Homocysteine Down-regulates Cellular Glutathione Peroxidase (GPx1) by Decreasing Translation*

Hyperhomocysteinemia contributes to vascular dysfunction and an increase in the risk of cardiovascular disease. An elevated level of homocysteine in vivo and in cell culture systems results in a decrease in the activity of cellular glutathione peroxidase (GPx1), an intracellular antioxidant enzyme that reduces hydrogen peroxide and lipid peroxides. In this study, we show that homocysteine interferes with GPx1 protein expression without affecting transcript levels. Expression of the selenocysteine (SEC)-containing GPx1 protein requires special translational cofactors to “read-through” a UGA-stop codon that specifies SEC incorporation at the active site of the enzyme. These factors include a selenocysteine incorporation sequence (SECIS) in the 3′-untranslated region of the GPx1 mRNA and cofactors involved in the biosynthesis and translational insertion of SEC. To monitor SEC incorporation, we used a reporter gene system that has a UGA codon within the protein-coding region of the luciferase mRNA. Addition of either the GPx1 or GPx3 SECIS element in the 3′-untranslated region of the luciferase gene stimulated read-through by 6–11-fold in selenium-replete cells; absence of selenium prevented translation. To alter cellular homocysteine production, we used methionine in the presence of aminopterin, a folate antagonist, co-administered with hypoxanthine and thymidine (HAT/Met). This treatment increased homocysteine levels in the media by 30% (p < 0.01) and decreased GPx1 enzyme activity by 45% (p = 0.0028). HAT/Met treatment decreased selenium-mediated read-through significantly (p < 0.001) in luciferase constructs containing the GPx1 or GPx3 SECIS element; most importantly, the suppression of selenium-dependent read-through was similar whether an SV40 promoter or the GPx1 promoter was used to drive transcription of the SECIS-containing constructs. Furthermore, HAT/Met had no effect on steady-state GPx1 mRNA levels but decreased GPx1 protein levels, suggesting that this effect is not transcriptionally mediated. These data support the conclusion that homocysteine decreases GPx1 activity by altering the translational mechanism essential for the synthesis of this selenocysteine-containing protein.

Hyperhomocysteinemia is a known risk factor for cardiovascular disease. Normal plasma homocysteine levels range from 5 to 15 M (1, 2); however, clinical studies suggest that high normal levels may contribute to cardiovascular disease risk. Elevated levels of homocysteine contribute to endothelial dysfunction, an early marker of vascular injury in atherogenesis (3). In animal models of hyperhomocysteinemia, such as the heterozygous cystathionine ␤-synthase-deficient (CBSϩ/Ϫ) mouse, a modest doubling of homocysteine levels is sufficient to impair endothelium-dependent vasodilator responses in mesenteric arteries that have normal endothelium-independent vasodilator responses (4). CBSϩ/Ϫ mice have other detectable signs of oxidative stress, including a deficiency in acetylcholine-stimulated cGMP accumulation in isolated aorta, an increase in immunodetectable aortic 3-nitrotyrosine, and an increase in the levels of plasma F 2 -isoprostanes (4 -6). These findings suggest a lack of bioavailable nitric oxide in these mice. We have proposed that homocysteine contributes to the accumulation of reactive oxygen species and the subsequent inactivation of nitric oxide by decreasing the activity of cellular glutathione peroxidase (GPx1), 1 a major intracellular antioxidant enzyme. In vivo, modest hyperhomocysteinemia decreases GPx1 activity (5,7). Similarly, in cultured endothelial cells, micromolar concentrations of homocysteine are sufficient to decrease the activity of this important antioxidant enzyme (6,8).
The molecular mechanism by which homocysteine decreases GPx1 activity is as yet unclear. GPx1 mRNA has been shown to be decreased in cells treated with supraphysiological (5 mM) concentrations of homocysteine (8,9), yet previous studies indicate that GPx1 activity is diminished by only physiological or pathological concentrations of homocysteine (6,8). GPx1 is one of several selenocysteine-containing proteins (10), the expression of which is known to be regulated post-transcriptionally by selenium, which increases GPx1 mRNA stability (11). Insertion of selenocysteine in the protein involves read-through of a UGA-stop codon that specifies selenocysteine incorporation. Many cofactors are involved in the translational incorporation of the amino acid selenocysteine, including cis-acting signals, such as a selenocysteine incorporation sequence (SECIS) in the 3Ј-untranslated region (UTR) of the GPx1 mRNA (12). Other cofactors for selenocysteine incorporation in eukaryotes include the following: selenium; enzymes involved in the synthesis of SEC; a tRNA specific for the selenocysteine amino acid (tRNA sec ) that has an anticodon to recognize the UGA; an elongation factor for selenocysteine incorporation (eF sec ); and a SECIS-binding protein (SBP2) (12)(13)(14)(15).
Because of the specialized mechanism of translation required for selenocysteine incorporation, we analyzed the potential interference of homocysteine with selenocysteine incorporation and GPx1 activity in cultured cells. Our data suggest that homocysteine interferes with GPx1 translation, resulting in a decrease in cellular production of this antioxidant enzyme.

EXPERIMENTAL PROCEDURES
Cell Culture-BAEC and COS7 cells were passaged in Dulbecco's modified Eagle's medium (DMEM) with high glucose (Invitrogen) and 10% fetal calf serum (FCS). For selenium depletion, COS7 cells were grown in DMEM with 0.1% FCS for 5 days prior to treatment. To increase levels of homocysteine in culture, we added the following: HAT/Met, which includes aminopterin (800 nM), a dihydrofolate reductase inhibitor, to decrease cellular folate and homocysteine remethylation; hypoxanthine (200 M) and thymidine (32 M), to compensate for adverse effects of aminopterin on nucleotide biosynthesis; and excess (1 mM) methionine (Met), to increase homocysteine synthesis.
Homocysteine Measurements-Cell culture media were collected after 48 h of HAT/Met treatment and briefly centrifuged to remove any cellular debris. Total homocysteine (free homocysteine plus homocysteine derived from homocystine or mixed disulfides) levels were measured by electrochemical detection of reduced samples that were deproteinated prior to injection into a C18 column on a BAS LC200A. Samples and standards were mixed with an internal standard to control for sample processing and recovery. Known concentrations of homocystine were reduced and used to generate a standard curve. As little as 2 pmol of homocysteine could be reproducibly detected in this system.
GPx1 Enzyme Activity-Cell pellets were collected and sonicated in 50 mM Tris, pH 7.5, 5 mM EDTA, and 1 mM dithiothreitol. After the removal of cellular debris by centrifugation, samples were stored at Ϫ80°C. Enzyme activity was determined by an indirect assay that links GPx1-mediated oxidation of glutathione with the recycled reduction of GSSG to GSH by glutathione reductase using NADPH as a reductant. Enzyme activity is calculated from the change of absorbance at 340 nm over time, a measure of NADPH oxidation.
Reverse Transcriptase (RT)-PCR-Total RNA was prepared from cells cultured in 100-mm tissue culture dishes by using the RNeasy mini kit (Qiagen) with the addition of an optional DNase I step to remove residual DNA. For the RT reactions, 1 g of RNA was used to generate cDNA with oligo(dT) primers and the Moloney murine leukemia virus polymerase according to the RT-for-PCR kit of Clontech. The 20 l of cDNA mixture was diluted with water to a final volume of 100 l, aliquoted, and stored frozen at Ϫ80°C. An 833-bp band specific for GPx1 cDNA was detected by PCR with the forward primer LP7F (CTC-CCCTTACAGTGCTTGTTCG) and the reverse primer LP10B (CTT-TATAGTGGGAACTCGCC). PCR conditions are as follows: an initial denaturation step at 94°C for 1 min, followed by 30 PCR cycles of 94°C for 45 s, 63°C for 45 s, and 72°C for 2 min, with a final extension at 72°C for 7 min. A 938-bp G3PDH fragment was generated in a separate PCR using commercially available human G3PDH amplimers (Clontech). DNA fragments were separated on 3% 3:1 Nusieve-agarose gels (BioWhittaker) and detected by ethidium bromide fluorescence.
Luciferase Constructs-The pGL2 UGA and pGL2 UAA vectors were a kind gift from Dr. Donna Driscoll (16). These vectors are modified from the Promega pGL2 control vector. Briefly, the pGL2 UGA vector contains a TGA codon substituted for the glycine 38 codon of the luciferase protein-coding region and has a PvuII site in the 3Ј-untranslated region (UTR) of the luciferase gene. The pGL2 UGA has an SV40 promoter and enhancer to drive transcription. To add the GPx1 SECIS element to the pGL2 UGA vector, a human GPx1 cDNA construct (5) was digested with AvrII and HindIII to isolate a 215-bp fragment that overlaps the portion of the GPx1 3Ј-UTR containing the SECIS element. After using Klenow polymerase in a reaction to fill in the restriction endonuclease overhangs, the fragment was ligated to the PvuII enzyme-digested pGL2 UGA vector. Clones were sequenced to identify those with a single copy of the insert in either the sense (UGA/hSE-CIS/S) or antisense (UGA/hSECIS/AS) orientation. A similar strategy was used to add the GPx1 SECIS element to the pGL2 UAA vector, which has a TAA codon substituted for the glycine 38 codon, creating UAA/hSECIS/S and UAA/hSECIS/AS. The starting vector and the UGA/hSECIS/S and UGA/hSECIS/AS were further modified by replacing the SV40 promoter with a GPx1 promoter fragment that includes ϳ540 bp upstream of the transcription initiation site and 46 bp of the 5Ј-untranslated region of the GPx1 gene. This modification was accomplished by replacing the KpnI/HindIII fragment in each of the SV40 promoter vectors with a KpnI/HindIII fragment containing the GPx1 promoter. The resulting constructs were denoted GPx1/UGA, GPx1/ hSECIS/S, and GPx1/hSECIS/AS.
A fragment of the plasma GPx (GPx3) 3Ј-UTR was amplified with the forward primer CCCAGCTGTGCCTACAGGTATGCGTGATTG and the reverse primer CCCAGCTGAAGCCAGTGGACCAGTGAGGGGTGA. The resulting 620-bp fragment has one internal PvuII site;: after PvuII digestion, the 500-bp fragment, which overlapped the SECIS element, was cloned into the PvuII site of the pGL2 UGA vector. The sense (GPx3/SECIS 500/S) and antisense (GPx3/SECIS 500/AS) constructs were used for transfections.
Transfections and Read-through Assays-COS7 cells were seeded at 50,000 cells/well in 12-well tissue culture plates in DMEM with 0.1% fetal calf serum (FCS), in the presence or absence of 10 ng/ml sodium selenite. After 5 days of growth, each well was transfected with 2 g of luciferase construct, 20 ng of cytomegalovirus Renilla construct, and 5 l of Superfect (Qiagen, Valencia, CA) for 2 h. After removal of the transfection media, cells were treated with the low serum media plus or minus additives. Cells were harvested 48 h later, and luciferase and Renilla luciferase measurements were made using dual-luciferase reporter assay system (Promega) and a Turner TD-20e luminometer. All luciferase activities were corrected for Renilla luciferase activities before any other comparisons were made.
Western Blot-Lysates were prepared as for the GPx1 assay with the addition of protease inhibitors (Calbiochem). Protein concentrations were measured by the bicinchoninic assay using reagents from Pierce. Samples (50 g) were separated by electrophoresis on 12% denaturing polyacrylamide gels. Proteins were transferred to nitrocellulose membranes (Hybond, Amersham Biosciences) that were subsequently incubated overnight at 4°C with a mouse anti-GPx1 monoclonal antibody (MBL, Woburn, MA) and a polyclonal rabbit anti-␤-actin antibody (Sigma). Membranes were then incubated with the appropriate horseradish peroxidase-conjugated secondary antibodies (Sigma or Cell Signaling, Beverly, MA). Enhanced chemiluminescence (Amersham Biosciences) was used to detect immune complexes.

GPx1
Enzyme Activity-GPx1 activity in bovine aortic endothelial cells (BAEC) can be reduced by as little as 25 M homocysteine (Fig. 1A). BAEC were treated with HAT, to inhibit dihydrofolate reductase and homocysteine remethylation, and Met, to increase intracellular levels of homocysteine. Both folate restriction (17) and methionine supplementation have been shown to increase production of homocysteine in a variety of cells (18). HAT/Met treatment resulted in a decrease in GPx1 activity similar to that achieved with 100 M homocysteine. The GPx1 activity of cells treated with 25 M homocysteine was further suppressed by the addition of adenosine (50 M), a cofactor required for S-adenosylation of homocysteine and methionine, in the presence of 10 M erythro-9-(2-hydroxy-3-nonyl)adenine, an adenosine deaminase inhibitor. COS7 cells could be grown in media supplemented with only 0.1% FCS, allowing for depletion of cellular selenium, which is present in normal fetal calf serum. After 7 days of selenium depletion, GPx1 activity is decreased by ϳ75% compared with the activity in cells cultured in 0.1% FCS media supplemented with 10 ng/ml sodium selenite. The activity in cells cultured in 0.1% FCS media supplemented with 10 ng/ml sodium selenite is nearly the same as the activity of cells grown in 10% FCS media without additional selenium supplementation (data not shown). Cells supplemented with 1 ng/ml sodium selenite have no apparent increase in GPx1 activity over those grown with no added selenium, suggesting that there is a threshold level of selenium supplementation necessary to increase the production of GPx1. HAT/Met treatment decreased GPx1 activity by 42% in COS7 cells cultured in selenium-replete cells (Fig. 1B). Under these conditions, we found that HAT/Met treatment of COS7 cells increases the cellular release of homocysteine into the media by 30% (Fig. 2).
GPx1 mRNA-To determine whether the effects of low levels of homocysteine or HAT/Met treatment affected GPx1 mRNA levels, we used RT-PCR of RNA isolated from COS7 cells that were treated for 48 h with HAT/Met or 100 M homocysteine (Fig. 3). As has been shown previously (11), selenium increases the levels of GPx1 mRNA by ϳ2-fold. Treatment with HAT/Met or 100 M homocysteine had no effect on the steady-state levels of GPx1 mRNA.
Selenium-dependent UGA Read-through-GPx1 expression relies on a unique translational mechanism to incorporate selenocysteine at a UGA codon; thus, we next determined whether treatments that increase homocysteine production can affect read-through of the UGA codon during mRNA translation. To accomplish this end, we utilized a luciferase reporter gene with a substitution of a UGA codon for a glycine codon (GGA) in the protein-coding region of the luciferase mRNA (pGL2 UGA). Theoretically, in the absence of selenium and other necessary cofactors for selenocysteine incorporation, read-through of the UGA to produce a full-length functional luciferase protein would occur infrequently. Therefore, to promote read-through of the UGA as a site for selenocysteine incorporation rather than as a site for termination of translation, the human GPx1 SECIS element was inserted in the 3Ј-UTR of the luciferase gene in the sense (hSECIS/S) orientation (Fig. 4A). As a negative control for random effects of the SECIS element on luciferase expression, constructs were made with the SECIS element in the antisense orientation (hSECIS/AS). To assess any effect of the GPx1 promoter on luciferase expression, the SV40 promoter was replaced by the GPx1 promoter, resulting in the vectors GPx1/ UGA, GPx1/hSECIS/S, and GPx1/hSECIS/AS (Fig. 4A). First, the activity of the pGL2 UGA was compared with that of a control vector without the UGA mutation in the luciferase gene. The activity of the control construct was over 100,000-fold higher than that of the pGL2 UGA (p Ͻ 0.0001), consistent with the role of the UGA mutation as a stop codon. The expression of the control vector was unaffected by the addition of selenium. In constructs with the UGA mutation, production of luciferase only occurs in the presence of selenium when a SECIS element is in the sense orientation, such as in the GPx1/hSECIS/S or UGA/hSECIS/S (Fig. 4B). By contrast, the pGL2 UAA constructs had no detectable luciferase activity in the presence or absence of selenium, and the SECIS element had no effect on luciferase activity. In the absence of selenium, the activity of GPx1/hSECIS/S is identical to the activity of the GPx1 construct that lacks the SECIS element (GPx1/UGA) or that of the construct that has the SECIS element in the antisense orientation (GPx1/hSECIS/AS), whether or not selenium is present. Similarly, the activity of the UGA/hSECIS/S in the absence of selenium is identical to the activity of the pGL2 UGA vector or the antisense construct (UGA/hSECIS/AS) with or without selenium. The levels of luciferase activity from cells transfected with the GPx1 promoter constructs is almost 2-fold higher than that of cells transfected with the corresponding SV40 promoter constructs, suggesting that the GPx1 promoter has higher transcriptional activity in COS7 than the SV40 promoter. The selenium-induced increase in UGA readthrough, however, is similar (nearly 6-fold) with either the UGA/hSECIS/S or the GPx1/hSECIS/S. Overall, these data show that expression involves selenocysteine incorporation as read-through production of luciferase requires both selenium and the SECIS element and that the read-through is specific for the UGA-codon as the UAA-codon did not allow for seleniumdependent expression.
To test the effects of homocysteine on read-through expression, we treated transfected cells with HAT/Met in the presence or absence of selenium (Fig. 4, C and D). In the presence of selenium, HAT/Met treatment caused a significant reduction of read-through of both the GPx1/hSECIS/S (Fig. 4C) and the UGA/hSECIS/S (Fig. 4D) constructs. Similar reduction in readthrough expression was obtained when cells were treated with 100 M homocysteine (data not shown). No significant changes were found with the antisense constructs or in the absence of selenium. These data suggest that the HAT/Met-mediated reduction of selenium-dependent UGA read-through is primarily a translational event in this cell system, as the construct with the GPx1 promoter showed no additional suppression by this treatment.
To test whether the HAT/Met suppression of read-through was specific for the GPx1 SECIS element, we compared the function of the GPx3 SECIS element in this assay system. The GPx1 SECIS element has been defined as a form 1 SECIS element, whereas the GPx3 SECIS element has been defined as a form 2 SECIS element on the basis of the predicted secondary structure of the stem-loop region that directs selenocysteine incorporation (19). As expected, there was no read-through if the GPx3 SECIS was in the antisense orientation (data not shown). Selenium stimulates read-through when the GPx3 SE-CIS element was in the sense orientation and, as with the GPx1 SECIS element, this read-through is significantly affected by HAT/Met treatment (p Ͻ 0.005) (Fig. 4E).
Immunodetection of GPx1-To confirm the effects of HAT/ Met treatment on the expression of the GPx1 protein, we performed Western blots (Fig. 5). After adjusting for the slight lane-to-lane variation in immunodetectable actin levels, HAT/ Met treatment leads to a maximum 50% reduction in immunodetectable GPx1, whereas 100 M homocysteine reduces GPx1 by ϳ20%. Overall, these data suggest that an increase in cellular homocysteine results in a decrease in GPx1 activity because of a decrease in the translation of this protein. DISCUSSION Current views hold that homocysteine increases vascular oxidative stress (20), thereby contributing to a proatherogenic and prothrombotic state. Clinical studies suggest that modest increases in plasma homocysteine levels increase the risk of cardiovascular disease. In vivo, a mere doubling of homocysteine levels is sufficient to cause some proatherogenic changes, such as endothelial dysfunction, reduced bioavailable nitric oxide, increased lipid peroxidation, and increased adhesion molecule expression (4 -6). In models of hyperlipidemia, moderate hyperhomocysteinemia (i.e. levels of total homocysteine of ϳ50 M or ϳ5-fold higher than normal homocysteine levels) has been found to accelerate atherogenesis (21)(22)(23). In cell culture studies, although some detrimental effects of homocysteine are apparent at physiologically (10 -50 M) or pathophysiologically (100 -300 M) relevant concentrations (24,25), other adverse effects can only be measured when cells are treated with supraphysiological (1-5 mM) concentrations of homocysteine (9). In our previous studies, we have shown that micromolar increases in homocysteine correlate with decreased GPx1 activity in vivo and in cell culture systems; however, nonphysiological levels, in the millimolar range, are necessary to decrease the expression of GPx1 mRNA in endothelial cells (4 -6, 8). The current study indicates that treatment with low micromolar concentrations of homocysteine does not affect GPx1 mRNA levels or GPx1 promoter activity, although this treatment reduces GPx1 enzyme activity and the levels of GPx1 protein. The reporter gene assays further establish that the unique mechanism of selenocysteine incorporation is reduced by homocysteine.
Homocysteine is produced by the hydrolysis of S-adenosylhomocysteine, which is formed after a transfer of a methyl group from S-adenosylmethionine to a methyl acceptor (26). Excess methionine is known to augment homocysteine production in vivo and in cultured cell systems. All cells have the capacity to form methionine by the transfer of a methyl group from the folate-containing methyl donor, 5-methyltetrahydrofolate, to homocysteine by the enzyme methionine synthase. Restriction of folate inhibits this remethylation pathway and contributes to the accumulation of homocysteine, whereas dietary folate supplementation has been shown to lower homocysteine levels (20). It is known that dihydrofolate reductase inhibitors increase homocysteine production by limiting the production of 5-methyltetrahydrofolate and other cellular fo-  4. Selenium-dependent read-through expression. A, vectors. The pGL2 UGA vector has a TGA mutation in the luciferase proteincoding region, a PvuII restriction endonuclease site in the 3Ј-untranslated region of the luciferase gene, and an SV40 promoter and enhancer. The GPx1/UGA vector has a GPx1 promoter substituted for the SV40 promoter. The human GPx1 selenocysteine incorporation sequence (hSECIS) was lates from dietary folates (27). In our cell system, we used the dihydrofolate reductase inhibitor aminopterin to decrease homocysteine remethylation. Extra methionine was also added to the cultures to increase homocysteine production. Treatment with HAT/Met decreased GPx1 activity in BAEC to a similar extent as 100 M homocysteine. We also found that in the presence of adenosine, the effects of 25 M homocysteine are augmented. These data suggest conditions that increase Sadenosylhomocysteine production (i.e. intracellular production of homocysteine or homocysteine plus adenosine) may enhance the detrimental effects of homocysteine. We were only able to measure a modest increase in homocysteine release into media in treated COS7 cells. This 30% increase is comparable with that found in many other cell systems after short term treat-ment in the absence of folate (17) or in the presence of excess methionine (18). As we have shown in our system, this incremental change is sufficient to alter GPx1 activity, and the modest decrease in antioxidant enzyme activity is consistent with the changes in GPx1 activity levels that are associated with mild hyperhomocysteinemia in the CBS(ϩ/Ϫ) mice (5,6).
The expression of GPx1 is under complex regulation that includes transcriptional (28,29), post-transcriptional (11), and translational mechanisms (12). Selenium has been shown previously to increase mRNA stability, and in the presence of selenium, we found a nearly 2-fold increase in the steady-state levels of GPx1 mRNA. Selenium is also necessary for the translation of GPx1 as it is a necessary precursor to selenocysteine formation. A specialized mechanism is involved in the incorporation of selenocysteine that involves decoding the UGA codon as a selenocysteine codon (12)(13)(14)(15). This translational process requires other cis-elements in the mRNA, notably the SECIS element that forms a stem-loop structure in the 3Ј-UTR. Transfer of the SECIS element to the 3Ј-UTR of heterologous mRNAs has been shown previously to promote recognition of a UGA embedded in the protein-coding region of a transcript as a site for SEC incorporation (16,30). An important requirement for read-through includes insertion of the SECIS element downstream of the protein-coding region in the proper orientation. In our system, selenium had no effect on the expression of constructs with the SECIS element in the antisense orientation. In fact, the activity of the antisense constructs was the same as the corresponding constructs that lacked the SECIS element, regardless of whether or not selenium was present. Selenium also had no effect on the activity of the UAA constructs and no effect on control constructs lacking nonsense mutations, indicating that the selenium effect is specific for the UGA codon.
In our studies, selenium increased the levels of GPx1 mRNA. Several studies have found that the stability of GPx1 mRNA is enhanced by selenium repletion (11,31,32); however, not all selenoprotein mRNAs are affected by selenium levels (32,33). Transcript stability is often mediated by sequences such as "AU-rich" elements in the 3Ј-UTR of the mRNA (34). Although mRNA instability elements have been found in the 3Ј-UTR of some selenoprotein transcripts (35), these sequence elements are not found in the 3Ј-UTR of GPx1 mRNA. In studies that have swapped the protein-coding regions and 3Ј-UTRs among GPx2, GPx4, and GPx1, the presence of either the GPx1 protein-coding region or the GPx1 3Ј-UTR correlated with mRNA instability, suggesting that more than one, as of yet unidentified, region of the GPx1 mRNA is involved in regulating the stability of GPx1 transcripts. Other studies suggest nonsensemediated decay (NMD) contributes to the degradation of transcripts with UGA codons that prematurely terminate translation. NMD appears to require splicing as intronless mRNAs with premature stop codons are exempt from NMD (36). NMD mechanisms play an important role in decreasing the levels of GPx1 transcripts when selenium is limiting (37); however, in selenium-replete conditions, NMD mechanisms do not influence GPx1 mRNA.
In selenium-replete conditions, HAT/Met or 100 M homocysteine did not affect GPx1 mRNA. These treatments also did were separated on 12% denaturing polyacrylamide gels and transferred to Hybond nitrocellulose membrane. Antibodies to the human GPx1 and actin were used in the Western blots. Horseradish peroxidase-conjugated second antibodies followed by ECL were used to detect proteins. Plots of densitometry analysis of this Western blot are shown. ADU, arbitrary densitometry units. not affect the expression of wild type luciferase in constructs that lacked the elements for selenocysteine incorporation (data not shown). In pGL2 UGA constructs, however, both the GPx1 SECIS and the GPx3 SECIS element conferred sensitivity to treatments that increased homocysteine levels. GPx3 is the extracellular GPx, a selenoprotein produced primarily in the kidney. GPx1 and GPx3 differ in their primary sequence, in the location of the selenocysteine, and in the forms of the SECIS elements responsible for selenocysteine incorporation. The GPx1 SECIS element has been described as a form 1 SECIS element, whereas the GPx3 SECIS element is a form 2 element (19). According to Grundner-Culemann et al. (19), the form 1 elements have a 10 -14-base loop at the apex of the SECIS stem, which contains three conserved adenosines. In the form 2 elements, there are two loops as follows: the conserved adenosines bulge forming a small loop proximal to the top of the stem, and there is a 3-6-base apical loop. Even though the SECIS elements differ between these transcripts, GPx1 mRNA and GPx3 mRNA are both sensitive to selenium depletion (38), and we found no statistically significant difference in the levels of selenium stimulation of read-through in constructs that used either the GPx1 or GPx3 SECIS elements. Minimal SECIS elements, like those studied in the context of a D1-selenoprotein reporter by Martin et al. (39), were not active in our heterologous reporter assay; thus, we cannot exclude the possibility that other elements present in the 3Ј-UTR may contribute to the sensitivity to HAT/Met-mediated suppression. Further analysis is necessary to determine whether a shared regulatory element or a common regulatory RNA-binding factor mediates the effects of homocysteine on selenium-dependent translation.
In the reporter gene system that we employed, the levels of expression of the TGA vectors do not approach the levels of activity of an unmutated control construct. Published reports suggest that incorporation of selenocysteine during translation is an inefficient event (40), as substitution of a cysteine codon for the selenocysteine codon results in a 20 -400-fold increase in expression (41,42). Others studies have found that location of the UGA in the context of the GPx1 transcript affects the efficiency of selenocysteine incorporation (43). In a recent report using a luciferase reporter gene assay to study the efficiency of selenocysteine incorporation (44), selenocysteine incorporation was also found to be altered by the location of the UGA-codon. Even after optimizing selenocysteine codon location, these authors found that the efficiency of expression of the selenocysteine-substituted luciferase was much lower than that of the normal, cysteine-containing luciferase.
High concentrations of homocysteine (in the millimolar range) have been shown to cause endoplasmic reticulum stress and activation of the unfolded protein response in endothelial cells and other cells grown in culture (9,45). Endoplasmic reticulum stress and the unfolded protein response result in a change in the translational program of cells (46). In vivo, mild hyperhomocysteinemia has been shown to promote the expression of some of the endoplasmic reticulum stress-response genes (9,45). These data suggest that mild-to-moderate hyperhomocysteinemia may also interfere with protein folding and normal translational events. Incorporation of selenocysteine may be extremely sensitive to events that affect translation. Martin and Berry (40) raise the possibility that there may be competition between recognition of the UGA as a selenocysteine codon and recognition of UGA as a termination site.
A necessary event for selenocysteine incorporation is the formation of a stable complex of the tRNA sec , eF sec , and selenocysteine-specific mRNA at the ribosome (12,42,47). RNAbinding proteins, such as SBP2, may serve to stabilize this complex. SBP2 has been shown to have a redox-sensitive thiol (48) that may be a target for homocysteine-induced alterations. Homocysteine has recently been found to alter the binding of a redox-sensitive RNA-binding protein, leading to enhanced translation and increased levels of folate receptor (17).
In the current report, we show that both GPx1 enzyme activity and GPx1 immunodetectable protein levels are reduced by increases in homocysteine levels, although there is no effect on GPx1 mRNA under these conditions. We have also found that increased production of homocysteine significantly reduced selenium-dependent UGA read-through in a reporter gene, whether the SV40 or the GPx1 promoter was used to drive transcription. Taken together, these data suggest that homocysteine regulates GPx1 expression by a mechanism that involves down-regulation of translation. This unique mechanism may contribute to the proatherogenic and prothrombotic effects of homocysteine by limiting the production of this and potentially other important selenoproteins.