepp+/− siresPup genotypeDam genotypeSepp+/−Sepp−/−Number of pups% of totalNumber of pups% of totalGenotype distribution of pups Sepp−/−88275142 Sepp+/−160497058 Sepp+/+81240Number of pups% SurvivingNumber of pups% SurvivingPup survival to weaning Sepp−/−881003569 Sepp+/−1599970100 Sepp+/+80990 Open table in a new tab Table IIFertility of Sepp−/− and Sepp+/− miceMouse genotypeSire:Sepp+/−Sepp−/−Dam:Sepp+/−Sepp−/−Sepp+/−Sepp−/−Breeding pairs2-aBreeding pairs were fed the chow diet and housed together for up to 6 months.4620108Litters742910Average pups/litter8772-a Breeding pairs were fed the chow diet and housed together for up to 6 months. Open table in a new tab In the initial experiments with our genetically altered mice (Tables I and II), we fed mouse chow containing 0.29 mg of selenium/kg. TheSepp−/− mice grew as well as theSepp+/+ mice (Fig.3A) and appeared healthy. However, when we began feeding a selenium-deficient diet containing <0.02 mg of selenium/kg, the Sepp−/− mice did not survive for more than a few weeks, whereasSepp+/− and Sepp+/+ mice survived. To investigate this sensitivity ofSepp−/− mice to selenium deficiency, selenium and glutathione peroxidase were measured in tissues from male mice of all three genotypes that for 8 weeks had been fed diets containing amounts of selenium ranging from deficient to 20 times the dietary requirement. These mice were obtained by matingSepp+/− males withSepp+/− females. Sepp−/− mice fed the selenium-deficient diet (no selenium supplementation) gained weight for about a week and then began to suffer from loss of motor coordination and weight (Fig.3B). When mice had lost 20% of their highest recorded body weight, they were euthanized. Their average survival time from weaning was 15 days. Sepp−/− mice fed the diet with 0.05 mg of selenium added/kg also had decreased survival times; 2 of 4 mice were sacrificed because of 20% weight loss, one at 3 weeks and the other at 6.5 weeks (Fig. 3C). Thus only 2Sepp−/− mice fed this diet remained alive at the 8-week time point. Animals fed an 0.1 mg of selenium or more/kg diet all survived to 8 weeks without loss of coordination, and weights were similar for all genotypes (Fig. 3D). Two selenoproteins contribute to plasma selenium: extracellular glutathione peroxidase (GSHPx-3) and Se-P. Fig.4A shows the selenium content and glutathione peroxidase activity of plasma at different levels of dietary selenium supplementation. The absence of Se-P from the plasma of Sepp−/− mice is reflected in the very low plasma selenium concentrations measured at all levels of selenium supplementation (Fig. 4A, left panel). Selenium concentrations in Sepp+/− mouse plasma were intermediate between the homozygote and wild type concentrations at each supplementation level. This finding is consistent with the gene-dose effect indicated in Fig. 2D by the Se-P concentrations in plasma. Plasma glutathione peroxidase activity increased as the dietary selenium supplement was increased from 0 to 0.1 mg/kg in wild type (Sepp+/+) mice and maintained that level as dietary selenium was increased further (Fig. 4A, right panel). This result is consistent with the established dietary requirement of 0.1 mg of selenium/kg in mice (26Prohaska J.R. Sunde R.A. Comp. Biochem. Physiol. B. 1993; 105: 111-116Crossref PubMed Scopus (29) Google Scholar, 27Weiss S.L. Evenson J.K. Thompson K.M. Sunde R.A. J. Nutr. 1996; 126: 2260-2267Crossref PubMed Scopus (84) Google Scholar).Sepp+/− plasma glutathione peroxidase activities were not significantly different from activities in theSepp+/+ mice. Sepp−/−mice had lower plasma glutathione peroxidase activities thanSepp+/+ mice at levels of selenium supplementation below 1.0 mg/kg. These results indicate that Se-P is needed to achieve normal plasma activity of GSHPx-3 at dietary selenium concentrations up to 5 times the requirement but not when the dietary selenium concentration is 10 times the required concentration. Liver selenium levels were not affected by deletion of Se-P, except when dietary selenium was below the selenium requirement (Fig.4B, left panel). The twoSepp−/− animals in the 0.05 mg selenium/kg diet group that survived for 8 weeks had levels that were greater than the averages of the other groups. The liver selenium levels of the twoSepp−/− animals that did not survive 8 weeks were also greater than the 8-week averages of the other animals (see legend for Fig. 4). Liver glutathione peroxidase activities had the same pattern (Fig. 4B, right panel) as liver selenium concentrations. These results indicate that theSepp−/− liver retains more selenium than theSepp+/+ liver during selenium deficiency and that deletion of Se-P does not cause a decrease in liver selenium content at any dietary selenium level. They also demonstrate that deletion of Se-P does not impair regulation of selenium concentration in liver when higher levels of selenium are fed. Fig. 5 shows the selenium content of testis, brain, kidney, and heart. Glutathione peroxidase activity was also determined in these tissues, and its pattern (not shown) was similar to that of the selenium concentrations. In each of these tissues the values for Sepp+/+ andSepp+/− mice were similar. The selenium difference between the Sepp+/+ orSepp+/− mice, on the one hand, and theSepp−/− mice, on the other hand, varied by tissue. It was greatest in testis, with selenium concentration being 19% of Sepp+/+ when selenium was fed at the dietary requirement. Moreover, increasing dietary selenium 20-fold did not raise the level of testis selenium into theSepp+/+ range. The brain was also severely affected with its selenium concentration being 43% of control at the dietary requirement. However, brain selenium was raised into theSepp+/+ range by increasing dietary selenium 20-fold. Kidney selenium was less affected, being 76% ofSepp+/+ when the dietary selenium requirement was fed. Moreover, increasing dietary selenium to just 2.5-fold the requirement raised the selenium content ofSepp−/− kidney to the same level found inSepp+/+ kidney. Heart selenium concentration was not significantly affected by genotype. These results indicate that Se-P facilitates selenium accumulation in testis, brain, and kidney. Heart selenium content does not appear to depend on Se-P. Thus, the tissues studied vary in their dependence on Se-P to maintain their selenium concentrations. To determine the effect of Se-P deletion on the disposition of a single dose of selenium, a tracer dose of 75Se as sodium selenite was administered by gavage to mice being fed the diet supplemented with 0.1 mg of selenium/kg. Tissues were harvested 24 h after administration, and their 75Se content was determined; Fig.6 shows the results. Liver showed a trend of containing more of the 75Se administered inSepp−/− mice than it did inSepp+/− or Sepp+/+ mice. Brain and testis contained less 75Se inSepp−/− mice than in the other two genotypes. These results show that absorbed selenium is readily taken up by the liver in Sepp−/− mice but is less well taken up and/or retained by the brain and testis. These results are consistent with the tissue selenium levels shown in Figs. 4 and 5. Mice with Se-P deleted have been produced by homologous recombination. These mice are viable but exhibit a profound alteration in selenium metabolism, which renders them intolerant of low dietary selenium intake. When intake is low they develop impaired movement and coordination and do not maintain their weight. In addition, males have sharply reduced fertility, even when fed enough selenium to prevent the movement abnormality and weight loss. These two phenotypes correlate with the effect of Se-P deletion on the selenium content of the brain and testis (Fig. 5, Band A, respectively). Low selenium in the brain correlates with the apparent neurological impairment, and low selenium in testis appears to underlie the impaired male fertility. Selenium is essential for spermatogenesis (28Behne D. Weiler H. Kyriakopoulos A. J. Reprod. Fertil. 1996; 106: 291-297Crossref PubMed Scopus (195) Google Scholar, 29Ursini F. Heim S. Kiess M. Maiorino M. Roveri A. Wissing J. Flohé L. Science. 1999; 285: 1393-1396Crossref PubMed Scopus (736) Google Scholar). Brain and testis have been identified as tissues that retain selenium well under conditions of extreme selenium deficiency (30Burk R.F. Brown D.G. Seely R.J. Scaief III, C.C. J. Nutr. 1972; 102: 1049-1055Crossref PubMed Scopus (109) Google Scholar, 31Behne D. Hilmert H. Scheid S. Gessner H. Elger W. Biochim. Biophys. Acta. 1988; 966: 12-21Crossref PubMed Scopus (414) Google Scholar). Therefore, a decade ago we compared the ability of these two tissues to take up selenium administered in the form of75Se-labeled Se-P (7Burk R.F. Hill K.E. Read R. Bellew T. Am. J. Physiol. 1991; 261: E26-E30Crossref PubMed Google Scholar). Both tissues took up the75Se avidly. The brain was able to increase its uptake more than 4-fold when selenium deficiency was imposed. In contrast, testis uptake, although brisk, was unaffected by selenium deficiency. These results and the present ones suggest that the brain and testis remove Se-P from plasma to acquire its selenium. Moreover, the brain would appear to be able to up-regulate this uptake mechanism in selenium deficiency. In addition, the neurological phenotype could be prevented by feeding selenium to Sepp−/− mice at the level of dietary requirement or higher. This suggests that the brain can take up other plasma forms of selenium than Se-P and therefore utilizes two or more mechanisms to obtain the selenium it needs. No evidence was found that the testis is able to utilize a form of selenium other than Se-P. The severity of the effect of Se-P deletion on selenium concentration forms a continuum in the tissues we examined. Although there were severe effects on the testis and brain, the kidney was only moderately affected and the heart was not significantly affected. This indicates that other transport forms of selenium than Se-P exist and that each tissue may have preferred forms of plasma selenium from which it obtains the element. In the mice studied here, plasma selenium was present as Se-P, GSHPx-3, and small molecule forms. The origin of plasma Se-P is mostly liver (6Kato T. Read R. Rozga J. Burk R.F. Am. J. Physiol. 1992; 262: G854-G858Crossref PubMed Google Scholar) and that of GSHPx-3 is kidney (32Avissar N. Ornt D.B. Yagil Y. Horowitz S. Watkins R.H. Kerl E.A. Takahashi K. Palmer I.S. Cohen H.J. Am. J. Physiol. 1994; 266: C367-C375Crossref PubMed Google Scholar). The intestine is known to release a small molecule form of selenium that is taken up by the liver (6Kato T. Read R. Rozga J. Burk R.F. Am. J. Physiol. 1992; 262: G854-G858Crossref PubMed Google Scholar), but additional small molecule forms of selenium, other than excretory metabolites, have not been detected. It can be inferred that the kidney can also take up a small molecule form of selenium, because kidney is the source of plasma GSHPx-3 and the selenium concentration of kidney responds to dietary selenium in animals lacking Se-P (Fig.5C). There is no direct evidence that GSHPx-3 serves to supply tissues with selenium. However, one finding in this study is consistent with such a function. Plasma glutathione peroxidase activity did not rise to its control level until 10 times the dietary requirement of selenium was fed (Fig. 4A), whereas kidney selenium level rose to control at only 2.5 times the requirement (Fig. 5C). If it is assumed that synthesis and secretion of GSHPx-3 was normalized when kidney selenium was at control levels, then increased removal of GSHPx-3 from plasma must be invoked to explain the low glutathione peroxidase activity in plasma at that point. Mice with deletion of GSHPx-3 will facilitate study of its putative transport role. The liver has a central role in selenium metabolism. It receives small molecule selenium directly from the intestine and is the major organ for removal of selenium from the dietary form, selenomethionine, via the trans-sulfuration pathway (33Esaki N. Nakamura T. Tanaka H. Suzuki T. Morino Y. Soda K. Biochemistry. 1981; 20: 4492-4500Crossref PubMed Scopus (112) Google Scholar). The liver also is the principal organ producing excretory metabolites of selenium to prevent the accumulation of toxic levels of the element. Thus, the liver is the portal through which selenium enters the body and the organ that maintains its homeostasis through excretion of the element. Secretion of Se-P into the plasma is another important function of the liver. The increase of liver selenium concentration in selenium-deficient mice with deletion of Se-P is therefore not surprising (Fig. 4B). Selenium entering the liver inSepp−/− mice is presumably diverted from export as Se-P to hepatic selenoproteins such as glutathione peroxidase. Se-P deletion did not interfere with the regulation of liver selenium content when levels above the dietary requirement for selenium were fed. Whether the liver selenium concentration was maintained by release of selenium in a form that could be taken up by other tissues or by production of excretory metabolites was not addressed in these experiments. Determination of this will be important in the elucidation of selenium homeostasis. The studies reported here address the selenium transport function of Se-P. They indicate that the testis and brain have mechanisms for acquiring selenium from plasma Se-P. The characterization of those mechanisms will require additional research, as will assessment of the putative oxidant defense role of Se-P. We are grateful to Dr. Kirk R. Thomas for help in generating the mutant mice as well as for very helpful discussions and to Dr. Marla J. Berry for suggesting the collaboration that led to this study.