|Home | About | Journals | Submit | Contact Us | Français|
Oxygen radicals regulate many physiological processes, such as signaling, proliferation, and apoptosis, and thus play a pivotal role in pathophysiology and disease development. There are at least two thioredoxin reductase/thioredoxin/peroxiredoxin systems participating in the cellular defense against oxygen radicals. At present, relatively little is known about the contribution of individual enzymes to the redox metabolism in different cell types. To begin to address this question, we generated and characterized mice lacking functional mitochondrial thioredoxin reductase (TrxR2). Ubiquitous Cre-mediated inactivation of TrxR2 is associated with embryonic death at embryonic day 13. TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos are smaller and severely anemic and show increased apoptosis in the liver. The size of hematopoietic colonies cultured ex vivo is dramatically reduced. TrxR2-deficient embryonic fibroblasts are highly sensitive to endogenous oxygen radicals when glutathione synthesis is inhibited. Besides the defect in hematopoiesis, the ventricular heart wall of TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos is thinned and proliferation of cardiomyocytes is decreased. Cardiac tissue-restricted ablation of TrxR2 results in fatal dilated cardiomyopathy, a condition reminiscent of that in Keshan disease and Friedreich's ataxia. We conclude that TrxR2 plays a pivotal role in both hematopoiesis and heart function.
Reactive oxygen species (ROS)—generated mainly as a by-product of the respiratory chain or by oxidases—are implicated in the pathogenesis and pathophysiology of a variety of human diseases such as cancer, cardiovascular, and degenerative disorders. A variety of cellular antioxidant systems control the balance of free intra- and extracellular oxygen radicals. Previous efforts have addressed the physiological role of superoxide dismutases, catalases, and glutathione (GSH) peroxidases in vivo, but the role of the thioredoxin/thioredoxin reductase/peroxiredoxin system in ROS removal has only recently attracted attention.
Thioredoxins are small redox-active proteins with an essential function in DNA metabolism and repair, transcription, and cell-cell communication (1). Acting through peroxiredoxins, they also efficiently protect cells from oxidative damage (27). Cytosolic (Trx1) and mitochondrial (Trx2) thioredoxins are required for proliferation and protection from apoptosis during early embryogenesis (26). Moreover, in chicken B cells, Trx2 is critically involved in the regulation of mitochondrion-dependent apoptosis (37). More recently, heart-specific overexpression of dominant-negative Trx1 was shown to be associated with increased oxidative stress and cardiac hypertrophy in mice (39).
Trx activities are governed by thioredoxin reductases (TrxRs) that, in turn, use NADPH/H+ as the reducing agent (23). TrxRs are members of the pyridine nucleotide-disulfide oxidoreductase family, form homodimers, and possess two interacting redox-active centers. The C-terminal redox center contains a catalytically important selenocysteine (SeCys) (9, 17, 41). In mammals, three TrxRs are known—one cytosolic (TrxR1) (8), one mitochondrial (TrxR2) (7), and one testis-specific GSH-TrxR (33)— each encoded by individual genes. In contrast to their well-studied biochemical properties, little is known about the individual contribution of the different TrxRs in living organisms or about possible redundancies among the different redox systems. To dissect the complex network of redox metabolism in vivo, we have established mice deficient for mitochondrial TrxR2. To overcome possible embryonic lethality, as observed in Trx1- and Trx2-null mice, and to investigate TrxR2 function in specific organs and at defined time points, we used a conditional ablation strategy. Ubiquitous deletion of TrxR2 results in embryonic lethality at embryonic day 13 (E13). TrxR2−/− embryos are severely anemic and exhibit a marked thinning of the ventricular heart wall. To further dissect the contribution of the two major defects in the overall phenotype, we generated mice with cardiac cell-specific inactivation of the TrxR2 gene. These mice developed fatal dilated cardiomyopathy and died shortly after birth. Analysis of fibroblasts isolated from TrxR2−/− embryos revealed a critical role of TrxR2 in the removal of toxic ROS species and the maintenance of mitochondrial integrity.
Mouse expressed sequence tag clone IMAGp998N131993 (RZPD, Berlin, Germany), covering exons 10 to 18 of TrxR2, was used to screen a 129/SV mouse lambda-phage library (Stratagene, Inc., La Jolla, Calif.). Five clones were isolated and cloned into pBluescript SK(+) (Stratagene). The insert of one clone containing the 3′ region of the TrxR2 gene was sequenced and used for the generation of the targeting constructs. In brief, a NheI/XmaI fragment representing the 5′ arm for homologous recombination was subcloned in pBluescript SK(+), and a loxP element, including an EcoRI site, was inserted in the unique FseI site. The 3′ arm was inserted as an XmaI/XmaI fragment in pPNT4, a vector containing an frt-flanked neomycin phosphotransferase (neo) gene and one loxP site (4). The targeting construct pPNT11 was completed by inserting the 5′ arm including the second loxP site. The conditional lacZ knockin construct is comprised of the same homologous arms but contains additional elements as depicted in Fig. Fig.1E.1E. Gene targeting of both constructs in E14 embryonic stem (ES) cells was carried out as described previously (14). ES cell clones with successful homologous recombination arose at a frequency of 5 to 8% when tested with both external probes pMC67 and pMC68. Additional integration sites were excluded by Southern blot analysis with a neo gene probe. Flp-mediated removal of the neo gene by transient overexpression of Flp recombinase in targeted ES cell clones occurred at a frequency of 6% (29). Backcross of chimeric animals to C57BL/6 mice, produced upon microinjection of ES cell clones into C57BL/6 blastocysts, gave rise to germ line transmission of the loxP flanked (floxed) TrxR2 allele (11). These mice were crossed to congenic C57BL/6 Cre deleter mice, resulting in deletion of the floxed allele (31). Mice with either the floxed or the deleted TrxR2 allele were backcrossed on a C57BL/6 background. To facilitate genotyping of mice, primer pairs were designed for detection of wild-type (WT), floxed, or deleted alleles (TrxR2flox1/2, CAGGTCACTAGGCTGTAGAGTTTGC/ATGTCCCAGTGTACTTATGATGAATC; TrxR2Del1/2, TGCTTCCAGGCCCAGTGCTCTGACTGG/CAGGCTCCTGTAGGCCCATTAAGGTGC). The Cre-specific primer pair was as follows: GATGCAACGAGTGATGAGGTTCGC/ACCCTGATCCTGGCAATTTCGGC (CreA/B).
TrxR2 was deleted in cardiomyocytes by using the MLC2a-Cre transgenic mouse line (38). TrxR2+/− mice were first crossed with MLC2a-Cre mice and then TrxR2+/−,Tg[MLC2a-Cre] and TrxR2fl/fl mice were mated to obtain TrxR2fl/−,Tg[MLC2a-Cre] mice.
Mice were kept under standard conditions with food and water ad libitum (Altromin GmbH, Lage, Germany). All animal experiments were performed in compliance with the German animal welfare law and have been approved by the institutional committee on animal experimentation and the government of Upper Bavaria.
Total RNA was isolated from homogenized embryonic tissue by using peqGOLDTriFast (peqLab, Erlangen, Germany). DNase-treated total RNA (DNase I, RNase I free, catalog no. 776785; Roche, Mannheim, Germany) was reverse transcribed by using a reverse transcription system (Promega Corp., Madison, Wis.). Primer pairs specific for Aldolase (Aldolase1/2, AGCTGTCTGACATCGCTCACCG/CACATACTGGCAGCGCTTCAAG)andTrxR2(TrxR2E6/10, CAGCTTTGTGGATGAGCACACAGTTCG/GATCCTCCCAAGTGACCTGCAGCTGG; and TrxR2E15/18, TTCACGGTGGCGGATAGGGATGC/TGCCCAGGCCATCATCATCTGACG) were used to detect the transcripts.
Detection of apoptotic cells was performed with the ApopTag kit (S7100; Serologicals Corp., Norcross, Ga.) on PFA-fixed (4% [wt/vol] paraformaldehyde [PFA] in phosphate-buffered saline), paraffin-embedded sections according to the manufacturer's instructions. Digoxigenin-deoxynucleoside triphosphate-labeled cells were incubated with an anti-digoxigenin peroxidase conjugate and visualized with 3,3′-diaminobenzidine (DAB, SK-4100; Vector Laboratories, Inc., Burlingame, Calif.).
Upon standard pretreatment and blocking with M.O.M. blocking reagent (M.O.M Immunodetection Kit, PK-2200; Vector Laboratories), cryosections were incubated overnight with an anti-β-galactosidase monoclonal antibody (1:50; MAB1802; Chemicon International, Inc., Temecula, Calif.). Visualization of avidin-biotin complexes was performed with 3-amino-9-ethylcarbazole (SK-4200; Vector Laboratories).
PFA-fixed paraffin-embedded sections were incubated for 1 h at 37°C with anti-PCNA monoclonal antibody (1:200; PC10; Dako, Glostrup, Denmark) and visualized with DAB.
Cryosections were incubated overnight at 4°C with a rat anti-mouse CD45 antibody (1:20; catalog no. 550539; BD Pharmingen, San Jose, Calif.). In the case of rat anti-K8 or -K18 antibodies, initial incubation was performed for 90 min at 37°C (1:10; Troma1 and Troma2 [kindly provided by Solange Magre, CNRS/Université Pierre et Marie Curie, Paris, France, and Rolf Kemler, Max Planck Institute, Freiburg, Germany]). Sections were subsequently incubated with a secondary biotinylated antibody for 1 h at room temperature. After avidin-biotin complex reaction was performed (ABC alkaline phosphatase staining kit, AK-5000; Vector Laboratories), bound antibodies were visualized with a substrate for alkaline phosphatase (Vector Blue, SK-5300; Vector Laboratories). Sections were washed and postfixed in cold ethanol-acetic acid (2:1) at −20°C, the endogenous peroxidase activity was blocked with 3% (vol/vol) H2O2, and the sections were subjected to the standard ISEL procedure.
Hearts were isolated and immediately fixed in 2.5% glutaraldehyde-2% PFA in 0.1 M sodium cacodylate buffer (pH 7.2) for 24 h, rinsed three times in the same buffer, and postfixed in 1% osmium tetroxide-1.5% potassium ferrocyanide in potassium cacodylate buffer for 2 h. Samples were washed three times, dehydrated, and embedded in Propylenoxyd and Polyembed 812 (Plano, Wetzlar, Germany). Ultrathin sections (60 nm) were stained with uranyl acetate and lead acetate and then examined on a Zeiss 902 electron microscope at 80 kV.
5-Bromodeoxyuridine (BrdU; catalog no. 100171; ICN Biomedicals, Aurora, Ohio) was injected intraperitoneally in pregnant mice at E13.0 at a dose of 100 mg/kg of body weight. After 2 h, isolated embryos were fixed in PFA and embedded in paraffin. Incorporated BrdU was visualized by immunohistochemistry with an anti-BrdU monoclonal antibody (catalog no. 1170376; Roche Diagnostics Corp.) and a secondary biotinylated antibody (ABC peroxidase staining).
Embryos were isolated from pregnant mice at E12.5, the body trunk was dissected away from other structures and then treated with trypsin, treated with DNase, minced with a syringe, and cultivated under standard conditions in Dulbecco modified Eagle medium (Invitrogen GmbH, Karlsruhe, Germany) with 10% fetal calf serum (FCS), glutamine, and penicillin-streptomycin. Buthionine sulfoximine (BSO; Sigma-Aldrich, Taufkirchen, Germany) was used at a 50 μM final concentration.
Cell nuclei were stained with 1 μM Hoechst 33342 (Sigma-Aldrich), and the mitochondria were stained with 0.5 μM MitoTracker Green FM (Molecular Probes, Inc., Oreg.). RedoxSensor Red CC-1 was used according to the manufacturer's protocol (2) (Molecular Probes). Stained cells were assessed with a Zeiss Axiovert 200 M microscope (Carl Zeiss, Jena, Germany) and Openlab software (version 3.0; Improvision, Coventry, United Kingdom).
CFU assays were performed as described previously (30). In brief, livers from E12.5 embryos were prepared, minced with a syringe in Iscove modified Dulbecco medium (including 20% FCS), centrifuged, and resuspended for hemolysis in 1 volume of medium without FCS and 1 volume of 2× hemolysis buffer (1.5 M NH4Cl, 0.1 M KHCO3, and 1 mM EDTA in 500 ml [pH 7.4] with KOH). The hemolysis step was repeated twice. Cells were plated at equal cell numbers on 35-mm dishes containing methylcellulose, 10% bovine serum albumin (catalog no. 652237; Roche), 20% FCS (pretested), 1% penicillin-streptomycin, 1% glutamine, 50 μM β-mercaptoethanol, and cytokines (100 U of mouse interleukin-3 [IL-3] conditioned medium, 2 U of erythropoietin, 10 ng of stem cell factor, 10 ng of FLIT ligand, 10 ng of thrombopoietin, 10 ng of granulocyte-macrophage colony-stimulating factor, and 10 ng of IL-6/ml). After 1 week of incubation, colonies with >50 cells were counted and subdivided into granulocytes, macrophages, mixed colonies (more than three different cell types), granulocytes and macrophages, and burst-forming units that were additionally visualized by benzidine staining (0.1 volume of benzidine, 0.1 volume of H2O2, and 0.5 volume of H2O).
The TrxR2 gene is composed of 18 exons spanning a region of 57 kb (22). The last four exons harboring the C-terminal catalytic center were flanked by two loxP sites (Fig. (Fig.1A).1A). The SeCys codon UGA is encoded by exon 17. The selenocysteine insertion sequence (SECIS) element, essential for cotranslational SeCys incorporation at the UGA codon, and the transcription termination signal are both encoded by exon 18. After homologous recombination of the targeting construct in ES cells (Fig. 1A and B), Flp-mediated removal of the frt-flanked neomycin phosphotransferase gene (neo) in ES cells results in a TrxR2 allele with two loxP and one frt inserted sites. Cre-mediated excision of the loxP-flanked exons 15 to 18 in mice leads to removal of the C-terminally located enzymatic domain of TrxR2 (Fig. (Fig.1C).1C). TrxR2+/− mice, obtained after breeding with Cre deleter mice, are viable and fertile, show no obvious phenotype, and have a normal life span compared to WT littermates. Since no specific antibody for the detection of TrxR2 was available, semiquantitative RT-PCR with embryonic mRNA was used to verify the knockout (KO) of the targeted region (Fig. (Fig.1D).1D). Two primer pairs, one covering the central region (TrxR2E6-10) and one covering the deleted region (TrxR2E15-18) of the TrxR2 gene, were used to study mRNA levels in embryos. Aldolase served as a control. The absence of the last four exons in TrxR2−/− embryos was confirmed with primer pair TrxR2E15-18. Reduction of mRNA levels was noticeable in TrxR2+/− embryos compared to WT siblings. Faint RT-PCR products were observed in TrxR2−/− embryos with primer pair TrxR2E6-10, indicating very low levels of truncated TrxR2 message.
To study the TrxR2 expression pattern, we designed a conditional lacZ knockin approach into the TrxR2 locus (Fig. (Fig.1E).1E). A translational and transcriptional stop/internal ribosomal entry site-β-galactosidase(lacZ) cassette was placed in the 3′-nontranscribed region of the TrxR2 gene. This brings LacZ under the control of the endogenous TrxR2 promoter upon Cre-mediated removal of exons 15 to 18, including the transcriptional stop signal (derived from the mouse phosphoglycerine kinase gene). Immunohistochemistry of TrxR2+/−(lacZ-k.i.) embryos with an anti-lacZ antibody revealed strong expression in the embryonic heart, especially in the myocardium and atrial walls, and to a lower extent in the embryonic liver (Fig. 1F to I). These findings are consistent with mRNA expression data obtained from adult tissues (21).
Intercross of TrxR2+/− mice did not yield TrxR2−/− offspring (of 137 viable offspring, 33.6% were genotypically WT and 66.4% hemizygous). Genotyping of littermates at different days of gestation revealed that TrxR2−/− embryos died between E12.5 and E13.5. TrxR2−/− embryos were reduced in size and had an anemic appearance (Fig. (Fig.2A).2A). The blood vessels of the yolk sac and the embryo proper were markedly less supplied with blood (Fig. 2B to E), and the bloodstream velocity was slower in living TrxR2−/− embryos at E13.0 (data not shown). The anemic appearance of the TrxR2−/− embryos and the comparable high expression of TrxR2 in embryonic heart and liver suggested a crucial role of TrxR2 in the respective organs.
Histological examination of TrxR2−/− embryos at E13.0 revealed dysplasia of cardiac tissue. The ventricular myocardium, the ventricular septum, and the trabeculae were thinner compared to WT siblings (Fig. 3A and B). No abnormalities of the valves and outflow tract were noticeable in TrxR2−/− hearts. To investigate whether reduced proliferation or increased cell death gave rise to the observed phenotype, we performed BrdU incorporation studies, as well as PCNA and ISEL staining. BrdU incorporation and PCNA staining revealed decreased proliferation in the myocardium of TrxR2−/− embryos (Fig. 3D and F). In contrast to TrxR2−/− mice (Fig. (Fig.3F),3F), myocardial PCNA staining in TrxR2+/+ littermates was more pronounced and predominantly nuclear, indicating that most cells were proliferating (Fig. (Fig.3E).3E). No differences in the proliferation of the endocardial cushions and, unexpectedly, no increase in apoptosis throughout the heart were observed (data not shown).
We compared expression levels between WT and TrxR2−/− embryos for a number of genes important for cardiovascular development and function (including erbB2, erbB4, erbB3, heregulin, GATA-4, GATA-6, Ang1, tie-2, VEGF, VEGFR2, ephrin-B2, Eph-B4, EGFR, TGFβ, TGFβRII, ecNOS, and iNOS), but we did not detect any differences (data not shown). Similarly, immunofluorescence analysis with antibodies against CD31, CD34, VEGFR2, laminin γ, collagen IV, erbB3, vWF, and VE-cadherin did not reveal any abnormal patterns in TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos (data not shown). Moreover, we did not observe any defects in vessel formation of placenta and yolk sacs of TrxR2−/− embryos (data not shown). We conclude that reduced proliferation, rather than loss, of cardiomyocytes is the cause of the thinning of the ventricular myocard in TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos.
We examined whether the anemic phenotype in TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos is due to reduced heart function or whether it is caused by perturbed hematopoiesis in the liver, the main site of fetal hematopoiesis after E11 (6). Noticeable was the pleiomorphic and spongiform appearance of liver tissue in TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos, suggesting perturbed proliferation and/or increased apoptosis of hepatocytes and/or hematopoietic cells (Fig. 3G and H). BrdU and PCNA staining of embryonic liver sections did not reveal any major differences (data not shown), whereas ISEL staining showed augmented apoptosis in the liver of TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos (Fig. 3I and K). Double staining, including ISEL, blood cell marker CD45, and hepatocyte markers keratins 8 and 18, revealed that mainly blood cells were affected by the loss of TrxR2 (Fig. 3L to N). CFU assays with liver cells isolated from E12.5 embryos were carried out to investigate whether hematopoietic differentiation is perturbed. In CFU assays, the composition of blood cell colonies was unaltered (Fig. (Fig.4A).4A). However, TrxR2TrxR2−/−minus;/TrxR2−/−minus; colonies were markedly smaller (Fig. 4B to E), indicating that reduced blood cell formation is the cause of the severe anemic phenotype of TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos.
The mitochondrial TrxR2/Trx2/peroxiredoxin-3 system is similar to the cytosolic TrxR1/Trx1/peroxiredoxin-1 system (25) and is thought to be involved in the elimination of oxygen radicals generated as a by-product of oxidative phosphorylation in the respiratory chain (Fig. (Fig.4G)4G) (15). To examine whether removal of ROS and consequently proliferation and/or apoptosis are affected in TrxR2TrxR2−/−minus;/TrxR2−/−minus; cells, MEFs were used as a cellular model system. Proliferation of TrxR2TrxR2−/−minus;/TrxR2−/−minus; MEFs was significantly slower compared to WT counterparts (Fig. (Fig.4F).4F). Embryonic fibroblasts were stained with the RedoxSensor Red CC-1 dye for intracellular ROS accumulation. RedoxSensor Red CC-1 stains intracellular ROS and, depending on the level of ROS, it localizes to different cellular compartments (2, 36). No major difference was observed in WT versus KO MEFs, indicating that other antioxidant systems substitute for the missing TrxR2 function (data not shown). However, when the de novo synthesis of GSH was inhibited with 50 μM BSO (Fig. (Fig.4G),4G), a competitive inhibitor of the γ-glutamyl-cysteinyl-synthetase that catalyzes the rate-limiting step in GSH synthesis, only WT but not KO MEFs were able to survive and proliferate (Fig. (Fig.4H).4H). To test whether increased levels of ROS in TrxR2TrxR2−/−minus;/TrxR2−/−minus; MEFs are responsible for the failure of cell survival when GSH synthesis is inhibited, we stained MEFs with the RedoxSensor Red CC-1 dye. In the presence of BSO, RedoxSensor Red CC-1 was predominantly localized in lysosomes of WT cells and in mitochondria of TrxR2TrxR2−/−minus;/TrxR2−/−minus; cells (Fig. 4I to L). Treatment of cells with the antioxidant N-acetylcysteine (NAC) rescued TrxR2TrxR2−/−minus;/TrxR2−/−minus; MEFs from BSO-induced apoptosis (Fig. 4H and M). These data indicate that under limiting GSH conditions TrxR2 is indispensable for sustaining cell survival in fibroblasts.
Targeted disruption of genes such as retinoid X receptor α (5) and the cytokine receptor gp130 (40) has suggested that these genes participate in heart and blood development. However, cardiomyocyte-specific inactivation of the respective genes did not support the conclusions drawn from the complete KO experiments (3, 10). To rule out the possibility that the failure in heart development is the consequence of a primary defect of TrxR2 deficiency in hematopoietic cells, TrxR2 was deleted specifically in cardiomyocytes by using the MLC2a-Cre transgenic mouse line (38). TrxR2+/− mice were first crossed with MLC2a-Cre mice and, subsequently, TrxR2+/−,Tg[MLC2a-Cre] and TrxR2fl/fl mice were mated to obtain TrxR2fl/−,Tg[MLC2a-Cre] mice. Mice that had both copies of the TrxR2 gene deleted in a cardiac tissue-specific manner showed clinical features of congestive heart failure, including generalized edema, liver congestion, globular heart shape, and atrial dilatation, and died within several hours after birth (data not shown). Histological examination revealed dilated heart cavities and thinning of the ventricular myocardium (Fig. 5B and C). We observed, at higher magnifications, severe distortion in the morphology of myocardial cells, such as pycnotic nuclei, cytoplasmic vacuolization, and reduced cross-striation (Fig. 5D to G). Electron microscopic analyses of neonatal hearts showed severe swelling and destruction of mitochondrial cristae in cardiomyocytes of cardiac tissue-restricted TrxR2-deficient mice (Fig. 5H to L). Vessel formation of yolk sac and placenta, as well as development of coronary arteries, as assessed by histological and immunohistochemical examination (vWF staining), was not affected by heart-specific inactivation of TrxR2 (data not shown). This demonstrates that TrxR2 is not only essential for normal embryonic hematopoiesis but also plays a crucial role in the maintenance of mitochondrial integrity in cardiomyocytes and is therefore indispensable for proper heart function of newborn mice.
Impairment or ablation of antioxidant enzymes leading to the accumulation of toxic levels of ROS might initiate or promote a variety of human diseases, e.g., neurodegeneration, cancer development, arthritis, and atherosclerosis. Many studies have addressed the role of GSH and GSH peroxidases as central regulators in the maintenance of the cellular redox balance. To assess the importance of the thioredoxin/thioredoxin reductase system in this process, we generated and analyzed mice with a nonfunctional TrxR2, a key enzyme of the mitochondrial redox system. At present, very little is known about the physiological role of TrxR2. TrxR2 belongs to the family of selenoproteins that are characterized by one or several catalytically indispensable SeCys residues (16). Since cotranslational SeCys incorporation at the UGA codon is very inefficient (35), overexpression studies are strongly limited in vitro and in vivo. For this reason, gene inactivation methods in mice offer a promising strategy.
Anticipating that loss of TrxR2 might be associated with embryonic death, we used the cre/loxP technology for the generation of TrxR2 KO mice. Cre-mediated excision of the last four exons leads to the removal of the final 100 amino acids including the C terminally located redox-center consisting of Cys-SeCys-Gly. Mutational and biochemical studies have previously shown that SeCys is essential for the enzymatic function of TrxRs (9, 17, 41). Also, this approach deemed necessary because the 5′ region of TrxR2 overlaps with the first exon of the catechol-o-methyltransferase gene, and there is also alternative first exon usage in the TrxR2 gene (22, 34).
We found that during development TrxR2 is mainly expressed in the embryonic heart and liver, reflecting the adult situation (22). This expression profile associates TrxR2 function with organs characterized by high metabolic activity and further corroborates a crucial role for TrxR2 in the control of harmful intracellular ROS.
Ubiquitous deletion of the TrxR2 gene leads to embryonic lethality at E13. The phenotypic characteristics of TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos reflected the expression pattern of TrxR2 in normal development. TrxR2-null embryos were highly anemic and smaller compared to WT littermates. Histological examination of the heart revealed that the ventricular walls and the trabeculae were thinner in KO mice than in WT and TrxR2+/− mice. The number of ISEL-positive cells was increased in the liver but not in the heart. In CFU assays established from fetal liver at E13, the size of all types of hematopoietic colonies was dramatically reduced, whereas hematopoietic differentiation was not affected by TrxR2 deficiency.
Outgrowth of TrxR2TrxR2−/−minus;/TrxR2−/−minus; MEFs in tissue culture was also impaired but not to the same extent as observed for hematopoietic cells. TrxR2TrxR2−/−minus;/TrxR2−/−minus; fibroblasts could be propagated in vitro; they were, however, much more sensitive to oxidative stress imposed to the cells by GSH depletion than TrxR2+/+ and TrxR2+/− cells. These data indicate that under standard cell culture conditions hematopoietic cells are more dependent on TrxR2 than fibroblasts. Several mechanisms, either alone or in combination, may account for the proposed hierarchy in TrxR2 dependence. Cells may differ in their ability to cope with oxygen radicals due to different degrees of pathway redundancy and/or overall antioxidant capacity. Alternatively, oxygen radical production of different cell types may vary greatly due to differences in their cellular metabolism. Finally, different cell types may differ widely in their intrinsic susceptibility versus resistance to oxygen radical-induced apoptosis due to different expression patterns of pro- and antiapoptotic proteins. The hematopoietic system is well known for its selective sensitivity to ionizing irradiation. Our finding that TrxR2 KO mice die of anemia at day 13 of embryonic development may thus emphasize the particular susceptibility of hematopoietic cells to oxygen radical-induced apoptosis. This interpretation is in line with the fact that deletion of Trx2 in the chicken B-cell line DT40 leads to oxygen-radical induced apoptosis mediated by cytochrome c release and caspase-3 activation (37).
The TrxR2 inactivation phenotype is less severe than the one observed in the KO mice of Trx2, the main substrate of TrxR2. Trx2−/− mice die at E10.5 due to massive apoptosis (26). This suggests that other enzymes, such as, for example, TrxR1 or the GSH-dependent redox system, can partially substitute for the TrxR2 deficiency.
Besides the defect in hematopoiesis, TrxR2 KO mice exhibited morphological changes in the heart. To discriminate whether TrxR2 plays an intrinsic role in heart development and function or whether the defect in heart morphology is a consequence of the defect in hematopoiesis, we bypassed the hematopoietic phenotype and established cardiac tissue-restricted TrxR2-deficient mice. These mice died shortly after birth from dilated cardiomyopathy and congestive heart failure. High-magnification and ultrastructural examination revealed severe distortion in the morphology of cardiomyocytes, i.e., pycnotic nuclei, partial loss of cross-striation, swelling of the mitochondria, and destruction of mitochondrial cristae. In contrast to the severe changes in morphology, no signs of apoptosis were detected in the myocardium by ISEL staining. Difficulties in detecting apoptotic cells in morphologically severely perturbed heart tissue have also been noted by others. Narula et al. found that cardiomyocytes of patients with idiopathic dilated or ischemic cardiomyopathy exhibit severe mitochondrial swelling and accumulation of cytochrome c in the cytosol in the absence of apoptotic changes (24). Likewise, the apoptotic index as defined by the TUNEL (terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling) assay was found to be very low in models of chronic heart failure (12, 18). We suggest that cardiac tissue-restricted TrxR2 KO mice die of congestive heart failure due to mitochondrial dysfunction of cardiomyocytes before DNA degradation and cell death of cardiomyocytes can be detected to a significant extent.
The fact that mice with cardiac tissue-specific deletion of TrxR2 die shortly after birth has two implications. First, it indicates that the defect in heart morphology observed in TrxR2TrxR2−/−minus;/TrxR2−/−minus; mice at E13 is not only a consequence of the hematopoietic defect. It demonstrates for the first time that loss of one member of the Trx/TrxR system is indispensable for normal heart development and function. Second, lethality in complete TrxR2TrxR2−/−minus;/TrxR2−/−minus; embryos is delayed from E13 until birth in mice with a cardiac tissue-specific deletion of TrxR2. This suggests that the heart-specific defect observed in TrxR2TrxR2−/−minus;/TrxR2−/−minus; mice at E13 is not limiting on its own for survival during embryonic development. However, TrxR2 deficiency in cardiomyocytes becomes fatal when an entirely self-sustaining circulatory system has to be maintained by the heart. It is conceivable that in mice with ubiquitous deletion of TrxR2, anemia and the heart-specific defect enhance each other and, thus, may lead to an earlier lethal compound phenotype compared to the respective tissue specific deletions. To clarify this issue, it would be necessary to generate a hematopoiesis-specific KO of TrxR2.
Using transgenic approaches, Yamamoto et al. demonstrated that Trx1 is involved in cardiac myocyte growth of mice (39). Heart-specific overexpression of a dominant-negative form of Trx1 led to increased oxidative stress in cardiomyocytes associated with cardiac hypertrophy, whereas overexpression of WT Trx1 reduced the extent of ROS-induced hypertrophy in response to pressure overload. Since the dominant-negative Trx1 may impair Trx2 activity as well, a putative contribution of Trx2 in protection of cardiomyocytes in this model seems likely. The perinatal lethality in our heart-specific TrxR2 KO model precludes at present a detailed study of heart function parameters. Crossing the conditional TrxR2 KO mice with the inducible heart-specific MERCreMER mice might generate a valuable tool for studying the impact of TrxR2 deficiency on adult heart physiology (32).
A first indication that selenium is essential for heart function came from patients suffering from Keshan disease, a selenium deficiency disease endemic in China. Severely selenium-depleted humans show dilated congestive cardiomyopathy resembling the histopathologic phenotype of Friedreich's cardiomyopathy (28) and the phenotype observed in the TrxR2TrxR2−/−minus;/TrxR2−/−minus; mice. It is plausible that TrxR2 is the most likely candidate whose function is impaired in Keshan disease due to selenium deficiency. Whether functional impairment of other selenoproteins also contributes to the pathophysiology of this disease remains an open question. It is also noteworthy that TrxR2 is localized to chromosome 22q11, a region deleted in the human haploinsufficiency velo-cardio-facial/DiGeorge Syndrome (VCS/DGS) (13, 19, 20). Hallmarks of VCS/DGS have not been noticed in hemizygous/homozygous or in cardiac tissue-restricted TrxR2 KO mice (data not shown). Further studies will address the question whether loss of one TrxR2 allele may contribute to the compound VCS/DGS phenotype.
We are grateful to C. Neumüller and J. Plitzko for help with the heart ultrastructure analysis, A. Geishauser and M. Semisch for excellent technical work, and the personnel of the animal facility and blastocyst injection unit at GSF for excellent assistance and support.
This study was supported by the DFG-Priority Programme SPP1087, by Fonds der Chemischen Industrie, and a fellowship from the Humboldt-Stiftung to S.G.M.; by the DFG Priority Programme 1069 “Angiogenesis” (HA 2983/1-2) and the German Human Genome Project (DHGP 9907) to A.K.H.; and by the BMBF to W.W.