Chemical-biology approaches to interrogate the selenoproteome

Conspectus:

Selenoproteins are the family of proteins that contain the amino acid selenocysteine. Many selenoproteins, including glutathione peroxidases and thioredoxin reductases, play a role in maintaining cellular redox homeostasis. There are a number of examples of homologs of selenoproteins that utilize cysteine residues, raising the question of why selenocysteines are utilized. One hypothesis is that incorporation of selenocysteine protects against irreversible overoxidation, typical of cysteine-containing homologs under high oxidative stress. Studies of selenocysteine function are hampered by challenges both in detection and recombinant expression of selenoproteins. In fact, about half of the 25 known human selenoproteins remain uncharacterized. Historically, selenoproteins were first detected via labeling with radioactive ⁷⁵Se, or by use of inductively coupled plasma-mass spectrometry to monitor non-radioactive selenium. More recently, tandem mass-spectrometry techniques have been developed to detect selenocysteine-containing peptides. For example, the isotopic distribution of selenium has been used as a unique signature to identify selenium-containing peptides from unenriched proteome samples. Additionally, selenocysteine-containing proteins and peptides were selectively enriched using thiol-reactive electrophiles by exploiting the increased reactivity of selenols relative to thiols, especially under low pH conditions. Importantly, the reactivity-based enrichment of selenoproteins can differentiate between oxidized and reduced selenoproteins, providing insight into the activity state. These mass spectrometry-based selenoprotein detection approaches have enabled: (1) production of selenoproteome expression atlases; (2) identification of aging-associated changes in selenoprotein expression; (3) characterization of selenocysteine reactivity across the selenoprotein family; and, (4) interrogation of selenoprotein targets of small-molecule drugs. Further investigations of selenoprotein function would benefit from recombinant expression of selenoproteins. However, the endogenous mechanism of selenoprotein production makes recombinant expression challenging. Primarily, selenocysteine is biosynthesized on its own tRNA, and is dependent on multiple enzymatic steps and is highly sensitive to selenium concentrations. Furthermore, selenocysteine is encoded by the stop codon UGA, and suppression of that stop codon requires a selenocysteine insertion sequence element in the selenoprotein mRNA. In order to circumvent the low efficiency of the endogenous machinery, selenoproteins have been produced in vitro through native chemical ligation and expressed protein ligation. Attempts have also been made to engineer the endogenous machinery for increased efficiency, including recoding the selenocysteine codon, and engineering the tRNA and the selenocysteine insertion sequence element. Alternatively, genetic code expansion can be used to generate selenoproteins. This approach allows for selenoprotein production directly within their native cellular environment, whilst bypassing the endogenous selenocysteine incorporation machinery. Furthermore, by incorporating a caged selenocysteine by genetic code expansion, selenoprotein activity can be spatially and temporally controlled. Genetic code expansion has allowed for the expression and uncaging of human selenoproteins in E. coli and more recently in mammalian cells. Together, advances in selenoprotein detection and expression should enable a better understanding of selenoprotein function and provide insight into the necessity for selenocysteine production.

Graphical Abstract

Introduction

Selenocysteine (Sec) is a cysteine (Cys) cognate with selenium (Se) in the place of sulfur (S) (Figure 1A), and proteins that contain Sec are categorized as selenoproteins. Selenoproteins are observed in all domains of life, though they are not observed in every organism within each domain. Functionally characterized selenoproteins primarily play a role in redox regulation in the cell, exemplified by glutathione peroxidases (Gpx)1–4 and 6, thioredoxin reductases (Txnrd)1–3, and methionine-R-sulfoxide reductase B1 (MsrB1).¹

Figure 1.

Selenoproteins and their expression. A) The structure of selenocysteine (Sec). B) The mammalian selenoproteome. C) The Sec biosynthesis pathways in E. coli and eukaryotic systems. D) The Sec insertion mechanism in E. coli and eukaryotic systems.

Intriguingly, many homologous proteins exist that perform the same functions as selenoproteins, but with Cys in the place of Sec. Such homologs include proteins from organisms that do not make selenoproteins, as well as related protein isoforms within the same organism. For example mammalian MsrB1 is a selenoprotein located in the cytosol and nucleus, while MsrB2 and MsrB3, localized to the mitochondria and ER, perform the same reductase function with an active site Cys.² Cys and Sec side chains are both nucleophilic and capable of catalyzing similar chemistries, but display significantly different pKa values (approximately 8.5 for Cys and 5.2 for Sec). Both Cys and Sec are susceptible to oxidation, however, Sec is resistant to irreversible oxidation, due to the lack of pi-bond character in the Se-O bond compared to S-O bonds.³ Human Gpx4 is an example of a selenoprotein demonstrated to be resistant to irreversible overoxidation. When the Gpx4 active site is mutated to Cys instead of Sec, the enzyme is subject to overoxidation to sulfinic and sulfonic acid forms, resulting in ferroptotic cell death of neuronal cells that harbor the mutant enzyme.⁴

Many selenoproteins are poorly characterized, especially with regards to their endogenous function and regulation.¹ Challenges in further characterization of these selenoproteins stem from difficulties in the detection and recombinant expression of selenoproteins. Difficulties in the detection of selenoproteins can be attributed to the low number and low abundance of these proteins. The selenoproteome is small; there are 25 human selenoproteins and 24 mouse selenoproteins (Figure 1B).¹ Furthermore, many of these selenoproteins have variable levels of protein expression based on tissue type¹ and Se concentrations.⁵ For these reasons, proteome-wide studies of selenoproteins have historically been challenging.

Recombinant expression of selenoproteins is a challenge due to the complexities in the endogenous selenoprotein translation mechanism. Compared to other proteinogenic amino acids, Sec has a noncanonical insertion mechanism. Sec is genetically encoded by the codon UGA, which also serves as a stop codon.⁶ Cells must suppress the stop codon in order to incorporate Sec. UGA is suppressed by a tRNA^[Ser]Sec harboring the ACU anticodon, and Sec is generated directly on this tRNA (Figure 1C). First, the tRNA is charged with a serine (Ser) residue by seryl-tRNA synthetase (SerS), generating Ser-tRNA^[Ser]Sec. In prokaryotes, the hydroxyl group in Ser-tRNA^[Ser]Sec is replaced with Se by the Sec synthase (SecS), generating Sec-tRNA^[Ser]Sec.⁷ In eukaryotes, Ser-tRNA^[Ser]Sec is first phosphorylated by phosphoseryl-tRNA^[Ser]Sec kinase (PSTK) to generate pSer-tRNA^[Ser]Sec, which is utilized by SecS to generate Sec-tRNA^[Ser]Sec.⁸

In order for Sec-tRNA^[Ser]Sec to successfully suppress the UGA stop codon, a number of cis- and trans-acting elements are required (Figure 1D). The cis-acting Sec insertion sequence (SECIS) element is a stem loop mRNA structure.⁹ In prokaryotes the SECIS element is found immediately downstream of the UGA codon, and in eukaryotes it appears in the 3’UTR of the mRNA.¹ Eukaryotic selenoprotein translation also minimally requires SECIS binding protein 2 (SBP2), which binds to the ribosome and SECIS element, as well as the Sec-specific elongation factor (eEFSec), which recruits Sec-tRNA^[Ser]Sec to the ribosome.⁹ Prokaryotes, however, utilize a Sec-specific translation elongation factor (called SelB in E. coli) as a single trans-acting element in selenoprotein translation.¹⁰ This complex mechanism of Sec incorporation leads to incompatibilities and inefficiencies in recombinant expression of selenoproteins.

Here we summarize recent advances in proteomic detection and recombinant expression of selenoproteins. We aim to frame these recent advances within the historical context for selenoprotein detection and expression, and highlight existing limitations and potential for future improvements. We believe that these techniques will be of great benefit to our understanding of selenoprotein biology.

Selenoprotein detection methods

Sec is a rarely used amino acid; it appears in only 24 proteins in mice and 25 proteins in humans. Typical selenoprotein expression levels are low, and vary based on tissue and available Se concentrations. Detection of selenoproteins is therefore challenging. Methods have been developed to overcome issues of selenoprotein abundance, taking advantage of the scarcity of Se in biological samples, the unique isotope pattern of selenium, and the specific reactivity of Sec. Together, these methods enable evaluation of selenium distribution across the proteome, and interrogation of the activity state of individual selenoproteins within a biological sample.

⁷⁵Se labeling

Selenium has long been known to be an essential element for health, and Sec in selenoproteins was identified as the predominant form of Se in the cell. Se localization was initially traced using radiolabeled ⁷⁵Se. ⁷⁵Se labeling has been used in a number of biological systems since the 1950s, and even evaluated as a tumor-localizing agent in humans in the 1960s.¹¹ The ⁷⁵Se radioisotope is a manually produced gamma emitter with a t_1/2 of approximately 120 days,¹² and is generally provided as selenious acid (H₂⁷⁵SeO₃). To detect selenoproteins, selenious acid is administered directly to a biological system (including tissue culture, or live animals). After an incubation period, often 1–3 days, cells or tissues are harvested, and lysates are analyzed by SDS-PAGE (either 1D or 2D) and phosphor imaging (Figure 2A).^{12 75}Se-labeled proteins can then be isolated from the gel and identified via tandem mass spectrometry. ⁷⁵Se labeling has been effectively used to interrogate the Se dependence of selenoprotein expression levels across different tissues in an organism,¹³ as well as within a single tissue like the brain.⁵ These studies have led to the hypothesis that there is a hierarchy of selenoprotein expression that includes proteins that are expressed at similar amounts in low or high Se environments and those that exhibit strongly Se-dependent expression. Limitations of ⁷⁵Se labeling include the hazardous nature of working with such a radioisotope, as well as limited sensitivity levels and model-system compatibility. In prokaryotic systems, ⁷⁵Se is erroneously incorporated into Cys and methionine (Met) residues, increasing background signals.¹²

Figure 2.

Methods of selenoprotein detection. A) Radioactive ⁷⁵Se labeling. B) ICP-MS-based selenoproteomics method. C) Computational identification of selenium-containing isotopic envelopes in MS/MS datasets. D) Activity-based detection of selenoproteins via Sec labeling with electrophilic probes, enrichment, and tandem MS.

Selenium detection by ICP

Monitoring non-radioactive selenium isotopes is also a viable method for selenoprotein detection. Both the unique isotopic signature of Se and the scarcity of Se in protein samples (besides in Sec) allows for specific detection of selenoproteins via inductively coupled plasma mass spectrometry (ICP-MS). This approach is advantageous compared to ⁷⁵Se labeling because it avoids the use of radioactivity and does not require exogenous Se addition to the system.¹⁴ ICP-MS can be adapted to analyze the selenoproteome through the following steps: (1) proteins in the lysate sample are separated by isoelectric focusing; (2) laser-ablation ICP-MS (LA-ICP-MS) is utilized to identify Se-containing bands in the isoelectric focusing gel; (3) the non-ablated portion of the band that corresponds to each Se-containing peak is cut from the gel and digested for LC-MS analysis; (4) one portion of the sample is analyzed by HPLC-ICP-MS, and the retention time of the Se-containing peak is noted; and, (5) the remaining sample is then analyzed by HPLC-ESI-MS/MS; at the retention time noted for Se signal, the precursor ion mass (MS1) is used to confirm the presence of a Se isotopic envelope and the fragmentation spectra (MS2) is used to manually sequence the peptide that contains Se (Figure 2B).¹⁵ This technique has been used to analyze the selenoproteome of multiple cell lines, and led to the successful identification of 5 human selenoproteins, Gpx1, Gpx4, Txnrd1, Txnrd2, and SelenoF.¹⁶ Limitations of this approach include the extensive manual analysis that is required, the lack of high quantitative capacity, the low coverage of the selenoproteome, and the inability to identify any modifications of Sec, such as oxidation, that could affect selenoprotein activity.

SESTAR

More recently, an approach has been developed to identify Sec-containing peptides automatically from LC-MS/MS samples, without the need for ICP. This approach, termed selenium-encoded isotopic signature targeted profiling (SESTAR), exploits the unique isotopic signature of Se.¹⁷ There are 6 naturally-occurring Se isotopes (⁷⁴Se, ⁷⁶Se, ⁷⁷Se, ⁷⁸Se, ⁸⁰Se, and ⁸²Se), and their distinctive distribution is readily identifiable in MS spectra (Figure 2C). The SESTAR technique is inspired and informed by algorithms that have previously been developed for calculating monoisotopic masses of Sec-containing peptides,¹⁸ and dibromide tagging and tracing methodology.¹⁹ In order to use SESTAR to analyze the selenoproteome, lysates are analyzed by LC-MS/MS using standard data-dependent scanning. Next, the SESTAR analysis is performed to identify the Se-dependent isotopic signature in the MS1 spectra, through envelope extraction and pattern recognition steps. Typically, since selenoproteins often exist at low abundance in a lysate, it is unlikely that MS2 spectra have been generated for Sec-containing peptides. If so, a replicate sample can be analyzed using a targeted mass inclusion list based on the results of the SESTAR analysis. SESTAR analysis has been used to analyze existing datasets from the Human Proteome Map to generate an atlas of selenoproteins across human cell and tissue types. Typical database searching resulted in 67 identifications of selenoproteins across 23 tissues or primary cell types, while SESTAR generated 133 identifications. SESTAR was also applied to identify selenoprotein targets of the ferroptosis-inducing agent RSL3. Proteomics studies had previously identified Gpx4 as a target of RSL3, though the exact site of RSL3 modification on Gpx4 was not directly identified, and was instead extrapolated via mutagenesis.^20,21 SESTAR analysis of an RSL3-alkyne labeled sample enabled the direct detection of the selenocysteine of Gpx4 as the site of RSL3 reactivity. Furthermore, SESTAR identified additional selenoprotein RSL3 targets including Txnrd1, SelenoK, SelenoS, SelenoM, and SelenoI.¹⁷ SESTAR greatly improves selenoprotein detection from complex biological samples. However, abundant non-Se containing isobaric species could mask the selenium isotope signature, especially for low-abundant selenopeptides. SESTAR also does not provide information regarding the native oxidation state of the Sec residue within the biological sample. However, SESTAR can be used in conjunction with the reactivity-based selenoprotein enrichment methods below in order to address these limitations.

Low pH iodoacetamide reactivity

The reactivity of Sec residues can be exploited to enrich Sec-containing proteins from complex mixtures, serving to remove abundant non-selenoproteins from the sample that could suppress selenopeptide MS signals. Enrichment is also selective for reduced and reactive Sec, and can differentiate reduced and oxidized selenoproteins within a sample. Electrophilic probes, including iodoacetamide (IA), have been widely used to monitor Cys in a reactivity-dependent manner. This approach has enabled the monitoring of Cys reactivity, posttranslational modifications, and inhibitor sensitivity.^22–26 Sec is reactive to IA,²⁷ but Sec-containing peptides are rarely identified in IA-based proteomics studies.²⁸ This lack of detection is likely due to the low abundance of selenoproteins in the proteome. Sec has a lower pKa than Cys (approximately 5.2 and 8.5, respectively), suggesting that at a lower pH, the majority of Cys residues will be protonated and therefore non-reactive with IA, whilst Sec residues will remain in the deprotonated form and therefore reactive to IA. In fact, Sec has been shown to be reactive with IA at pH conditions as low as 2.²⁷ Therefore, IA labeling can be performed at a low pH (4–5.75) in order to target Sec over Cys, enrich labeled peptides, and analyze by LC-MS/MS for reactivity-based protein profiling (Figure 2D).

In the first example of low-pH IA labeling for proteomic selenoprotein detection, lysates were labeled with IA-alkyne at pH 5.75. A cleavable biotin tag was then attached to alkyne-labeled proteins via copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC). Labeled proteins were enriched via streptavidin beads and subjected to on-bead trypsin digestion. The IA-labeled peptides were then eluted from the beads and analyzed by LC/LC-MS/MS.²⁹ First, this technique was used to monitor selenoprotein expression across mouse tissues. Analysis of 5 mouse tissues and 1 mouse cell line (RAW264.7 macrophages) identified 14 of 24 mouse selenoproteins. These 14 selenoproteins comprised 12 of the 14 soluble selenoproteins (all but SelenoO and SelenoV) and 2 membrane selenoproteins (SelenoS and SelenoT). Second, a quantitative approach was utilized to verify the previously observed hierarchy of selenoprotein production under limited selenium concentrations, first reported with ⁷⁵Se labeling. Isotopically heavy and light IA-alkyne³⁰ was used to label lysates from cells that had been grown with or without exogenous Se supplementation. Consistent with previous reports, it was observed that some proteins, including Txnrd1, exhibited no Se-dependence, while other proteins, such as SelenoW, exhibited strong Se-dependence. Third, Sec reactivity was characterized; comparing the extent of IA-alkyne labeling in a concentration-dependent manner, enables ranking of Cys/Sec residues by reactivity.²² This design was replicated in low pH buffer “to rank selenoproteins by Sec reactivity, and it was found that Sephs2, Txnrd1, Gpx1, and Gpx4 contained the most reactive Sec residues. These reactive Sec residues were compared to Cys residues that were also labeled at low pH, and it was observed that all Sec residues identified fall in the most reactive quartile of the Cys residues identified. It can therefore be concluded that Sec residues have higher inherent reactivity relative to Cys at low pH. Low-pH IA labeling was also used to elucidate the targets of the drug Auranofin, which has been shown to inhibit the selenoprotein Txnrd. Auranofin binding was monitored by observing a loss of IA labeling. It was found that Auranofin was not selective for Trxnd, and potential other targets were identified, including the selenoprotein Gpx4 as well as Cys targets in Tpi1, Clic1, Ctsb, Iqgap1, and Txndc5.²⁹ Recently, this selenoprotein detection approach was used to monitor mis-incorporation of Sec at Cys-encoded residues in E. coli. Such mis-incorporation is due to the inability of the endogenous Cys aminoacyl tRNA synthetase (CysRS) to sufficiently distinguish between Cys and Sec. However, the CysRS from the Se-accumulating plant Astragalus bisulcatus has the ability to reject Sec, and mutation of E. coli CysRS so that it mimics the A. bisulcatus CysRS leads to a decrease Sec mis-incorporation.³¹

Similarly, low-pH labeling with an IA-PEG₂-biotin probe has been used to detect selenoproteins in mouse proteomes.³² First, this approach was used to generate an atlas of mouse selenoproteins, whereby 22 of the 24 mouse selenoproteins were successfully detected. The only selenoproteins not detected were Dio2 and Gpx3. Second, tandem mass tag (TMT)-based quantitation was used to monitor changes in selenoprotein levels in cultured BV2 cells, with and without reactive oxygen species (ROS), and with and without Sec supplementation. It was found that supplementing with Sec both increased selenoprotein abundance and protected against ROS, consistent with the antioxidant role of many characterized selenoproteins. Third, changes in selenoprotein enrichment in mice was monitored as a function of age. It was found that in the brain and heart, the levels of many enriched selenoproteins decrease with age. In the heart there is a decrease in SelenoM, SelenoN, SelenoK, Txnrd1, Txnrd2, SelenoP, and SelenoH; in the brain there is a decrease in SelenoK, SelenoT, SelenoP, and SelenoS. Finally, low-pH IA reactivity was used to identify candidate selenoproteins, irrespective of SECIS elements. It is important to note that the current selenoproteome was originally identified and defined by bioinformatic screens based on SECIS element identification.³³ By generating the SECIS-independent selenoprotein (SIS) database, where all UGA codons are translated to Sec and only UAA and UAG are stop codons, other potential sites of Sec incorporation could be identified. Peptides with high-confidence MS2 assignments that also contained the observable selenium isotopic envelopes were identified. Using this strategy, 5 candidate selenoproteins were identified: FXYD2 in kidney, ATP5B in kidney and liver, SCGB1A1 in lung, MUP and MT2 in liver.

While low-pH IA labeling has proven to be an effective tool for enrichment of selenoproteins containing reduced and reactive Sec residues, the requirement of non-physiological pH could likely influence endogenous protein structure, modifications, and function. This limitation could be overcome by the generation of Sec-specific electrophilic probes that bind Sec at physiological pH, and preferably directly within living cells. Fluorescent and colorimetric Sec-specific probes (dinitrophenyl ethers and sulfonamides) that label Sec at physiological pH have been generated and used for imaging and visualization studies,^34,35 but the ability of such reactive groups to selectively enrich Sec-containing species from complex samples has not been demonstrated. The development of Sec-specific chemical probes for proteomic studies would be of great benefit to selenoprotein research.

Selenoprotein expression methods

While there are useful tools for the study of selenoprotein expression and activity on the proteomic level, follow up of proteomic results is hindered by the difficulties in recombinantly expressing selenoproteins. As mentioned previously, the endogenous mechanism of Sec insertion makes traditional approaches of recombinant expression, such as transforming or transfecting cells simply with cDNA of selenoproteins inefficient. Therefore, a number of approaches have been developed to optimize or bypass the endogenous selenoprotein translation machinery.

In vitro selenoprotein production

Selenoproteins can be produced in vitro through Sec-mediated native chemical ligation (NCL) and expressed protein ligation (EPL) (Figure 3A), and these approaches have been reviewed in detail recently.³⁶ NCL and EPL more traditionally utilize Cys residues to form the amide bond between two protein fragments. In fact, Sec-mediated NCL and EPL techniques are more efficient than Cys-mediated ligation approaches. The Sec-containing fragments for in vitro selenoprotein production can be generated by solid phase peptide synthesis (SPPS), or expressed in E. coli. Expression of Sec-containing fragments in E. coli requires mutation of Sec codons to Cys codons, and culturing E. coli in the presence of chemically defined media containing L-Sec, forcing the mischarging of Cys tRNA with Sec. This approach leads to global mis-incorporation of Sec in the place of Cys. Recently, SelenoM has been synthesized independently by two groups, one using exclusively SPPS,³⁷ and the other using exclusively E. coli expression,³⁸ for fragment production. These in vitro approaches, however, are limited by the requirement for proper protein folding after ligation, as well as challenges associated with introducing these proteins into cells for cellular studies.

Figure 3.

Methods of selenoprotein expression. A) Native chemical ligation and expression protein ligation. B) Engineering of the E. coli endogenous Sec insertion machinery. C) Engineering of eukaryotic endogenous Sec insertion machinery. D) Genetic code expansion to incorporate caged Sec, and uncaging to generate selenoproteins.

Prokaryotic systems for selenoprotein expression

Prokaryotic model organisms, especially E. coli, are often used as recombinant protein expression systems. The endogenous selenoprotein mechanism, and especially the differences in the eukaryotic and prokaryotic Sec insertion mechanisms, make traditional recombinant techniques inefficient if not impossible. Many approaches have been developed to allow for SECIS element-independent incorporation of Sec in E. coli (Figure 3B). Details of successful strategies to avoid SECIS dependence in E. coli, through modification of tRNA and/or the Sec codon, are well summarized by a recent review.³⁹ These strategies to increase the efficiency of selenoprotein expression in E. coli include: (1) synthetic tRNAs that utilize the general elongation factor EFTu instead of Sec-specific SelB, thereby bypassing the SECIS element;^40–43 (2) a synthetic allo-tRNA, co-expressed with the Sec synthase of Aeromonas salmonicida, for efficient incorporation of multiple Sec residues in a single protein;⁴⁴ (3) engineering of E. coli to incorporate Sec at sense codons, which increases efficiency by eliminating competition with termination factors;⁴⁵ (4) recoding Sec to UAG codons and depleting E. coli of release factor 1,⁴⁶ and (5) recoding of the Sec codon followed by whole-genome engineering, to generate an E. coli strain with high Sec incorporation efficiency.^47,48

The Sec insertion and biosynthesis mechanisms can be bypassed entirely by the use of genetic code expansion (GCE) technology. Furthermore, the genetic encoding of a caged Sec residue provides the ability to regulate selenoprotein activity via uncaging. GCE has been used to incorporate a number of unnatural amino acids, using orthogonal engineered aminoacyl tRNA synthetase (RS)/tRNA pairs generated through directed evolution.^49,50 Two groups have utilized engineered orthogonal RS/tRNA pairs to incorporate a caged Sec residue into proteins in E. coli by suppressing the amber stop codon UAG. The caging group can then be removed to generate proteins harboring Sec (Figure 3D). The use of a caging group is a practical consideration, since evolving an RS able to orthogonally distinguish between Sec and Cys would be challenging due to structural similarity. Furthermore, given the high sensitivity of Sec to oxidation, the caging group provides protection from oxidative inactivation and allows for temporal control over Sec uncaging and therefore selenoprotein activity. These strategies avoid the endogenous Sec insertion machinery, including the synthesis of Sec on the tRNA, rendering selenoprotein production independent of Se concentration.

A photocaged Sec amino acid (PSc) was incorporated into proteins in E. coli using an engineered pyrrolysyl RS/tRNA pair. PSc was then successfully uncaged by UV irradiation (Figure 3D).⁵¹ However, this system has so far only been used to incorporate PSc into non-native selenoproteins where Sec is utilized as a substrate for site-specific protein labeling. This system has not been utilized to produce a native selenoprotein. An engineered pyrrolysyl RS/tRNA pair was also used to incorporate the amino acid Se-allyl Sec (ASec) into proteins in E. coli. ASec can then be uncaged in a palladium (Pd)-catalyzed reaction (Figure 3D).⁵² This approach was used successfully to express Gpx1, a native human selenoprotein, in the E. coli expression system. The SECIS element-independent and GCE-based techniques for selenoprotein in E. coli are beneficial for providing relatively large amounts of selenoproteins for in vitro evaluation, but prevent the study of eukaryotic selenoproteins in their native environment.

Eukaryotic systems for selenoprotein expression

Ideally, eukaryotic selenoproteins could be efficiently expressed in eukaryotic cell systems. It has been demonstrated that expressing selenoprotein cDNA, and including a non-native SECIS-element in the 3’ UTR of the mRNA, increases the efficiency of selenoprotein expression in mammalian cells (Figure 3C).⁵³ This approach makes use of the SECIS element that is natively found in Toxoplasma gondii, and has been used routinely to express selenoproteins such as SelenoS.⁵⁴ Furthermore, this method has been used to introduce Sec into non-native selenoproteins; for example, for incorporating Sec into roGFP in order to improve redox sensitivity.⁵⁵

As with the E. coli system, the entire Sec insertion mechanism in eukaryotic cells can be bypassed through the use of GCE. The first demonstration of incorporation of a caged Sec amino acid by genetic code expansion was the incorporation of the photoprotected 4,5-dimethoxy-2-nitrobenzyl (DMNB)-Sec amino acid in yeast using an engineered leucyl RS/tRNA pair (Figure 3D).⁵⁶ This system was used to incorporate DMNB-Sec into the model protein GFP, and was not used to express native selenoproteins. Additionally, yeast do not possess any native selenoproteins so yeast is a non-ideal organism for investigating endogenous selenoprotein function.

The ASec amino acid described previously has been successfully incorporated into proteins in mammalian cells and uncaged via Pd, yielding a Sec-containing protein.⁵² However, in mammalian cells this technology has been limited to the incorporation of ASec into the model protein GFP, and has not been used to express a native selenoprotein. Furthermore, this system requires the addition of exogenous Cys in order to mitigate ASec toxicity, and this addition will likely complicate any studies of cellular redox function mediated by the selenoprotein of interest. Recently, it has been demonstrated that DMNB-Sec can be incorporated into proteins in mammalian cells. Through photo-uncaging of DMNB-Sec, both GFP-Sec and the native selenoprotein MsrB1 have been generated.⁵⁷ Successful MsrB1-DMNBSec uncaging in living cells was demonstrated using the low-pH IA-labeling selenoprotein detection strategy.²⁹ Together, these studies utilize GCE based incorporation of caged Sec analogs to allow for Seindependent expression of native selenoproteins as well as temporal control of selenoprotein activity in mammalian cells.

A major limitation of the eukaryotic GCE platforms is the low yield of the target selenoprotein. Yields can likely be improved by further engineering of the RS/tRNA pairs used for incorporation. Optimized GCE approaches should enable the further study of both characterized and functionally unannotated native selenoproteins. Furthermore, the ability to robustly incorporate Sec site-specifically should be a great benefit in studying the function and advantage of Sec compared to Cys.

Conclusions

Selenoproteins are a relatively small but essential family of proteins. Family members that have been characterized mostly play a role in redox chemistry, however, a number of annotated selenoproteins remain poorly characterized. Methods for detection and expression of selenoproteins have the potential to further our understanding of selenoprotein function and the necessity for Sec incorporation over Cys. Recently, targeting of reactive selenocysteine residues with electrophilic probes, along with advances in computational selenium detection in proteomic samples, has significantly broadened the scope of questions that can be addressed in selenoproteomics. For example, these methods enable the identification of selenocysteine at non-SECIS sites across the proteome. Recombinant expression of selenoproteins is necessary to facilitate experimental follow-up on proteomic findings. Because of the complicated mechanism of endogenous selenoprotein expression, traditional experiments such as overexpression and mutant generation have been challenging to apply to the study of selenoproteins. Recently, GCE has enabled expression of selenoproteins in cells while avoiding the endogenous Sec incorporation pathway. Furthermore, the incorporation of caged Sec through GCE allows for temporal control over Sec activity, which could serve to accelerate studies into annotating the function of uncharacterized selenoproteins. We predict that advances in selenoprotein detection and expression together will enable a new wave of selenoprotein research.

Biography

Jennifer C. Peeler earned her bachelor’s degree at Franklin & Marshall College. As an undergraduate she developed tools for genetic code expansion in the laboratory of Ryan A. Mehl. She then earned her PhD from the Rockefeller University, using chemical biology tools to study G protein-coupled receptors in the laboratory of Thomas P. Sakmar. She is currently a postdoctoral fellow in the laboratory of Eranthie Weerapana at Boston College, using chemical biology approaches including proteomics and genetic code expansion to study cysteine and selenocysteine functionality.

Eranthie Weerapana is an Associate Professor of Chemistry at Boston College. She received her B.S. in Chemistry from Yale University, and her Ph.D. in Chemistry from MIT, where she worked with Professor Barbara Imperiali, investigating glycosyltransferases involved in N-linked glycosylation in the gram negative bacterium Campylobacter jejuni. She then performed postdoctoral studies at The Scripps Research Institute, La Jolla where she worked with Professor Benjamin F. Cravatt to develop chemical-proteomic methods to investigate reactive cysteines in complex proteomes. Her interdisciplinary research program focuses on applying mass-spectrometry methods to identify reactive and functional cysteine and selenocysteine residues in the human proteome, and chemical biology approaches to develop covalent small-molecule modulators for (seleno)cysteine-mediated protein activities.