The utilization of selenocysteine-tRNA^[Ser]Sec isoforms is regulated in part at the level of translation in vitro

ABSTRACT

The tRNA for the 21st proteinogenic amino acid, selenocysteine, exists in mammalian cells as 2 isoforms differing by a single 2′-O-methylribosyl moiety at position 34 (Um34). These isoforms contain either 5-methoxycarbonylmethyluridine (mcm⁵U) or 5-methoxycarbonylmethyl-2′-O-methyluridine (mcm⁵Um) at position 34. The accumulation of the mcm⁵Um isoform is tightly correlated with the expression of nonessential “stress response” selenoproteins such as glutathione peroxidase 1 (GPX1). The expression of essential selenoproteins, such as thioredoxin reductase 1 (TXNRD1), is not affected by changes in Sec-tRNA^[Ser]Sec isoform accumulation. In this work we used purified mcm⁵U and mcm⁵Um Sec-tRNA^[Ser]Sec isoforms to analyze possible differences in binding to the selenocysteine-specific elongation factor, EEFSEC, and the translation of GPX1 and TXNRD1 in vitro. Our results indicate that no major distinction between mcm⁵U and mcm⁵Um isoforms is made by the translation machinery, but a small consistent increase in GPX1 translation is associated with the mcm⁵Um isoform. These results implicate fundamental differences in translation efficiency in playing a role in regulating selenoprotein expression as a function of isoform accumulation.

KEYWORDS:

• glutathione peroxidase
• isoforms
• methylation
• selenium
• selenocysteine
• thioredoxin reductase
• translation
• tRNA modification

Introduction

The 21st proteinogenic amino acid, selenocysteine (Sec), is brought to the elongating ribosome as Sec-tRNA^[Ser]Sec. Although Sec-tRNA^[Ser]Sec is undermodified relative to other tRNAs,¹ it does contain a dynamic methylation in the anticodon loop at uridine 34 (U34), which is synthesized on 5-methoxycarbonylmethyluridine (mcm⁵U) giving rise to 5-methoxycarbonylmethyl-2′-O-methyluridine (mcm⁵Um). The synthesis of the mcm⁵Um isoform of Sec-tRNA^[Ser]Sec (hereafter referred to simply as either mcm⁵U or mcm⁵Um) depends on the unique tertiary structure of tRNA^[Ser]Sec,² and its presence has a dramatic impact on the overall structure of the tRNA.^3,4 It has been recently established at the biochemical and genetic levels that methylcarboxymethlylation at position U34 is important for the stable interaction of AAA, CAA and GAA codons in yeast.^5,6 The additional functionality conferred by the further methylation to mcm⁵Um has also been linked to translational fidelity,⁷ but the mechanism has not been studied.

In mammals, the regulated synthesis of mcm⁵Um is apparently unique to Sec-tRNA^[Ser]Sec, and it is dependent on the presence of an adenosine residue at position 37, prior base modifications at positions 55 (pseudouridine) and 58 (1-methyladenosine), intact secondary and tertiary structures, and aminoacylation of tRNA^[Ser]Sec,^2,8-11 as well as the prior methylcarboxymethylation at position U34.¹² The isopentenylation of Sec-tRNA^[Ser]Sec is performed by the TRIT1 enzyme, genomic deletion of which results in a complete loss of the mcm⁵Um isoform.¹³ The production of mcm⁵U is catalyzed by the ALKBH8 methyltransferase, which has also been demonstrated to be essential for mcm⁵Um production in ALKBH8 null mice.¹² The mcm⁵U and mcm⁵Um isoforms have been proposed to be differentially used by the nonessential “stress-related” selenoprotein genes such as GPX1. When mcm⁵Um production is inhibited by mutating the U34 or A37 positions of Sec-tRNA^[Ser]Sec, the production of the stress-related selenoproteins, including GPX1, is dramatically reduced.^9,10,14 Interestingly, the extent of mcm⁵Um modification is dependent on the cellular selenium concentration, thus providing a mechanistic basis for the well known hierarchy of selenoprotein expression during selenium deficiency.¹⁵ Consistent with these findings, mice lacking TRIT1 activity have reduced expression of the nonessential selenoproteins,¹³ and the ALKBH8 null mice also showed reduced GPX1 levels, albeit to a lesser extent.¹²

The mechanism by which nonessential selenoprotein expression is regulated by the presence or absence of mcm⁵Um Sec-tRNA^[Ser]Sec has not been explored. Using GPX1 as an example, one possibility is that the mcm⁵U isoform is excluded from recruitment to ribosomes loaded with GPX1 mRNA, which might result from differential affinity to the Sec-specific elongation factor EEFSEC. Alternatively, it could be that utilization of the mcm⁵U isoform for GPX1 translation signals the degradation of GPX1 mRNA in a process akin to nonsense mediated mRNA decay (NMD).^16-18 This latter hypothesis is supported by the fact that several studies have demonstrated that GPX1 mRNA is targeted for destruction during selenium deficiency,^19-22 which correlates with the loss of mcm⁵Um Sec-tRNA^[Ser]Sec.^3,23,24

In this study, we used the purified Sec-tRNA^[Ser]Sec isoforms to determine whether the mcm⁵Um isoform is preferentially used during in vitro translation and/or elongation factor binding. Although we observed no difference between mcm⁵U and mcm⁵Um Sec-tRNA^[Ser]Sec utilization in rabbit reticulocyte lysate, we did observe a small and consistent increase in translation efficiency for GPX1 (nonessential), but not for the essential thioredoxin reductase (TXNRD1) mRNA in a wheat germ system that lacks endogenous Sec-tRNA^[Ser]Sec. However, we also did not observe preferential binding of either of the 2 isoforms by EEFSEC, which is consistent with the idea that the regulation of mcm⁵Um utilization must occur after this binding event.

Results & discussion

Recombinant EEFSEC does not show a preference for binding to either mcm⁵U or mcm⁵Um isoforms

GPX1 protein expression is found to be dramatically reduced in vivo when the expression of the mcm⁵Um isoform is inhibited.^10,14 One explanation for this result may be that the mcmU/EEFSEC ternary complex may be intrinsically less stable than the mcm⁵Um/EEFSEC complex. To determine whether the Sec-tRNA^[Ser]Sec isoforms are differentially bound by the Sec-specific elongation factor, EEFSEC, we took advantage of the well-established method of determining the ratio of mcm⁵U to mcm⁵Um by RPC5 chromatography.^25-27 Fig. 1A shows the characteristic separation of wild-type mouse liver-derived tRNA^[Ser]Sec that was aminoacylated in vitro with [³H]Ser, the early and late-eluting peaks corresponding to the mcm⁵U and mcm⁵Um isoforms, respectively. To test EEFSEC binding, we performed affinity purification of Sec-tRNA^[Ser]Sec from mouse liver total RNA with immobilized FLAG tagged recombinant EEFSEC. The EEFSEC-affinity purified tRNA was subject to deacylation and reacylation with [³H]Ser to verify integrity (Fig. 1B), and the distribution of Sec-tRNA^[Ser]Sec isoforms that were recovered by EEFSEC affinity chromatography was determined by RPC-5 chromatography. Fig. 1C and D show the profiles of Sec-tRNA^[Ser]Sec isoforms derived from total liver tRNA (Fig. 1C) versus that from EEFSEC affinity (Fig. 1D). The fact the EEFSEC-purified and total liver Sec-tRNA^[Ser]Sec profiles are similar indicates that EEFSEC does not preferentially bind to one isoform or the other under these conditions. These results indicate that the strong cellular discrimination between the utilization of mcm⁵U and mcm⁵Um isoforms in protein synthesis^9,10,14 is not likely occurring at the level of EEFSEC binding due to the slight differences we observed in their binding levels to this cellular component. We performed this experiment a single time due to limiting access to equipment and material. However, we are unable to discern a significant bias in EEFSEC binding, and the results observed are well within the typical variance we have found in this assay since it was introduced by us and used for quantifying the levels of the 2 Sec tRNA isoforms under many different physiologic conditions.^9,10,14,28 In addition, considering this assay used purified components, we cannot rule out the contribution of cellular factors that may regulate EEFSEC affinity.

Figure 1. Analysis of Sec-tRNA^[Ser]Sec isoforms by RPC-5 chromatography. (A) Purified [3H]Ser-tRNA^[Ser]Sec was resolved into the 2 characteristic isoform peaks by RPC-5 column chromatography. Fractions were analyzed by scintillation counting. (B) Sec-tRNA^[Ser]Sec purified by EEFSEC affinity was aminoacylated in vitro with [3H]Ser as described in the experimental procedures. Total liver tRNA (C) or tRNA eluted from anti-FLAG beads loaded with FLAG-EEFSEC (D) was applied to an RPC-5 column and tRNA^[Ser]Sec was detected in fractions by dot blot hybridization with a [³²P] labeled probe.

In vitro translation of GPX1 and Selenop mRNAs in rabbit reticulocyte lysate is not dependent on the Sec-tRNA^[Ser]Sec isoform

Since EEFSEC appears to form a ternary complex with mcm⁵U just as well as mcm⁵Um, it is possible that the differential utilization observed in cells occurs at the level of translation. To test whether there is an intrinsic difference in GPX1 translation as a function of tRNA^[Ser]Sec isoform, we performed in vitro translation experiments in rabbit reticulocyte lysate. For this experiment, ⁷⁵Se-labeled mcm⁵U and mcm⁵Um isoforms were purified by RPC-5 chromatography. As shown in Fig. 2A, we obtained 2 well resolved peaks, the first corresponding to mcm⁵U and the latter to mcm⁵Um. Early and late fractions from each peak were pooled so as to reduce the chance of isoform mixing. These purified forms of Sec-tRNA^[Ser]Sec were added to rabbit reticulocyte lysate in vitro translation reactions containing either GPX1 or SELENOP mRNA. As shown in Fig. 2B, ⁷⁵Se-labeled bands corresponding to the expected molecular weights for both SELENOP and GPX1 were observed. Interestingly, we observed no discernible difference in the efficiency of translation for either mRNA as a function of isoform. As a control, reactions were also performed in the presence of free ⁷⁵Se-selenite which is incorporated into the endogenous tRNA^[Ser]Sec in rabbit reticulocyte lysate. We chose to use the SELENOP mRNA because it contains 10 Sec codons, and thus represents a unique challenge to the Sec incorporation machinery. As such, we expected that SELENOP translation might be a sensitive indicator of differences in tRNA isoform utilization, but under these conditions, such a difference was not observed. Overall, these results do not reveal a difference in isoform utilization, but the fact that substantial endogenous Sec-tRNA^[Ser]Sec is present in rabbit reticulocyte lysate²⁹ is a complicating factor if pre-formed EEFSEC/Sec-tRNA^[Ser]Sec ternary complex is somehow affecting isoform utilization.

Figure 2. In vitro translation of GPX1 and SELENOP mRNAs in rabbit reticulocyte lysate using purified ⁷⁵Se-labeled Sec-tRNA^[Ser]Sec. (A) Large scale purification of the mcm⁵U and mcm⁵Um isoforms of Sec-tRNA^[Ser]Sec by RPC-5 chromatography. The shaded areas represent the fractions that were pooled for each isoform. (B) 100 ng each of GPX1 and SELENOP mRNAs were translated in rabbit reticulocyte lysate in the presence of inorganic ⁷⁵Se or the Sec-tRNA^[Ser]Sec isoforms as indicated. The fold increase in SELENOP and GPX1 translation as a function of the presence of mcm⁵Um Sec-tRNA^Sec is indicated below.

In vitro translation of GPX1 mRNA in a reconstituted Sec incorporation system is slightly enhanced with the mcm⁵Um isoform

Because the rabbit reticulocyte lysate system contains endogenous Sec-tRNA^[Ser]Sec, we endeavored to use a more defined system to analyze any potential difference between the 2 tRNA^[Ser]Sec isoforms. To that end, we used the wheat germ lysate system that we recently developed to characterize the function of EEFSEC.³⁰ Since higher plants do not utilize Sec and do not possess any of the Sec incorporation components, it is an ideal system in which to analyze the core mechanism of Sec incorporation factors. We are unable to use the SELENOP mRNA in this system as the incorporation of multiple Sec residues in a single protein does not occur in these conditions.³¹ Instead, we generated mRNA encoding mouse TXNRD1, the expression of which is not regulated in vivo by altered Sec-tRNA^[Ser]Sec isoforms.^10,14 Based on the cellular and in vivo data, we expected that GPX1 would be translated with higher efficiency with the mcm⁵Um-containing tRNA while TXNRD1 mRNA should be translated with equal efficiency regardless of isoform.^10,14 We used a range of tRNA concentrations to examine the dynamic response of GPX1 vs. TXNRD1 translation in wheat germ lysate. As shown in Fig. 3, GPX1 translation appears to be slightly enhanced in the presence of the mcm⁵Um isoform by a maximum of ∼2-fold. Interestingly, the overall efficiency of TXNRD1 translation is much higher than that of GPX1 in wheat germ lysate, regardless of isoform. To determine if the difference in efficiency of GPX1 vs. TXNRD1 mRNA translation was specific for the wheat germ system, we used rabbit reticulocyte lysate and wheat germ lysate in the same experiment for a direct comparison. As seen in Fig. 3B, the translation efficiency of GPX1 and TXNRD1 was similar in rabbit reticulocyte lysate but TXNRD1 synthesis was up to 31-fold higher in wheat germ. Considering the difference in TXNRD1 translation efficiency is specific to the wheat germ lysate, we expect this is due to the longer 3′ UTR on TXNRD1 mRNA (663 nucleotides) compared with that on GPX1 (181 nucleotides), which is a known factor regulating translation efficiency in the wheat system.³²

Figure 3. In vitro translation of GPX1 and TXNRD1 in wheat germ lysate using purified ⁷⁵Se-labeled Sec-tRNA^[Ser]Sec. (A) A range of mcm⁵U and mcm⁵Um Sec-tRNA^[Ser]Sec amounts from 2000–6000 cpm were added to wheat germ in vitro translation reactions programmed with 4 nM of either GPX1 (lanes 1–7) or TXNRD1 mRNA (lanes 8–14). The fold increase in GPX1 and TXNRD1 translation as a function of the presence of mcm⁵Um Sec-tRNA^[Ser]Sec is indicated below. (B) 1 and 4 nM GPX1 or TXNRD1 mRNAs were translated in rabbit reticulocyte lysate and wheat germ lysate. The fold increase of GPX1 and TXNRD1 in reticulocyte lysate vs. wheat germ lysate is indicated below.

TXNRD1 translation outcompetes that of GPX1

To further analyze the dynamics of GPX1 vs. TXNRD1 translation, we performed a mixing experiment where the mRNAs were translated either alone or as an equimolar mixture. Interestingly, as shown in Fig. 4, we observed a significant decrease in GPX1 synthesis but the slight enhancement with mcm⁵Um is retained (Fig. 4A, compare lanes 3 and 6). This result is consistent with prior work in vivo that the translational efficiency of TXNRD1 appears to be significantly higher than that of GPX1.^33,34

Figure 4. In vitro translation of GPX1 and TXNRD1 together in wheat germ lysate. (A) GPX1 and TXNRD1 mRNAs were translated in the presence of mcm⁵U and mcm⁵Um Sec-tRNA^[Ser]Sec isoforms as described in Figure 3 but in lanes 3 and 6, equal amounts of mRNA were combined in the same reaction. Phosphorimage quantitation of bands plotting the percent increase as a function of using the mcm⁵Um Sec-tRNA^[Ser]Sec isoform is shown below. (B) Phosphorimage quantitation of multiple experiments comparing the amount of GPX1 or TXNRD1 translation in the presence of mcm⁵U vs. mcm⁵Um Sec-tRNA^[Ser]Sec. A total of 19 experiments are included in this analysis, which is an unpaired 2-tailed t-test.

To maximize the use of a small supply of purified isoforms, we quantified multiple non-identical experiments in which we compared GPX1 translation vs. TXNRD1 translation in the presence of mcm⁵U or mcm⁵Um (n = 19). We found an average of 1.7-fold increased GPX1 translation with mcm⁵Um but little detectable increase (1.1-fold) for TXNRD1. The experiments included in this analysis include all of those shown in Figs. 3 and 4, plus additional experiments that only differed in the amounts of tRNA or mRNA added. This difference was found to be statistically significant in a 2-tailed unpaired t-test (Fig. 4B; p = .002). Whether the small increase in efficiency in the presence of the mcm⁵Um isoform is functionally significant remains to be tested.

Conclusion

In vivo, the preferential utilization of mcm⁵Um in stress-related selenoprotein expression has been shown to be stringent, since mice with mutations in either position 34 or 37 of Sec-tRNA^[Ser]Sec lack the mcm⁵Um isoform,² resulting in a low level of expression for this subclass of selenoproteins.^10,14 The purpose of the present study was to elucidate at which step in translation the involvement of mcm⁵Um occurs in regulating stress-related selenoprotein expression. Our data clearly show that there is no discrimination between mcm⁵U and mcm⁵Um at the level of EEFSEC selection, and thus, the distinction must occur at a subsequent step in translation. Removing many of the variables associated with cell and animal based systems, we found that both GPX1 and TXNRD1 can be translated by using either the mcm⁵U or mcm⁵Um isoforms in rabbit reticulocyte lysate. However, this observation may indicate that the selectivity observed in vivo for preferential utilization of mcm⁵Um has been lost in lysate preparation and/or in carrying out translation under these conditions. On the other hand, we do observe a small but reproducible increase in GPX1 translation in the presence of the mcm⁵Um isoform in wheat germ extract, which may indicate, at least in part, the basis for a cellular mechanism that distinguishes the translation of essential and nonessential selenoprotein mRNAs. This intrinsic difference in efficiency, for example, may be just enough to trigger a response that ultimately leads to GPX1 mRNA decay, thus eliminating GPX1 expression when mcm⁵Um concentrations are low.

Experimental procedures

Binding of Sec-tRNA^[Ser]Sec to EEFSEC

Total tRNA was purified from mouse liver as described previously.²⁶ Sec-tRNA^[Ser]Sec was purified by EEFSEC affinity as described previously.³⁰ Briefly, FLAG-EEFSEC was bound to anti-FLAG beads (Sigma-Aldrich) in Buffer B (20 mM Tris-HCl pH 7.5, 20 mM KCl, 0.1 mM EDTA, 25% glycerol) and incubated with mouse liver amino acyl tRNA in the presence of 0.5 mM GTP at 4°C for 1 h. The EEFSEC/Sec-tRNA^[Ser]Sec complex was eluted in 200 μl of Buffer B with 250 μg/ml of 3X FLAG peptide for 30 min at 4°C.

Analysis of EEFSEC-bound Sec-tRNA^[Ser]Sec

Purified Sec-tRNA^[Ser]Sec was deacylated in 1 M Tris (pH 8.0) at 37°C for 1 h. Deacylated tRNA^[Ser]Sec was precipitated with 3 volumes of ethanol, collected by centrifugation at 14,000 RPM for 15 min, washed with 75% ethanol, dried and dissolved in nuclease-free H₂O. Yeast total tRNA (600 µg) was added to the tRNA^[Ser]Sec as a carrier and tRNA was chromatographed on an RPC-5 column in a linear gradient from 0.525 M sodium chloride to 0.675 M sodium chloride containing 10 mM sodium acetate, 10 mM magnesium acetate, 1 mM EDTA, pH 4.5²⁶.

Dot blot analysis was performed by spotting 5 µl of each column fraction on a Hybond-N+ membrane (GE Healthcare Life Sciences) followed by UV cross-linking. Hybridization was performed using Quikhyb solution (Stratagene) for 4 h at 58 ˚C in a rotating hybridization oven with a [³²P]-end labeled oligo probe complementary to tRNA^[Ser]Sec (5′-CAG ACC ACT GAG GAT CAT CCG-3′), which was prepared using T4 polynucleotide kinase (New England Biolabs) and γ⁻³²P ATP (Specific activity 3000Ci/mmol, Perkin Elmer) according to manufacturer's instructions. Following hybridization, the membrane was washed 3 times with 2X SSC, 0.1% SDS, then washed 2 times with 0.1X SSC, 0.1% SDS, exposed to a PhosphorImager screen and spots were quantitated using Imagequant software (GE Healthcare Life Sciences). Aminoacylation of purified tRNA^[Ser]Sec was performed using [³H]Ser (specific activity 20Ci/mmol, from Moravek Biochemicals) as described²⁶ and labeled tRNA^[Ser]Sec was chromatographed on an RPC-5 column as described above.

Isolation of [⁷⁵Se]-labeled Sec-tRNA^[Ser]Sec

HL-60 cells were grown in the presence of 300 nM sodium selenite as described previously.²⁴ Cells (∼5 g) were collected, washed with PBS and re-suspended in 75 ml of RPMI-1640 medium containing 1% FBS without sodium selenite. Five millicuries of neutralized ⁷⁵Se (Specific activity ∼1000 Ci/mmol, University of Missouri Research Reactor Center, Columbia, MO) were added, and the cells were gently shaken for 3 h at 37°C. Cycloheximide (100 µM final concentration) was then added and cells were incubated for an additional 45 min. Cells were collected by centrifugation at 1,200 RPM for 5 min at 4°C, washed with ice cold PBS, and stored at −80°C until tRNA extraction. Phenol/chloroform extraction and purification of [⁷⁵Se]-labeled Sec-tRNA^[Ser]Sec by RPC-5 chromatography was performed as described.²⁴

GPX1 and TXNRD1 cloning and mRNA synthesis

Full length mouse GPX1 (coding region plus 181 base pair 3′ UTR) was isolated from mouse liver cDNA by PCR with a Kozak consensus sequence-containing 5′ primer (5′ GCC ACC ATG TGT GCT GCT CGG CTC TCC-3′), and a reverse primer ending immediately upstream of the polyadenylation signal (5′ CTT AGT AGT GAA AC ACC TTT-3′). Partial length mouse TXNRD1 (full coding region plus 663 base pair 3′ UTR) was cloned from mouse cortex cDNA using a Kozak consensus sequence-containing 5′ primer (5′-GCC ACC ATG CCA GTT GAT GAC TGC TGG-3′) and 3′ primer ending 983 base pairs upstream of the polyadenylation signal. Plasmid DNA was linearized with XbaI and 5′ capped mRNA was synthesized by in vitro transcription (mMessage, Ambion).

In vitro translation

Wheat germ in vitro translation reactions (12 μl) contained 6.5 μl of wheat germ extract (Promega), GPX1, SELENOP or TXNRD1 mRNA as indicated in the figure legends, [⁷⁵Se]-labeled tRNA (∼6000 cpm unless otherwise indicated) and 320 nM each of C-terminal SECIS binding protein 2 and EEFSEC recombinant proteins as described previously.³⁰ Reactions were incubated at 25 ˚C for 2 hours followed by treatment with 10 μg of RNAse A for 15 min. at 37 ˚C. Rabbit reticulocyte lysate (Promega) reactions (12 ul) contained 8 μl of lysate, mRNA as indicated in the figure legends, 320 nM C-terminal SECIS binding protein 2 and ⁷⁵Se-labeled Sec-tRNA^[Ser]Sec or inorganic ⁷⁵Se as indicated. Radiolabeled proteins were resolved by 12% SDS-PAGE followed by Phosphorimaging.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by a grant from the National Institutes of Health, GM077073 to P.R.C. and in part by the Intramural Research Program of the National Institutes of Health, NCI, Center for Cancer Research to D.L.H.