STAF [Sec (selenocysteine) tRNA gene transcription activating factor] is a transcription activating factor for a number of RNA Pol III- and RNA Pol II-dependent genes including the Trsp [Sec tRNA gene], which in turn controls the expression of all selenoproteins. Here, the role of STAF in regulating expression of Sec tRNA and selenoproteins was examined. We generated transgenic mice expressing the Trsp transgene lacking the STAF-binding site and made these mice dependent on the transgene for survival by removing the wild-type Trsp. The level of Sec tRNA was unaffected or slightly elevated in heart and testis, but reduced ∼60% in liver and kidney, ∼70% in lung and spleen and ∼80% in brain and muscle compared with the corresponding organs in control mice. Moreover, the ratio of the two isoforms of Sec tRNA that differ by methylation at position 34 (Um34) was altered significantly, and the Um34-containing form was substantially reduced in all tissues examined. Selenoprotein expression in these animals was most affected in tissues in which the Sec tRNA levels were most severely reduced. Importantly, mice had a neurological phenotype strikingly similar to that of mice in which the selenoprotein P gene had been removed and their life span was substantially reduced. The results indicate that STAF influences selenoprotein expression by enhancing Trsp synthesis in an organ-specific manner and by controlling Sec tRNA modification in each tissue examined.
There are 24 known selenoprotein genes [i.e. genes coding for proteins containing the amino acid Sec (selenocysteine)] in rodents and 25 in humans . One of the key factors required in the expression of selenoproteins is Sec tRNA. Since the biosynthesis of Sec occurs on its tRNA, and Sec synthesis initiates with the aminoacylation of serine, this tRNA has been designated tRNA[Ser]Sec . The tRNA[Ser]Sec population consists of two isoforms that differ by a single methyl group located on the 2′-O-hydroxyribosyl moiety at position 34 (designated Um34) (reviewed in ). The modified base at position 34 in both isoforms is mcm5U (5′-methylcarboxylmethyluracil) and the nucleoside at position 34 in the Um34-containing isoform is mcm5Um (5′-methylcarboxylmethyluridine-2′-O-methylribose). Thus the two isoforms are designated mcm5U and mcm5Um. Synthesis of Um34 on mcm5U to form mcm5Um is the last step in the maturation of tRNA[Ser]Sec and its attachment is dependent on selenium status .
We have taken advantage of the fact that selenoprotein synthesis is dependent on the presence of tRNA[Ser]Sec and have thus generated several mouse models that have altered tRNA[Ser]Sec levels as a tool to modulate selenoprotein expression and elucidate the role of this specialized class of proteins in health and development ([4,5] and reviewed in ). For example, transgenic mice encoding wild-type or mutant tRNA[Ser]Sec transgenes were generated that resulted in an overexpression of the corresponding wild-type or mutant tRNA[Ser]Sec . Increased levels of wild-type Trsp (Sec tRNA gene) caused only minor changes in the selenoprotein population in specific tissues, suggesting that the Sec tRNA[Ser]Sec population in most mammalian cells is not limiting (reviewed in ). Transgenic mice encoding mutant Trsp transgenes, however, resulted in reduced expression of a subclass of selenoproteins that function primarily in stress-related phenomena [6–8]. One of these mutant transgenic mouse lines has been used to show that these stress-related selenoproteins have a role in preventing colon  and prostate  cancers.
The Trsp gene has also been conditionally deleted in specific tissues and organs using loxP-Cre technology , resulting in the complete loss of selenoprotein expression after recombination in targeted cells [5,11,12]. The conditional knockout of Trsp has been used to show that selenoproteins have a role in proper liver function , endothelial cell development and heart disease prevention , neuronal development and survival (E. K. Wirth, M. Conrad, B. S. Bharathi, J. Winterer, C. Wozny, B. A. Carlson, S. Roth, D. Schmitz, G. W. Bornkamm, M. Brielmeier, V. Coppola, L. Tessarollo, E. Pohl, L. Schomburg, J. Kohrle, D. L. Hatfield and U. Schweizer, unpublished work) and immune function .
Regulation of Trsp expression is unique among tRNAs. In contrast with other tRNA genes, Trsp transcription is governed by three upstream regulatory regions: the TATA box motif at −30, the PSE (proximal sequence element) located at ∼−70 [14–17] and a DSE (distal sequence element) located at ∼−200 in mice . The latter regulatory region is composed of an AE (activator element) containing an SPH motif and an octamer sequence . The DSE is required for optimal transcription of Trsp in vivo in Xenopus oocytes , although it does not function in the enhancement of Trsp transcription in vitro ( and reviewed in ). The DSE has been disrupted in mice by inserting a 3.2 kb fragment between this regulatory region and the PSE . Disruption of the DSE in this manner was embryonic lethal resulting from loss of Trsp transcription and severely decreased amounts of selenoprotein transcripts. A protein factor, designated the STAF (Sec tRNA gene transcription activating factor), binds to the AE. STAF was previously cloned and characterized in Xenopus  and mice [22–24]. This transcription factor was subsequently found to have roles in the expression of many other genes transcribed by RNA polymerases II and III [25–28].
In the present study, we took an alternative approach in investigating the role of STAF and the AE region in Trsp transcription. The role of STAF in tRNA[Ser]Sec transcription in an in vivo model was examined by generating mutant Trsp transgenic mice in which the STAF-binding site of Trsp, the AE region, was deleted. We then used the deleted AE allele to genetically complement Trsp-knockout mice [5,29]. The resulting AE− transgenic/Trsp-knockout mice expressed tRNA[Ser]Sec in substantially reduced amounts in liver, kidney, brain, lung, spleen and muscle, but in similar or slightly elevated levels in heart and testes. Interestingly, the level of the mcm5Um isoform was less than that of the mcm5U isoform in all tissues, and even in heart and testes wherein the Sec tRNA[Ser]Sec population was not reduced. Selenoprotein expression was down-regulated in tissues that were most affected by Sec tRNA[Ser]Sec reduction, and brain, muscle, lung and spleen appeared to be the most affected tissues. AE− animals manifested a neurological phenotype similar to that of mice in which the selenoprotein P gene (Sepp) had been targeted for removal [29–33]. These mice had a wide, waddling gait [31,33,34], tremor  and hyperexcitability with occasional seizures . The results of these studies are presented and discussed with regard to mechanism of regulation of selenoprotein expression by STAF and the two tRNA[Ser]Sec isoforms.
Materials and animals
75Se (specific radioactivity 1000 Ci/mmol) was obtained from the Research Reactor Facility, University of Missouri (Columbia, MO, U.S.A.), [α-32P]dCTP (specific radioactivity ∼6000 Ci/mmol) from PerkinElmer and [3H]serine (specific radioactivity 30 Ci/mmol), Ready-to-Go DNA Labelled Beads and Hybond Nylon N+ membranes from GE Healthcare. NuPAGE 10% polyacrylamide gels, 10% TBE (Tris/borate/EDTA; 1×TBE=45 mM Tris/borate and 1 mM EDTA)/urea gels, PVDF membranes, TRIzol® and SeeBlue Plus2-protein markers were purchased from Invitrogen. SuperSignal West Dura Extended Duration Substrate was obtained from Pierce, the GeneEditor In Vitro Site-Directed Mutagenesis System from Promega, antibodies against GPx1 (glutathione peroxidase 1) from Abcam, antibodies against GPx4 (glutathione peroxidase 4), SelR (selenoprotein R) and SelT (selenoprotein T) from our stock supplies and anti-rabbit-HRP (horseradish peroxidase)-conjugated secondary antibodies from Sigma. All other reagents were obtained commercially and were products of the highest grade available. Heterozygous Trsp-knockout mice (Trsp+/−) in a C57Bl/6 genetic background were from a previous study in our laboratory . Mice were maintained on standard rodent diet, NIH-31A, containing adequate amounts of selenium as described in  and the care of animals was in accordance with the National Institutes of Health Institutional Guidelines under the expert direction of Dr Kyle Stump (NCI, NIH, Bethesda, MD, U.S.A.).
Preparation of mutant transgene, generation of experimental mouse lines, breeding, genotyping, survival rates and pathology
A 2158 bp DNA fragment encoding Trsp  with a 14 bp deletion of the AE region (bases −213 to −226) was used to generate transgenic mice by zygote injection as described elsewhere . The 14 bp deletion was prepared using the GeneEditor In Vitro Site-directed Mutagenesis System according to the manufacturer's instructions. Three independent founder transgenic mice carrying the 2158 bp transgene with a deletion of the AE allele were established in strain FVB/N and designated TrsptAE−. DNA was extracted from mouse-tail clippings and the TrsptAE− copy number initially assessed in the three founder mice by Southern blotting of DNA that was digested with XhoI, probed with [32P]CTP-labelled probe and estimated to be one copy in one founder and five copies in the other two founders by procedures described previously [4,37]. The low-copy-number TrsptAE− founder and one of the higher number copy founders were bred to obtain the corresponding homozygous mice for further examination. The presence or absence of Trsp or TrsptAE− was determined in all subsequent studies by PCR and the primers used have been described in detail previously [6,7]. These primers were designated CKNO2 (forward primer) and RES1 and VP1 (reverse primers). CKNO2-RES1 yields a 980 bp PCR Trsp fragment and a 500 bp PCR Trsp− fragment and CKNO2-VP1 yields a 1058 bp PCR TrsptAE− fragment.
Survival rates of TrsptAE−/−/Trsp−/− mice carrying two or ten copies of TrsptAE−/− and control mice were assessed as described in the legend to Figure 1.
Survival rates of TrsptAE−/−/Trsp−/− and control mice
TrsptAE−/−/Trsp−/− homozygous mice (five males and three females encoding two copies of TrsptAE−/−) and control mice (two males and three females), 1–2 months old, were subjected to pathological analysis. Tissues were fixed in 10% formalin, embedded in paraffin, sectioned 5-μm-thick, stained with H/E (haematoxylin/eosin), coverslipped and examined using an Olympus BX41 microscope.
Isolation, aminoacylation, fractionation and quantification of tRNA
Total tRNA was isolated from mouse tissues (see Figure 2 for tissues examined) of TrsptAE−/−/Trsp−/− mice carrying two or ten copies of TrsptAE−/− and control mice, and the amounts of tRNASer were quantified after gel electrophoresis and hybridization with the appropriate 32P-labelled probe by Northern-blot analysis using a PhosphorImager (Molecular Dynamics) . Only TrsptAE−/− mice carrying two copies of the mutant transgene and control mice were examined in all subsequent studies. Total tRNA from tissues of TrsptAE−/−/Trsp−/− mice and control mice was aminoacylated with [3H]serine and 19 unlabelled amino acids in the presence of rabbit reticulocyte synthetases . The resulting aminoacylated tRNA was fractionated on an RPC-5 column , first in the absence and then in the presence of Mg2+ as given in [4,5], the amounts of tRNA[Ser]Sec relative to the total Ser tRNA population, and of the distributions of the two isoforms, mcm5U and mcm5Um, were determined as described in detail previously [4,5].
Northern-blot analysis of TrsptAE−/−/Trsp−/− and control mice
Labelling of selenoproteins and selenium levels
Mice were given intraperitoneal injections of 75Se (50 μCi/g) and killed 48 h after injection. Tissues and organs were excised, frozen in liquid nitrogen and stored at −80 °C until use. Tissues were homogenized, the protein extracts were electrophoresed on NuPAGE 10% polyacrylamide gels, along with molecular mass markers, gels were stained with Coomassie Brilliant Blue, dried and exposed to a PhosphorImager and band intensities were assessed as described in .
Protein extracts were prepared from liver, kidney, brain, testes, heart, lung, spleen and muscle from three TrsptAE−/−/Trsp−/− mice and three control mice, all 4–5 weeks old, were electrophoresed on NuPAGE 10% polyacrylamide gels, and the resulting fractionated proteins were transferred to PVDF membranes and immunoblotted with antibodies against GPx1 (1:1000 dilution), GPx4 (1:1000), SelR (1:1000) and SelT (1:800). Anti-rabbit-HRP-conjugated secondary antibody (1:20000) was used in all Western blots. After the incubation with the secondary antibody, membranes were washed with 0.1% TBS-T (20 mM Tris/HCl, pH 7.5, 150 mM NaCl and 0.1% Tween 20), incubated in SuperSignal West Dura Extended Duration Substrate and then exposed to an X-ray film. Band intensities were assessed by PhosphorImager analysis . Separate gels were used for each antibody, and blots were stripped and probed with β-tubulin (1:1000) to serve as a loading control.
Brains of 5-week-old TrsptAE−/−/Trsp−/− and control mice were immersion-fixed in 4% (w/v) paraformaldehyde/0.1 M sodium phosphate buffer, pH 7.4 (PB). After cryoprotection in 30% sucrose in PB, brains were frozen on dry ice and stored at −80 °C. Sections of 35 μm thickness were cut on a cryostat. Nissl staining was done as described in . Immunostaining on free-floating sections was performed with the following antibodies: rabbit anti-PV (parvalbumin) (1:5000; Swant, Bellinzona, Switzerland) and mouse anti-GFAP (glial fibrillary acidic protein; Sigma, Munich, Germany) (1:1000). Immunoreactivity was visualized with HRP using the Vectastain reagents (Vector, Burlingame, VT, U.S.A.) and the MOM kit (Vector) with diaminobenzidine as the substrate.
Generation of TrsptAE−/−/Trsp−/− mice and survival rates
To examine the role of STAF in Trsp transcription, we deleted the STAF activating (AE) region in a Trsp fragment and generated transgenic mice carrying two or ten copies of the mutant transgene (homozygous animals designated TrsptAE−/−). We then used these transgenic mice to genetically complement Trsp-knockout (Trsp−/−) mice with the mutant transgenes. The resulting mice were viable, and therefore, the transgenic AE-lacking Sec tRNA[Ser]Sec rescued the lethal phenotype of Trsp−/−.
Since mice encoding Trsp−/− were dependent on the mutant transgene for survival, this provided us with a model for examining the role of the AE region in tRNA[Ser]Sec expression in various tissues. The breeding scheme used in obtaining the TrsptAE−/−/Trsp−/− and control mice is shown in Table 1. The survival rates of high- and low-copy-number mutant and control transgenic mice were compared (see Figure 1). Most TrsptAE−/−/Trsp−/− mice died prematurely (Figure 1). By 46 and 48 days, half of the TrsptAE−/−/Trsp−/− mice carrying two and ten copies of mutant transgene respectively died. There were no differences observed between survival rates of males and females. The high death rate was characteristic of young mice, whereas those who survived to adulthood, whether they carried two or ten copies of the mutant transgene, had much better chances of long-term survival.
|1st breeding||TrsptAE−/−* × Trsp+/−†|
|F1 (2nd breeding)||TrsptAE−/Trsp+/− × TrsptAE−/Trsp+/−|
|F2 (3rd breeding)‡||TrsptAE−/−/Trsp+/− × TrsptAE−/−/Trsp+/−||Trsp+/+ × Trsp+/+|
|1st breeding||TrsptAE−/−* × Trsp+/−†|
|F1 (2nd breeding)||TrsptAE−/Trsp+/− × TrsptAE−/Trsp+/−|
|F2 (3rd breeding)‡||TrsptAE−/−/Trsp+/− × TrsptAE−/−/Trsp+/−||Trsp+/+ × Trsp+/+|
*TrsptAE−/− transgenic mice carrying either two or ten copies of the mutant transgene wherein superscript t denotes that Trsp is a transgene and superscript tAE−/− denotes that both transgene alleles encode deleted AE.
†Trsp+/− mice (strain C57BL6) were obtained as described in .
‡Only those offspring that were used for further breeding are shown.
§Only those offspring that were used experimentally are shown.
Sec tRNA[Ser]Sec expression and isoform distribution in TrsptAE−/−/Trsp−/− and control mice
We examined the expression and distribution of tRNA[Ser]Sec and its isoforms, mcm5U and mcm5Um, in TrsptAE−/−/Trsp−/− mice. Total tRNA was isolated from liver, kidney, brain, heart, testes, lung, spleen and muscle of TrsptAE−/−/Trsp−/− and control mice and the expression of tRNA[Ser]Sec was examined by Northern-blot analysis in control and TrsptAE−/−/Trsp−/− mice carrying two or ten copies of the mutant transgene (Figure 2). The analysis was repeated in triplicate with the control and low-copy-number transgenic mice. The levels of tRNA[Ser]Sec from tissues of control mice were averaged and assigned a value of 1.0 (legend to Figure 2). The levels of tRNA[Ser]Sec from tissues of TrsptAE−/−/Trsp−/− mice carrying two copies of transgene relative to the wild-type control are shown in the upper panel of Figure 2. With the exception of heart and testes, which had similar levels of tRNA[Ser]Sec in control and mutant transgenic mice, the levels of tRNA[Ser]Sec were substantially reduced in all other tissues of mutant transgenic mice compared with the control wherein liver, brain, lung spleen and muscle manifested less than 50% expression. In addition, tRNA[Ser]Sec levels were examined in thymus of mice carrying two copies of mutant transgene and found to be approx. 50% of the wild-type level. Although the Northern-blot analysis was only carried out with tRNA[Ser]Sec from tissues of a single mouse carrying the higher number of mutant transgenes, the pattern of tRNA[Ser]Sec reduction was similar in the two mouse lines in that the reduction of tRNA[Ser]Sec occurred in all tissues except heart and testes, and muscle showed the lowest level relative to the control.
The above observations on survival rates (Figure 1) and tRNA[Ser]Sec expression (Figure 2) suggested that mice carrying either the high- or low-copy-number mutant transgenes, wherein the mice were dependent on the transgene for survival, produced similar effects on mice. Thus only the low-copy-number TrsptAE−/−/Trsp−/− mice were further examined.
tRNA[Ser]Sec expression in TrsptAE−/−/Trsp−/− control mice was also examined by RPC-5 chromatography, and the levels of the total tRNA[Ser]Sec population relative to the total seryl-tRNASer population were determined (Table 2). The amount of the tRNA[Ser]Sec population appeared to be slightly elevated in heart and was approx. 1.4 times higher in testes of TrsptAE−/−/Trsp−/− animals compared with controls (Table 2, column 3). However, tRNA[Ser]Sec levels were substantially lower in liver (64% reduction), kidney (56% reduction), brain (79% reduction), lung (69% reduction), spleen (67% reduction) and muscle (83% reduction) of TrsptAE−/−/Trsp−/− animals.
|% (Total)‡||% of control mouse§|
|% (Total)‡||% of control mouse§|
*Total tRNA was isolated from the tissues listed (column 1) and fractionated and then the amounts of total tRNA[Ser]Sec (column 3) and the percentage of tRNA[Ser]SecmcmU and tRNA[Ser]SecmcmUm isoforms (columns 4 and 5) determined as described in the Experimental section.
†Percentage of tRNA[Ser]Sec population within the total seryl-tRNA population.
‡Percentage of tRNA[Ser]SecmcmU and tRNA[Ser]SecmcmUm isoforms within the total seryl-tRNA population.
§Percentage of tRNA[Ser]SecmcmU and percentage of tRNA[Ser]SecmcmUm were determined by dividing the percentage of the TrsptAE−/−/Trsp−/− isoform by the percentage of the respective control isoform.
The distributions of the mcm5U and mcm5Um isoforms within the tRNA[Ser]Sec population were assessed by RPC chromatography in liver, kidney, brain, heart, lung, testes, muscle and spleen (Table 2). The percentage of mcm5U and mcm5Um within the total tRNA[Ser]Sec population is shown in columns 4 and 5 respectively. The distributions were dramatically altered in each tissue from TrsptAE−/−/Trsp−/− animals compared with those of control animals in that the mcm5Um isoform was reduced substantially relative to the mcm5U isoform. The percentage change of both isoforms in each of these tissues in TrsptAE−/−/Trsp−/− compared with control mice is shown in columns 6 and 7 of Table 2. These data show the reduced amount of mcm5Um in each tissue and either the increase (heart and testes) or decrease (liver, kidney, brain, muscle and spleen) of mcm5U. The most pronounced reduction in both isoforms occurred in brain, lung, muscle and spleen, whereas large losses of mcm5U occurred in all tissues except heart and testes. The lowest levels of mcm5Um expression occurred in brain and muscle.
75Se-labelling in TrsptAE−/−/Trsp−/− and control mice
Since the level of tRNA[Ser]Sec is not limiting in tissues of normal mice (reviewed in ), but the levels of this tRNA were reduced substantially in most of the tissues examined in the present study, we analysed selenoprotein expression by 75Se-labelling of TrsptAE−/−/Trsp−/− and control mice (Figure 3A). Coomassie Brilliant Blue-stained SDS/PAGE gels revealed no visible differences in total protein levels in the corresponding tissues of TrsptAE−/−/Trsp−/− and control animals (results not shown). In the PhosphorImager analyses that visualize 75Se-labelled proteins, those selenoproteins that were identified previously [6,42] are indicated in the Figure along with the molecular mass markers on the liver and plasma panels (Figure 3). Although the labelling was weakest in brain, the overall selenoprotein labelling pattern appeared to be highly affected in this tissue and possibly in lung and spleen of TrsptAE−/−/Trsp−/− mice. GPx1 in liver appeared to be slightly reduced relative to other selenoproteins in tissues from mutant mice. Selenoprotein labelling in muscle was not examined as selenoproteins were very poorly labelled in this tissue in earlier studies . The intensity of the various bands relative to that found in the corresponding control proteins was quantified by PhosphorImager analysis (Figure 3B). Values obtained for labelling of the selenoproteins analysed were assigned a value of 1.0 in each control tissue. The levels of GPx1, GPx4 and TR1 appeared to be lowest in spleen, lung and brain and less affected in the other tissues. GPx3 was slightly reduced in plasma and SelP was somewhat more reduced in this tissue.
Metabolic 75Se-labelling of selenoproteins in TrsptAE−/−/Trsp−/− and control mice
Analyses of selenoprotein expression by Western blotting
Metabolic labelling of selenoproteins may be influenced by protein turnover, differential delivery of the isotope to specific organs and the level of available selenium in respective tissues. We therefore placed greater emphasis on analysing selenoproteins by Western blotting to assess changes in their levels in TrsptAE−/−/Trsp−/− mice. Several selenoproteins were selected that were readily detectable by immunoblotting and are affected by the methylation status of tRNA[Ser]Sec [6,7,11] for further analysis (Figure 4). Western blots were carried out in triplicate from three individual TrsptAE−/−/Trsp−/− and control mice and a representative blot from one of the affected and control animals is shown in Figure 4(A). However, since the Western blots were done in triplicate, we quantified the results from the tissues of the three TrsptAE−/−/Trsp−/− and control mice and did a statistical analysis on the three runs (Figure 4B). GPx1 levels were reduced statistically in all tissues examined except kidney compared with controls, while GPx4 was reduced statistically only in brain, lung and spleen. SelR levels were down in liver, brain, testes, heart and muscle and the variations were too extreme in lung and spleen to assess whether SelR levels were down statistically in these tissues compared with controls. SelT levels were reduced only significantly in lung, spleen and muscle compared with the controls. Possible reasons for the variations in levels of these selenoproteins in the different tissues are further considered in the Discussion section.
Western-blot analysis of selenoproteins affected by reduction of the mcm5Um isoform
Influence of the STAF-binding region on selenium status of mouse tissues
Selenium amounts were determined in liver, kidney, brain, heart, testes, lung, spleen, plasma and muscle of TrsptAE−/−/Trsp−/− and control animals (Table 3). Although selenium levels were numerically lower in each tissue from TrsptAE−/−/Trsp−/− animals compared with controls, statistical analysis showed that Se contents were significantly lower in liver, brain, lung and muscle in TrsptAE−/−/Trsp−/− mice, but not in the other tissues.
|Selenium levels (ng/g)*|
|Selenium levels (ng/g)*|
*Selenium levels were measured as described in the Experimental section. Values represent the mean±S.E.M. for three different mice of each genotype with the exception of testes for which the experiment was carried out in duplicate.
†This symbol indicates that the differences in selenium levels were significant (P<0.05) between control and affected tissues as determined by the Student's t test (unpaired, two-sided) using GraphPAD Prism 4.0 as described in the Experimental section.
Phenotypic characteristics and pathology of TrsptAE−/−/Trsp−/− mice
Phenotypically, the TrsptAE−/−/Trsp−/− mice were smaller in size than their normal littermates, which was readily noticeable after approx. 3 weeks of age. Pups were weighed at 3–5 weeks after birth and the average weights of TrsptAE−/−/Trsp−/− mice during this time frame were 71% of those of control mice.
Several pathological abnormalities were observed in TrsptAE−/−/Trsp−/− mice (Figure 5). Six of eight TrsptAE−/−/Trsp−/− mice had smaller spleens with little haematopoietic activity and smaller follicles, which also lacked a marginal zone compared with control mice (Figure 5A). Five of eight TrsptAE−/−/Trsp−/− mice exhibited acute cerebral neuronal necrosis, while none of the five control mice had this lesion (Figure 5B). Less frequent findings in TrsptAE−/−/Trsp−/− mice included an observation that three of eight mice had either single cell necrosis/apoptosis or larger areas of coagulative hepatic necrosis (Figure 5C). One control had mild hepatocellular single cell necrosis. Three of eight AE mutant mice had gastric ulcers and/or gastritis. In addition, mineralization of skeletal muscle, tongue, heart and adipose tissue was present in six of eight mutant mice, whereas only one control mouse had mild mineral deposition in the heart.
Pathological analysis of spleen, brain and liver in TrsptAE−/−/Trsp−/− and control mice
Reduced survival, growth retardation, mineralization of muscle, small spleens and liver abnormalities were all previously found in Sepp−/− mice ([30,31,33,34,43]; and U. Schweizer, unpublished work). Moreover, the TrsptAE−/−/Trsp−/− mice appeared hyperactive, with tremor, periodically exhibited seizures (which could be induced by handling) and a wide waddling gait, and occasional backward movement, which are also phenotypes observed in Sepp−/− mice ([30–33]; U. Schweizer, unpublished work).
Immunohistological analysis of the brain
In order to investigate the neurological phenotype of TrsptAE−/−/Trsp−/− mice in more detail, we performed further histological studies of their brains. As judged by Cresyl Violet staining, the brains of these mice appeared structurally normal (Figure 6A). However, if neurons die by apoptosis and at a low rate, their remnants will be cleared from the brain parenchyma within days. Therefore we probed brain sections for increased expression of GFAP, which is a marker of reactive astrocytes and an established indicator of neurodegenerative processes. In TrsptAE−/−/Trsp−/− mice, GFAP immunoreactivity was massively increased throughout the brain, in particular in the cerebral cortex grey matter where GFAP is normally very low and confined to astrocytes surrounding blood vessels (Figure 6B). We speculated that the hyperexcitable phenotype of TrsptAE−/−/Trsp−/− mice may be related to an impaired inhibitory system. In the cerebral cortex and hippocampus, GABA (γ-aminobutyric acid) is the predominant inhibitory neurotransmitter. Most of the GABAergic neurons within the cerebral cortex express the small calcium-binding protein PV, which can be used as a marker for these neurons . In TrsptAE−/−/Trsp−/− mice, the number of PV-immunopositive neurons was greatly reduced both in the cerebral cortex and in the hippocampus (Figure 6C). Thus the neurological phenotype of TrsptAE−/−/Trsp−/− mutant mice may be partially caused by a disturbance of the GABAergic inhibitory system. On the other hand, it has been shown that epilepsy leads to the loss of GABAergic interneurons . Thus we cannot be certain at present whether the loss of PV cells is the cause or consequence of the observed phenotype. Nevertheless, increased and widespread astrogliosis as shown by GFAP immunostaining argues for a mechanism of diffuse neurodegeneration throughout the brain.
Histological analysis of brain tissues from TrsptAE−/−/Trsp−/− and control mice
In the present study, the role of STAF in Trsp transcription was investigated in vivo by genetically complementing Trsp−/− mice with TrsptAE−/− alleles, which rescued these knockout mice from lethality. The loss of STAF responsiveness of transgenic Trsp expression in two mutant mouse lines carrying two or ten copies of TrsptAE−/− resulted in tissue-specific Trsp transcription changes wherein testes and heart appeared to have similar or slightly enriched levels of Sec tRNA[Ser]Sec and liver, kidney, brain, lung, muscle and spleen had reduced levels. The low- and high-copy-number TrsptAE−/−/Trsp−/− mouse lines had reduced but similar rates of survival compared with control animals, and the levels of tRNA[Ser]Sec expression in the various tissues examined by Northern-blot analysis were similar. In light of these observations and the fact that the TrsptAE−/−/Trsp−/− mouse carrying two copies of the mutant transgene would more naturally mimic a mutation in the AE region precluding STAF attachment, further studies were only pursued with the low-copy-number transgenic mouse. In addition, the possibility that the effect of the deleted AE region may be influenced by the position of insertion of the mutant fragment in the genome rather than due solely to the loss of this regulatory site seems unlikely, since another founder mouse encoding a higher copy number of mutant transgenes manifested a similar survival rate and synthesized similar levels of tRNA[Ser]Sec. Surprisingly, mice carrying two or ten copies of TrsptAE− generated similar levels of tRNA[Ser]Sec transgene product (see Figure 2), whereas previous studies had shown that the amount of tRNA[Ser]Sec synthesized from such genes is proportional to the number of transgenes [4,37]. The synthesis of similar amounts of tRNA[Ser]Sec from single and multiple TrsptAE− transgenes may be related to the reduced expression of the mcm5Um isoform as selenium-deficient cells and tissues synthesize less of this isoform and less tRNA[Ser]Sec overall (reviewed in ). Furthermore, Secp43 is known to have a role in Um34 synthesis (and possibly a role in shuttling SecS from the nucleus to the cytoplasm) . Whether Secp43 may be involved in the overall level of tRNA[Ser]Sec expression from multiple (trans)genes remains to be established, but its role in Um34 synthesis and the reduced expression of the mcm5Um isoform observed in the present study in AE− mice would seem to be more than just coincidental.
The above observations raise a question of whether the effects seen by the removal of the STAF-binding region is related to selenium status and may be only indirectly related to the loss of STAF binding. This seems unlikely as the levels of selenium were reduced significantly in liver, brain, lung and muscle, but not in kidney, heart, testes, spleen or plasma, whereas the levels of mcm5Um were reduced dramatically in all tissues examined regardless of whether the overall levels of tRNA[Ser]Sec and selenium were, or were not, reduced. Clearly, the overall levels of tRNA[Ser]Sec appeared to be increased in heart and testes, while the levels of selenium were not significantly changed in affected compared with control mice.
We have used both Northern blotting and RPC-5 chromatography to assess tRNA[Ser]Sec levels in mammalian cells (e.g. see  and the present study), while other investigators have used Northern blotting exclusively [20,29]. Although the two methods appear to yield similar results, a major advantage of RPC chromatography is that it permits an assessment of the distributions of the two isoforms, whereas Northern blotting can be used on much smaller amounts of tissue.
Kelly et al.  generated a mouse with a disrupted DSE wherein a 3.2 kb fragment was inserted between this region and the PSE regulatory region. Sec tRNA[Ser]Sec transcription was severely impaired in the mouse carrying the insertion on both alleles and embryo death occurred at day 7.5. The heterozygous mouse survived normally and, despite reduced levels of Sec tRNA in kidney and liver, this animal expressed selenoprotein mRNAs and the major mammalian selenoprotein, GPx1, in normal amounts. It is not surprising that the disrupted DSE generated by Kelly et al.  had a more pronounced effect on reducing Trsp transcription than the TrsptAE−/−/Trsp−/− mouse, since the entire DSE was disrupted in their study, while our present study focused on the STAF-binding region by deleting only AE.
It is not clear why TrsptAE−/−/Trsp−/− animals were smaller and their survival rates were reduced compared with their normal littermates. Most likely, reduced selenoprotein expression in those tissues most affected, e.g. brain and muscle, played a role in these abnormalities. The phenotypic similarity between Sepp−/− mice and TrsptAE−/−/Trsp−/− mice was, however, striking. Both mouse models are characterized by growth retardation, small spleen, tissue calcification, liver and brain defects [31–35]. The exact mechanism why reduced selenoprotein expression in a particular cell type eventually leads to a growth defect in the mice is not yet understood. One possibility is that brain appears to be among the most affected organs in both models. Therefore we investigated brain at the molecular level and in more detail. We did not find evidence for developmental abnormalities within the brains of TrsptAE−/−/Trsp−/− mice. Rather, the alterations that we observed pointed towards degenerative processes as evidenced by widespread astrogliosis and hypereosinophilic neurons in several brain regions. The most specific finding was the apparent reduction of PV expressing interneurons in cerebral cortex and hippocampus. This finding may be related to the observed neurological phenotype, in particular to the hyperexcitability, the abnormal gait and impaired righting reflex. However, a more detailed study would be required to elucidate the mechanism by which reduced selenoprotein expression causes neurodegeneration. Moreover, a dedicated neuropathological study would be needed to determine whether TrsptAE−/−/Trsp−/− mice also suffer from axonal degeneration .
One of the more intriguing findings in the present study was the observed alteration in mcm5U and mcm5Um distribution in TrsptAE−/−/Trsp−/− mice wherein the latter isoform was more dramatically reduced in the tissues examined. In liver, kidney, brain, heart and lung of control animals, the ratio of mcm5Um/mcm5U was 1–1.5. However, upon reduction of Trsp expression in TrsptAE−/−/Trsp−/− mice, the ratio fell to ≤1. In heart and testes, mcm5U was increased in the mutant, but mcm5Um decreased even at similar or slightly higher levels of the tRNA[Ser]Sec population. In both the latter tissues of the TrsptAE−/−/Trsp−/− mice, selenoprotein expression was partially disrupted. Thus it appears as if the bulk levels of tRNA[Ser]Sec isoforms are not governing selenoprotein expression alone. Possibly, additional cellular compartments exist in these organs that respond differently to disruption of the activation element in Trsp. These compartments could express different sets of selenoproteins and thus a selenoprotein may appear reduced in spite of increased tRNA[Ser]Sec levels. In the case of testis, there is evidence for such a possibility. In Sepp−/− mice, testicular GPx4 expression is undetectable, while GPx1 is almost normal . Usually, GPx4 is less dependent on Se supply than GPx1. Thus GPx4 may be expressed in an SePP (selenoprotein P)-dependent compartment, whereas GPx1 is not. It is otherwise difficult to explain how deletion of the AE region and/or the failure of STAF to properly recognize this region may be involved in the methylation of the mcm5U isoform unless Secp43 may be involved as noted above.
GPx1 was not detected in muscle by Western blotting. In liver, the levels of GPx1 and SelR, but not SelT and GPx4, were reduced significantly in TrsptAE−/−/Trsp−/− mice compared with control mice as assessed by Western blotting. In other tissues, GPx1 was also statistically reduced in brain, heart, testes, lung, spleen and muscle, while SelR appeared to be reduced in these same tissues, but SelT only in lung, spleen and muscle and GPx4 in brain, lung and spleen in affected compared with control mice. mcm5Um amounts were reduced more substantially than mcm5U amounts in all tissues examined. These results provide interesting insights into the complex nature of expression of two subclasses of selenoproteins at the level of translation that is carried out by two separate isoforms.
We have also rescued Trsp−/− mice with a mutant Trsp transgene  wherein the tRNA gene product was incapable of generating the mcm5Um isoform due to a mutation at position 37 . Several selenoproteins serving stress-related functions, such as GPx1, GPx3, SelR, SelT and SelW, were poorly expressed in the rescued mouse, but other selenoproteins serving largely housekeeping functions, such as TR1, TR3 and GPx4, were expressed in normal or slightly reduced amounts . Mice dependent on the mutant transgene for survival were phenotypically normal with the exception that males had defective sperm and produced fewer offspring than their wild-type littermates, while rescued females appeared to have some aberration with regard to fertility and/or carrying pregnancies to term. Although rescued Trsp−/− mice in the present study also expressed mcm5Um poorly, they additionally expressed mcm5U poorly in several tissues that apparently resulted in several housekeeping selenoproteins being synthesized in reduced but sufficient levels to permit the resulting mice to survive at least to an early age, as shown in Figure 1. The significance of the transgenic mouse model described in the present study over those used previously in which the tRNA[Ser]Sec levels were also altered resulting in differential expression of the subclasses of selenoproteins (reviewed in ) is that the present model down-regulates both tRNA[Ser]Sec isoforms in specific organs and tissues. The present study also emphasizes the more important role of housekeeping selenoproteins in the day-to-day function with regard to overall health and life span than stress-related selenoproteins.
distal sequence element
glial fibrillary acidic protein
glutathione peroxidase 1
proximal sequence element
Sec tRNA gene
Sec tRNA gene transcription activating factor
This work was supported by the Intramural Research Program of the Center for Cancer Research, NCI, NIH, to D. L. H., NIH grants to V. N. G. and Deutsche Forschungsgemeinschaft grants [DFG SFB665, Scho849/1-2] to U.S.