Introducing Selenocysteine into Recombinant Proteins in Escherichia coli

Abstract

Selenoproteins contain the 21st amino acid, selenocysteine. Selenocysteine is the only amino acid that is synthesized on its cognate tRNA, and it is inserted at specific recoded UGA stop codons via a complex translation system. Although highly similar to cysteine, selenocysteine has unique properties, including a stronger nucleophilic ability and lower reduction potential. Efforts to site-specifically incorporate selenocysteine to create recombinant selenoproteins involve a recoded UAG stop codon and expression of the necessary selenocysteine translation machinery. This article presents a protocol for expressing and purifying selenoproteins in Escherichia coli. © 2021 Wiley Periodicals LLC.

INTRODUCTION

This article provides a general workflow and describes key issues to be considered in making recombinant selenoproteins in Escherichia coli. Selenocysteine (Sec, U), discovered after the genetic code (containing the 20 canonical amino acids) was initially deciphered, was found to recode a stop codon in specific instances (Labunskyy, Hatfield, & Gladyshev, 2014; Yoshizawa & Böck, 2009). Sec has been found in proteins throughout all three domains of life. It has a unique and complex translation pathway that makes its site-specific incorporation difficult, unlike the process for canonical amino acids (Fig. 1). The strategy described below reduces the complexity of the natural Sec incorporation pathway, yielding a system applicable to any protein of interest (details are discussed in the Background Information).

Figure 1.

Selenocysteine incorporation in bacterial cells. (A) Chemical structures of cysteine and selenocysteine, with the single difference between the structures highlighted. (B) Plasmid map of pSecUAG-Evol2 generated through SnapGene (available from the Addgene repository, cat. no. 163148). The plasmid encodes allo-tRNA^UTu2D (blue) as well as the necessary Sec machinery for incorporation using the rewired pathway. (C) In the native bacterial Sec incorporation pathway, tRNA^Sec is first serylated by SerRS and then converted to Sec-tRNA^Sec by SelA. SelB and the SECIS element bring Sec-tRNA^Sec to the ribosome to recode a UGA codon as Sec. (D) In the rewired Sec incorporation pathway, allo-tRNA^Sec follows the same aminoacylation steps as tRNA^Sec. The difference resides in elongation, when Sec-allo-tRNA^Sec is brought to the ribosome by EF-Tu to recode UAG codons as Sec.

As a result, this protocol is designed to be used both to recombinantly produce larger quantities of natural selenoproteins and to enable site-specific incorporation of Sec into a protein of interest. Cysteine (Cys)-to-Sec substitution mutations are of interest for numerous reasons (Metanis & Hilvert, 2014), with some of the more notable applications including investigating mechanisms of redox enzymes, potentially improving the catalytic rate of enzymes reliant upon Cys as a nucleophile, and increasing the strength of disulfide bonds. Chemically, Sec is highly similar to the canonical amino acid Cys, differing by only a single atom (Fig. 1A). As Cys and Sec have similar size, shape, and oxidation states, they can be substituted for one another without significantly perturbing a protein’s structure (Johansson, Gafvelin, & Arner, 2005; Reich & Hondal, 2016). Sec is the only naturally occurring amino acid that contains selenium, giving this residue a variety of unique properties. Compared to the thiol group in Cys, the selenol found in Sec has a lower reduction potential, which is thought to account for Sec’s natural role in a variety of redox enzymes (Reich & Hondal, 2016). Selenium is also more strongly nucleophilic than sulfur (Johansson et al., 2005), and there is evidence suggesting that Sec acts as a nucleophile within the active site of rat thioredoxin reductase (Brandt & Wessjohann, 2005). Sec further exhibits greater acidity than Cys due to its lower pKa (5.2 vs. 8.5), and is therefore is predominantly found in its ionized (deprotonated) form under physiological conditions, whereas Cys is predominantly protonated (Reich & Hondal, 2016). Lastly, diselenide bonds formed from Sec are stronger than disulfide bonds, and the rapidly formed diselenide bonds have also been shown to improve protein folding (Arai et al., 2017; Weil-Ktorza et al., 2019).

STRATEGIC PLANNING

When deciding to incorporate Sec into a protein of interest, it is imperative that this protein has an established expression and purification protocol that produces pure, active protein. The more knowledge available regarding protein assembly and necessary factors for activity, the better, as this information will be an asset when designing the expression platform. Although this protocol can be applied to the expression of any recombinant protein, there are specific strategies that facilitate the purification of full-length selenoproteins, which are discussed below.

Choosing a compatible expression plasmid

All Sec incorporation machinery needed for expression is provided on a single plasmid (pSecUAG-Evol2; Mukai, Sevostyanova, Suzuki, Fu, & Söll, 2018; Fig. 1B). For successful co-transformation with more than one plasmid in a cell, it is important to confirm that the two plasmids are compatible with regard to antibiotic resistance, inducibility, and origin of replication. pSecUAG-Evol2 is kanamycin resistant and induced with arabinose, and gene expression levels have been optimized with its RSF origin of replication.

Designing an efficient protein expression platform for purification

When expressing selenoproteins, the position of the affinity tag for purification will affect the amount of downstream processing necessary for clean product. If a UAG (stop) codon is used for Sec incorporation, translation does not always go to completion; instead the UAG can be read as a stop, terminating translation, or the processivity may be reduced so that translation stops shortly after incorporation of the Sec. This results in many truncated products that will contaminate the final selenoprotein. Therefore, the most ideal location for an affinity tag on a single-subunit protein is at the C-terminus (Fig. 2). This immediately removes all truncated products, which are unable to translate the affinity tag. Unfortunately, not all proteins can be actively expressed or purified with a C-terminal tag. If this problem arises, an N-terminal tag can be used; however, additional purification steps such as size-exclusion chromatography will be necessary to remove any truncated proteins. Depending on the protein size and the point of truncation (which typically occurs around the Sec incorporation site), separation based on size may not yield a clean product.

Figure 2.

Choosing an optimal position for an affinity tag. (A) Schematic of the process to strategically tag the selenoprotein for maximum yield. The number of protein subunits and the positioning of the Sec residues incorporated will affect where the affinity tag should be placed. The most optimal positioning is designated as 1, with subsequent options being less efficient. (B) Diagram describing the effect of the affinity tag position on downstream processing to achieve a clean final product.

This strategy is also valid for multi-subunit proteins: preferentially have a C-terminal tag on the subunit with the Sec incorporated, otherwise aim for an N-terminal tag (Fig. 2). Because of the complexity of multi-subunit protein assemblies, it may not be possible to place a tag anywhere on the Sec-containing subunit, especially if it is at the core of a multi-subunit assembly. In such cases, the next best option is to place the tag on the C-terminus of the last protein expressed that interacts with the Sec-containing subunit (this may not correlate to the last protein expressed in the plasmid if there are helper proteins present). This strategy should also result in full-length selenoprotein with some possible contamination of only the tagged protein (provided it does not associate with any other subunit in the absence of the Sec-containing subunit). If there is enough difference in size between the protein contaminant(s) and the selenoprotein, size-exclusion chromatography should suffice to purify out the full-length selenoprotein. A tag can also be placed on the C-terminus (or N-terminus) of other subunits; however, this results in more complications regarding the various products that could be obtained from affinity purification. With enough information about the multi-subunit assembly and appropriate downstream purification protocols, however, a tag could be strategically placed on any subunit.

Further considerations must be taken into account if multiple Sec residues are to be incorporated into a multi-subunit protein. If all the Sec residues are in one subunit, then the same strategies discussed above are appropriate. When the Sec residues are dispersed throughout more than one subunit, however, this becomes more complicated. If it is known that one of the Sec-containing subunits is expressed at a lower yield than the rest in the wild-type protein, then the tag should be located on the C-terminus of that subunit (Fig. 2). If all subunits are expressed equally, then the last Sec-containing subunit that is expressed should contain a C-terminal tag.

It is highly recommended to optimize the affinity tag position before beginning selenoprotein production to aid in the detection and purification of pure protein.

Consideration of contamination and protein yields

Because UAG is used for Sec incorporation, it cannot be used as a stop codon for any of the proteins expressed. Therefore, prior mutagenesis should be performed to remove any UAG codons from the expression plasmids, replacing them with UGA or UAA.

As shown in Figure 1C, incorporation of Sec first requires the tRNA to be serylated by seryl-tRNA synthetase (SerRS) before the sulfur moiety is replaced by a selenium through SelA. If SelA does not perform this conversion efficiently, or there is not enough selenium precursor, then it is possible that Ser will be incorporated at the UAG instead of Sec. In many cases this contamination can be kept quite low (<1%); however, this may not be possible in every situation. Therefore, it is necessary to consider whether Ser contamination will be a problem in further studies, or whether downstream processing can separate the Ser-containing protein from the Sec-containing protein.

Furthermore, the overall protein yield will be significantly less than what is obtained with wild-type proteins (before any modifications). In best-case scenarios, one can obtain a 25% yield with a single Sec substitution; however, it can be as low as 5%, depending on the system. The yield decreases further with multiple Sec substitutions and multiple plasmids (more than two). This loss in yield may not be an issue for high-expressing proteins, but when wild-type protein expression is already low, significant culture volumes may be required to obtain the selenoprotein yield desired.

For projects seeking to incorporate a single Sec residue into a protein of interest, standard E. coli expression strains such as BL21(DE3) are sufficient. However, when seeking to insert two or more Sec residues into a protein, premature termination becomes increasingly problematic. In such instances, use of a recoded E. coli strain that does not contain release factor 1 (RF-1), such as C321.ΔA.exp or B-95.ΔAΔfabR (Mukai et al., 2015), is preferred as this reduces premature protein termination.

BASIC PROTOCOL RECOMBINANT SELENOPROTEIN PRODUCTION IN E. coli USING A REWIRED TRANSLATION SYSTEM

Once the protein of interest is designed with the affinity tag in the most optimal position for purification and is actively expressed in high yield (see Strategic Planning), it is ready to be used for selenoprotein production. This protocol describes in detail the method used to site-specifically incorporate Sec into a protein of interest using technology engineered in our lab (Mukai et al., 2018). One plasmid (pSecUAG-Evol2) encodes the machinery necessary for site-specific incorporation of Sec at a UAG codon. The pSecUAG plasmid provides kanamycin resistance and is arabinose inducible, encoding SelA^Evol and SelD from Aeromonas salmonicida, a Sec-containing thioredoxin from Treponema denticola, and allo-tRNA^UTu2D (see Background Information for details on the development of this plasmid).

Materials

PfuUltra II Fusion High-fidelity DNA Polymerase (Agilent cat. no. 600670)

10× PfuUltra II Fusion Buffer (supplied with Agilent cat. no. 600670)

dNTP mix (10 mM each)

Forward and reverse QuikChange primers (see recipe for Primers)

IPTG-inducible protein expression plasmid (compatible with RSF origin of replication and kanamycin resistance; extracted from a dam⁺/dcm⁺ cell line)

Nuclease-free water

Agarose

Gel buffer (see recipe)

Ethidium bromide or SYBR™ Safe (Thermo Fisher Scientific cat. no. S33102)

DNA gel loading dye (homemade or commercial)

DNA ladder (1 kb plus)

PCR cleanup kit (optional)

DpnI restriction endonuclease

Competent DH5α cells (or equivalent), for cloning

Growth medium (commercially available Luria Broth [LB] or Super Optimal broth with Catabolite repression [SOC] are recommended)

LB agar plates with appropriate antibiotic(s)

Plasmid DNA miniprep kit

pSecUAG-Evol2 plasmid (Addgene cat. no. 163148; Mukai et al., 2018)

Electrocompetent cells: ME6 (Mukai et al., 2016), C321.ΔA.exp (Addgene cat. no. 49018), or B95.ΔAΔfabR (Mukai et al., 2015), for protein expression

Antibiotics for protein expression plasmid (sterile filtered using a 0.2-μm-pore-size filter and syringe)

50 mg/ml kanamycin (sterile filtered as above)

20% (w/v) arabinose solution (sterile filtered as above)

100 mM sodium selenite solution (sterile filtered as above)

1 M isopropyl β-d-1-thiogalactopyranoside (IPTG; sterile filtered as above)

0-10 μl, 20-200 μl, and 200-1000 μl micropipets

Appropriate pipet tips (sterilized)

0.5-ml thin-wall PCR microtubes

PCR machine

Agarose gel running apparatus

Standard microwave, for melting agarose

Culture tubes

Microcentrifuge

1.5-ml microcentrifuge tubes

0.2-μm-pore-size sterile filters

Syringes

Autoclave

NanoDrop spectrophotometer (or equivalent), for checking DNA concentration

Electroporation cuvettes

Electroporator

37°C incubator

Floor shaker

Expression vessel

Additional reagents and equipment for SDS-PAGE (Gallagher, 2006)

Mutate codon in desired position to TAG

1. Set up a 25-μl reaction in 0.5-ml thin-wall PCR microtubes in the following order (adapted from the manufacturer’s protocol):

10× PfuUltra II Fusion buffer	2.5 μl (1×)
10 mM dNTP mix	0.6 μl (240 μM)
10 μM forward primer	2.5 μl (1 μM)
10 μM reverse primer	2.5 μl (1 μM)
100 ng protein expression plasmid	0.2-5 μl
PfuUltra II Fusion High-Fidelity DNA polymerase	1 μl
Nuclease-free water	Up to 25 μl.

Run PCR as follows:

Initial step:	3 min	95°C	(denaturation)
25 cycles:	30 s	95°C	(denaturation)
	30 s	50-65°C*	(annealing)
	1 min/kb**	72°C**	(extension)
Final step:	5 min	72°C	(extension).

^*The annealing temperature will depend on the melting temperature of the primers (see, e.g., https://www.sigmaaldrich.com/technical-documents/articles/biology/oligosmelting-temp.html). For best results, set up multiple PCRs at different melting temperatures and choose one that provides a single clean band on the gel.

^**If the plasmid is >10 kb, increase extension time to 2 min/kb and reduce temperature to 68°C.

• 2.While waiting for the PCR to complete, measure out enough agarose for a 1% (w/v) gel. Add the appropriate amount of gel buffer and microwave the solution until it is bubbling and agarose has dissolved. Add 0.25 μg/ml ethidium bromide or add SYBR Safe™ to 1×, and pour gel into the casting equipment with an appropriate comb.
• 3.Mix 5 μl of the PCR product with DNA gel loading dye and nuclease-free water. Prepare the DNA ladder in the same way with loading dye (if ladder is not obtained ready to use). Load samples and ladder onto agarose gel from step 2, and fill gel running apparatus with enough gel buffer to cover the gel. Run the gel at 140 V for 20-25 min.

  A 1% agarose gel run for 20-25 min will allow visualization of most plasmids ranging from 1-10 kb. If the expression plasmid is larger, the gel can be made with a lower percentage of agarose or run a bit longer to confirm that the band observed is the correct size. Observation of a single band at the correct size signifies a successful PCR.

• 4.Add 1.7 μl DpnI to the remaining 20 μl of PCR product. Incubate at least 1 hr at 37°C (can be left overnight).

  To avoid evaporation, it is best to add the DpnI directly to the PCR tube and then place the tube in a 37°C incubator or use the incubation feature on the PCR machine, rather than using a heat block. This step removes any of the parent plasmid that is methylated, leaving only the product, which should contain the mutation of interest. If the expression plasmid used as a template for the PCR was extracted from dam⁻/dcm⁻ cells, this step will not work: Colonies produced from such a transformation will be a mixture of original template and mutated product.

• 5.Add 3-10 μl of the digested PCR product to chemically competent DH5α cells (or equivalent) on ice and incubate for 30 min. Heat shock the cells for 30-45 s at 42°C, and place them back on ice for 5 min. Add 1 ml LB or SOC medium and transfer to 37°C with shaking for 1 hr to allow the cells to recover. Spin down the recovered cells and resuspend in 100 μl LB. Plate the entire 100 μl of transformed cells onto an LB agar plate with the appropriate antibiotic for the expression plasmid. Incubate overnight at 37°C.

  Typically, 5 μl of digested PCR product will yield sufficient colonies to screen. If colonies are not observed after 24 hr, the transformation can be repeated with more PCR product. An alternative is to use a PCR cleanup kit after DpnI digestion in step 4 to remove the digested plasmid and quantify the mutated product using the NanoDrop. From knowing the concentration of mutated product, the appropriate volume can be transformed to equal 100 ng of DNA. With plasmids close to 10 kb or larger, transformation efficiency is further reduced due to plasmid size. Therefore, transformation should be done with higher-efficiency cells (e.g., Stellar cells) and/or with electrocompetent cells.

• 6.Inoculate a single colony into 3-5 ml LB with antibiotic in a culture tube. Repeat for a total of three or four colonies. Grow 12-18 hr at 37°C in a shaker.
• 7.Spin down the cells and follow the plasmid miniprep kit manufacturer’s protocol for purifying DNA. Measure the DNA concentration using a NanoDrop spectrophotometer (or equivalent).

  Cells can be spun down all at once at 3000 × g for 10 min. Alternatively, cells can be transferred to microcentrifuge tubes and 1-ml aliquots can be spun down in a microcentrifuge at full speed for 30 s. The latter yields more DNA; however, the amount obtained through spinning down the cells all at once is usually sufficient for sequencing and transformation.

• 8.Send samples for sequencing with appropriate sequencing primer.

  The sequencing primer should be ≥100 bp upstream of the TAG mutation and be designed according to the rules designated by the sequencing facility. The primer is typically 20 bases in length and will have an approximate GC content of 50% and a melting temperature of 55-60°C. In addition, the primer will be specific for each protein, depending on where the TAG mutation is located.

Transform expression plasmid with pSecUAG-Evol2 for expression

• 9.Transform 100 ng pSecUAG-Evol2 plasmid with 100 ng of expression plasmid containing successful TAG mutation (from step 8) into cells for expression. Allow cells to recover in 1 ml LB for 1 hr at 37°C before plating 100-200 μl on LB agar with appropriate antibiotics.

  Use of electrocompetent cells for transformation is recommended to increase efficiency. Standard expression cell lines are suitable for selenoprotein production; however, optimized cell lines with selABC and fdhF removed, as in ME6 (Mukai et al., 2016), or with some of the UAG stop codons recoded and RF-1 removed, as in B-95.ΔAΔfabR (Mukai et al., 2015), tend to produce higher protein yields. Our optimized cell line has the natural Sec machinery from B-95.ΔAΔfabR removed (B-95.ΔAΔfabRΔselABC). Use of a recoded strain is particularly recommended for projects seeking to incorporate multiple Sec residues into a protein of interest, as the deletion of RF-1 in these strains reduces issues arising from premature protein termination. If few to no colonies are observed, incubating the LB agar plate at 37°C before plating or plating a larger number of cells may help colony formation.

Expression of selenoprotein

• 10.Pick a colony from the transformation plate, and prepare an overnight culture with LB and antibiotics as would be done for expression of the wild-type protein.
• 11.The next day, prepare an appropriate vessel containing the volume of medium required for expression. Be sure to include all necessary antibiotics and medium additives for cell growth (not protein expression), as well as 50 μg/ml kanamycin, 0.1%-0.2% arabinose, and 5-10 μM sodium selenite.

  Concentrations of arabinose and sodium selenite used should be optimized for each individual protein.

• 12.Inoculate prepared medium with the overnight culture at the desired dilution (typically a 1/100 dilution is used; however, lower is also acceptable). Grow cells aerobically at 25°C-37°C for 8-24 hr until the desired cell density is achieved. A NanoDrop spectrophotometer (or equivalent) can be used to measure the cell density at OD₆₀₀.
• 13.Add IPTG (40 μM-1 mM) to induce protein expression. Express protein at desired temperature and conditions for 24-48 hr.

  It is necessary to first induce expression of the Sec machinery before inducing protein expression. This allows the accumulation of selenocystenylated tRNA to be incorporated at the UAG codon once protein expression is induced. Although inducing expression at an OD₆₀₀ = 0.4-0.6 yields the highest protein yields for normal proteins, this may not hold true for the Sec versions. Growth for an extended period allows sufficient incorporation of Sec into the protein. Step 11-13 need to be optimized for each protein to achieve maximum yield. Additional sodium selenite can be added at this point if Ser contamination is high.

• 14.Pellet cells by centrifugation and store cell pellet at −80°C until ready to purify.

  In situations where the cells need to be kept anaerobic (see annotation to step 15), be sure to follow the necessary protocols for storage of the cell pellet.

Purify the selenoprotein

• 15.Purify protein using the same protocol as for the wild-type protein.

  As mentioned in Strategic Planning, protein yields will be significantly lower than for the wild-type protein. Therefore, if there are activity assays or other methods to check for the presence of active protein (or the protein in general), consider increasing the amount used by 10- to 100-fold before ruling that no selenoprotein was produced. Otherwise, it may be beneficial to wait until the protein is fully purified before confirming successful expression. Furthermore, insertion of Sec can make enzymes oxygen sensitive, and therefore additional precautions (or additives) may have to be used to retain enzyme activity (e.g., working in an anaerobic chamber or adding KNO₃ to the buffers).

• 16.Run SDS-PAGE gel to confirm protein is present and pure.

Confirm incorporation of Sec by mass spectrometry

• 17.Cut out gel band for digestion and liquid chromatography–tandem mass spectromtry (LC MS/MS) analysis, or send protein solution for intact mass analysis.

  Preparation of material for analysis will be outlined by the facility used. The selenol group is highly reactive and therefore should be alkylated to protect it before analysis (Peeler & Weerapana, 2019). Besides the expected mass change, selenocysteine incorporation can be further confirmed by examination of the isotopic envelope observed in peptide MS data (see Fig. 3).

Figure 3.

Sample data. (A) Example of intact mass spectrometry data obtained using time-of-flight mass spectrometry. A single peak is observed for the protein of interest that indicates complete incorporation of Sec at the desired position. A gel insert shows that the protein was not clean, explaining the additional contaminating peak at 47,018 Da. (B) Example of the isotopic envelope observed from the mass spectrometry data of a peptide containing Sec. This differs considerably from (C) the isotopic envelope of the same peptide containing Ser instead. (Generated by P R. Baker and K. R. Clauser using http://prospector.ucsf.edu.)

REAGENTS AND SOLUTIONS

Gel buffer (1×)

40 mM Tris·HCl

2 mM EDTA

5 mM sodium acetate

Adjust to pH 8 with acetic acid

Buffer can be kept at room temperature for an extended period.

Primers

To mutate bases in the expression plasmid to the TAG codon for Sec incorporation, two primers are needed that each carry the mutation and are complimentary to the rest of the plasmid. Forward and reverse primers should be designed to have a minimum of ten complimentary bases 5′ of the TAG codon and 15 bases 3′ of the TAG codon. Ensure that both primers begin and end on a G or C base to enhance binding to the template. Try to keep the melting temperatures of the two primers within 5°C of one another, if possible.

COMMENTARY

Background Information

Bacterial selenoprotein production

In nature, protein incorporation of Sec occurs through a highly restrained process, which until recently has limited investigations into this amino acid. Sec incorporation begins with its biosynthesis, which uniquely occurs upon its tRNA. In the first step of Sec biosynthesis, Ser is ligated to tRNA^Sec (also known as selC) by SerRS. In bacteria, Ser is converted to Sec by the endogenous protein SelA. Next, the aminoacylated Sec-tRNA^Sec is recognized by a specialized elongation factor, of which the variant endogenous to bacteria is termed SelB. This elongation factor is specific for Sec-tRNA^Sec, ensuring that its biosynthetic precursor Ser is not incorporated at targeted insertion sites. SelB mediates Sec insertion at opal (UGA) stop codons that lie upstream from a specialized RNA motif known as the SElenoCysteine Insertion Sequence (SECIS) element (Labunskyy et al., 2014; Yoshizawa & Böck, 2009; Fig. 1C). SECIS elements found in nature form a convergent mRNA secondary structure that helps recruit SelB and mediate correct translation of Sec, despite exhibiting divergent sequences. Although it is possible to mediate bacterial selenocysteine insertion at new positions by adding a SECIS element sequence at the 3′ untranslated region (3′ UTR), this approach is generally unreliable as the SECIS element sequence in bacteria is typically encoded within the protein coding sequence. Thus, the SECIS element cannot readily be inserted at new positions of interest without significantly altering the protein sequence. In laboratory strains of E. coli (including BL21 and K12 derivatives), formate dehydrogenase H (FdhH) is the only endogenous selenoprotein, within which the SECIS element encodes amino acid residues 141-153. As tRNA^Sec is not recognized by other elongation factors, efforts to insert Sec at new positions that do not encode this residue in nature are limited by the firm requirement for the SECIS element.

In recent years, the bacterial Sec insertion pathway has been successfully engineered to utilize the general elongation factor EF-Tu instead of SelB (Fig. 1D). In the first effort, a tRNA^Ser was engineered to serve as a substrate for SelA and to mediate Sec incorporation programmed by UAG in the absence of the SECIS element (Aldag et al., 2013). This tRNA, termed tRNA^UTu, was further optimized to act as a better substrate for SelA, yielding tRNA^UTuX (Miller et al., 2015). In a separate study, tRNA^Sec was engineered by directed evolution to become a substrate for EF-Tu, but not of SelB, producing tRNA^SecUX (Thyer, Robotham, Brodbelt, & Ellington, 2015). Although tRNA^UTuX and tRNA^SecUX originated from separate scaffolds, they both contain a 13-bp acceptor branch (comprising the acceptor stem plus the T-stem; Miller et al., 2015; Thyer et al., 2015). Though such 13-bp acceptor branches are efficiently recognized by SelA, both tRNAs are relatively poor substrates for EF-Tu, which normally recognizes tRNAs with a 12-bp acceptor branch. However, a bioinformatic search identified a SelA variant from A. salmonicida (As SelA) that efficiently recognizes tRNAs with a 12-bp acceptor branch (Mukai et al., 2018). Furthermore, a metagenomic allo-tRNA variant (allo-tRNA^UTu2D) was isolated that carried identity elements for both E. coli SerRS and As SelA. To ensure sufficient selenium donor levels, A. salmonicida SelD (As SelD) and the T. denticola (Td) Sec-containing thioredoxin (Trx1) were also cloned (Mukai et al., 2018). Together, these four genes (As SelA, allo-tRNA^UTu2D, As SelD, and Td Trx1; cloned together in the pSecUAG-Evol2 plasmid) provide the most efficient EF-Tu-dependent Sec insertion system described to date (Mukai et al., 2018). It has been used for the expression of a variety of hydrogenases and dehydrogenases (unpublished data from our group) and a class I ribonucleotide reductase (Greene, Stubbe, & Nocera, 2019).

Mammalian selenoprotein production

The selenocysteine incorporation pathway in mammalian cells is more complicated than the bacterial system, involving additional components for Sec insertion (Howard & Copeland, 2019; Labunskyy et al., 2014; Yoshizawa & Böck, 2009). As in bacteria, mammalian tRNA^Sec is first charged by SerRS with Ser. However, the conversion to Sec-tRNA^Sec is a two-step process: Ser-tRNA^Sec is phosphorylated by O-phosphoseryl-tRNA kinase (PSTK) to phosphoseryl-tRNA Sec, which is then converted to Sec-tRNA^Sec by SepSecS. Efficient decoding of the UGA codon requires a specialized elongation factor (EF-Sec) and SECIS element. The mammalian SECIS element is mostly located in the 3′ UTR and can direct Sec-tRNA^Sec to multiple UGA codons. For instance, the human selenoprotein P mRNA contains ten UGA codons and two SECIS elements in its 3′ UTR. Additional components are required in the mammalian Sec pathway, many of which are still unknown. These complexities make engineering any part of the mammalian pathway difficult.

A few synthetic (man-made) strategies for making and interrogating selenoproteins in mammalian cells have been reported (Peeler & Weerapana, 2019). In one, the endogenous selenoprotein synthesis apparatus is supplied with additional selenite in order to overexpress natural selenoproteins whose gene, along with the 3′ UTR, is cloned into an expression plasmid (Kim & Gladyshev, 2005; Li et al., 2018). This can also be used as a route to new selenoproteins by mutating the site of interest to TGA in a gene that is placed in front of a 3′ UTR that contains a SECIS element (Novoselov et al., 2007; Stanford, Ajmo, Bahia, Hadley, & Taylor-Clark, 2018). Plasmids encoding a gene followed by a 3′ UTR containing a SECIS element are commercially available (e.g., LUC01 (L/Sec/P), Addgene cat. no. 112144; Mehta, Rebsch, Kinzy, Fletcher, & Copeland, 2004). The encoded gene can be easily replaced with the gene of interest bearing a UGA codon at the desired position for Sec incorporation (Stanford et al., 2018).

More recently, genetic code expansion strategies have inspired mammalian selenoprotein synthesis; orthogonal aminoacyl tRNA synthetases, orthogonal tRNAs that recognize the UAG codon, elongation factor EF-1α, and caged Sec analogs are used in these approaches. Selenium-allyl Sec (ASec) and 4,5-dimethoxy-2-nitrobenzyl (DMNB)-Sec have been site-specifically incorporated and uncaged in HeLa and HEK293T cells, respectively (Liu et al., 2018; Peeler et al., 2020). Such studies show promise for development into synthetic tools.

Alternative selenoprotein production methods

Although the above protocols involve the biological production of custom selenoproteins in various hosts, selenoproteins can also be produced using organic synthesis. Two studies used peptide synthesis to generate seleno-insulin variants (Arai et al., 2017; Weil-Ktorza et al., 2019). However, custom peptide synthesis remains limited to relatively small peptides, as this method becomes increasingly more challenging with the addition of amino acids.

Critical Parameters

First and foremost, UAG cannot be used as a stop codon for any proteins expressed using this protocol. UAG is reserved for incorporation of Sec and will cause problems as proteins that are supposed to terminate instead continue translating due to the incorporation of a Sec residue. As discussed in Strategic Planning, the position of the affinity tag is crucial to how easily clean product will be obtained. Given the small yields obtained through selenoprotein production, it is advisable to reduce the downstream processing as much as possible to obtain clean product. In some cases this may not be possible, and then larger culture volumes will be required to obtain the desired yields. This protocol can be applied to virtually any protein; however, the amount of optimization required to obtain significant yields may differ from one protein to another.

Troubleshooting

Inefficient transformation of expression plasmid and pSecUAG-Evol2

After transformation, it is possible that colonies will not grow. This is due to poor transformation efficiency, which can be optimized in several ways. These include transforming freshly made (not pre-frozen) competent cells, using a different electroporation cuvette (0.2 vs. 0.1 cm), using a richer medium for recovery (such as SOC), increasing the length of the recovery phase, and/or increasing the amount of cells plated. Furthermore, plating on a prewarmed plate will also promote cell growth.

Insufficient selenoprotein expression

As mentioned earlier, selenoproteins expressed with this protocol are produced in significantly lower abundance than wild-type protein. The exact yield is protein dependent, and further parameters may need to be optimized to try and increase yield. Although B-95.ΔAΔfabR is mentioned as the preferred cell line, it may not be optimal for producing maximal yields of all selenoproteins, and other cell lines should be tested in the case of unsatisfactory yield. Furthermore, the pSecUAG-Evol2 tRNA system may not be the best option. Currently allo-tRNA^Utu2D (pSecUAG-Evol2) is the tRNA found to incorporate the highest amount of Sec into proteins; however, using other systems (Evol 1, 3, and 4) that contain different promoters may increase selenoprotein production (Mukai et al., 2018). When optimizing the expression protocol, it should be noted that sufficient time must be given to allow a build-up of the available tRNA pool. The temperature and duration of this step require optimization. The amount of sodium selenite added to the medium will also affect protein yields: too much sodium selenite is toxic to the cells, whereas insufficient amounts will cause Ser contamination. Dosing with sodium selenite has been a desirable method to ensure a large enough pool of selenium donor without significantly affecting cell growth.

Understanding Results

Once the selenoprotein is purified, the best way to confirm whether Sec is incorporated is to analyze the protein by mass spectrometry. Intact mass analysis will help quantify how much of the protein contains Sec and how much is contaminated with Ser. A single peak at the correct mass signifies complete Sec incorporation (Fig. 3A). If a second peak is seen with a mass corresponding to Ser instead of Sec, complete Sec incorporation did not occur. Further analysis using enzymatic digestion followed by tandem mass spectrometry will determine the exact position of the Sec residue. These data can also be verified for Sec incorporation by the isotopic envelope of the peak(s) of interest. Ser and Sec have isotopic envelopes that differ from one another, allowing Ser contamination to also be observed in this way (Fig. 3B and C). More details on this topic as well as other ways to detect Sec incorporation have been reviewed in detail elsewhere (Peeler & Weerapana, 2019).

Time Considerations

Initial mutagenesis of the expression plasmid to contain the TAG codon at the desired position will take ~3-4 days. Additional time will be required if the initial mutagenesis is unsuccessful or if additional TAG codons are required. Protein expression, from transformation to storage of the cell pellet, can take 4-5 days, depending on the duration of protein expression. The time for protein purification and confirmation of Sec incorporation will vary depending on the protein-specific purification protocol and on access to a mass spectrometry facility. Please note that this is a general timeline that does not include time to prepare materials (e.g., autoclave medium, make buffers) and optimize each step in the protocol (Fig. 4).

Figure 4.

General timeline and order of steps required to produce selenoproteins.