The 2A Story: The End of the Beginning

Abstract

Translational control of viral gene expression is a fundamental process essential for the vitality of all viruses. In special cases, signals encoded in the mRNA reprogram the ribosome to read the message in a different way, a process termed “translational recoding”. The 2A region of the foot-and-mouth disease virus (FMDV) encodes a short sequence, only 18 amino acids, that mediates self-processing by a novel translational effect “ribosome skipping” rather than proteolysis. Briefly, 2A interacts with the ribosome exit tunnel to inhibit peptide bond formation at the C terminus of the 2A sequence. Translation terminates at this point, but then resumes elongation, creating a second independent protein product. Thus, discrete proteins can be produced from a single transcript. The 2A sequence is particularly useful in vector strategies (AAV and retroviral vectors) where the capacity to incorporate foreign DNA is limited. Use of 2A and “2A-like” peptides to link the sequences encoding several proteins in the same open reading frame has led to their increasing use as important tools in biotechnology and biomedicine. This technology has been crucial for the visual tracking of expressed proteins, human gene therapies targeting cancer, production of induced human pluripotent stem cells for regenerative medicine, creation of transgenic animals and plants and the improvement of CRISPR-Cas9 and TALEN genome editing methods.

1. Introduction

As obligate intracellular parasites with relatively limited coding capacity, viral strategies to sequester the host translation machinery target the initiation, elongation, and termination steps of protein synthesis. Although most translational control strategies operate at the rate-limiting initiation step, different regulatory mechanisms target elongation and termination. Commonly these involve translational control or “translational recoding” defined as instances in which “the rules for decoding are temporarily altered through the action of specific signals built into the mRNA sequences” [1, 2]. The two most common types of recoding elements program ribosomes to either shift the reading frame of an mRNA, typically by one base in either the 5′(−1) or the 3′(+1) direction (programmed ribosomal frameshifting) or reassign the “sense” of stop codons (UAA, UAG or UGA) leading to stop codon “Read-Through”. Less common are recoding elements that cause ribosomes to bypass defined segments of mRNAs (ribosomal “hopping”) or those that skip the synthesis of a specific peptide bond (ribosomal “skipping”). Ribosomal skipping, mediated by a “self-cleaving” 2A oligopeptide that manipulates the ribosome to “skip” the synthesis of the glycyl-prolyl peptide bond at its own carboxyl terminus leading to release of the nascent protein and translation of the downstream sequence. In this chapter we outline the development of this efficient methodology for the co-expression of multiple proteins from a single gene using a small 2A peptide originally identified in viruses.

2. Ribosome “skipping”

Among the five distinct types of picornavirus 2A proteins, a group of oligopeptide sequences collectively known as Aphthovirus-like 2A mediate a translational recoding event in cis known as “ribosome skipping”, “StopGo” or “Stop Carry-on” translation [3, 4, 5, 6]. Here a termination event occurs at a sense codon, followed by release of the nascent polypeptide and ribosomal translocation to the next in-frame codon. The function of the 2A sequence was first characterised from the positive-stranded RNA picornavirus Foot-and-Mouth Disease Virus (FMDV). The 2A oligopeptide is only 18 amino acids (aa) long (-LLNFDLLKLAGDVESNPG-), delineated by 3C^pro post-translational cleavage at its N-terminus and co-translational “cleavage” at its C-terminus [7, 8, 9]. The core sequence at the C-terminus of 2A is strongly conserved and contains the canonical motif D¹²(V/I)E(S/T)NPG_2A^↓P¹⁹_2B (where P¹⁹ refers to the completely conserved first residue of the downstream FMDV protein 2B). The less conserved part of the 2A sequence, located upstream of the motif, appears to be essential for 2A function [9, 10]. During this period the number of genome sequences was limited, but it was apparent that a similar motif was also present at the C-terminus of the longer cardiovirus 2A protein [11]. Analysis of recombinant FMDV polyproteins indicated FMDV 2A, together with the N-terminal proline of protein 2B, appeared sufficient for a co-translational “cleavage” between the upstream (capsid proteins) and downstream (RNA replication protein) domains of the polyprotein (-NPG^↓P-) [7, 12]. Following cleavage, the 2A “tag” remains as a C-terminal extension of the upstream protein. The single proline at the N-terminus of the downstream protein confers a long half-life (>20 h) which should help increase protein stability [13]. Early studies showed the FMDV 2A region was not simply a substrate for a virus proteinase (L^pro, 3C^pro), nor a substrate for a host-cell proteinase—“self-cleavage” was a novel translational effect [3, 7, 12]. Due to its mode of action within the ribosome the acronym CHYSEL (cis-acting hydrolase element) was proposed as an alternative name for 2A and 2A-like sequences to avoid confusion with protease mediated cleavage [14].

To monitor gene expression, artificial “self-processing” polyproteins were constructed with these 18 codons, plus the first aa (proline) of 2B (referred to as “2A”), separating green fluorescent protein (GFP—stop codon deleted) and β-glucuronidase (GUS) reporter-encoding sequences in a single open reading frame (ORF) encoding [GFP2AGUS] (Figure 1, pSTA1) [3, 15, 16]. In the absence of all other FMDV proteins, such artificial polyproteins comprised well-characterised, metabolically stable “reporter” proteins and used to program in vitro translation systems. These produced a simple translation profile that could be readily interpreted—and the accumulation of the processing products quantified. A control construct encoding [GFPGUS] produced only the fusion protein [GFPGUS], whereas the (GFP2AGUS) was shown to be highly active in mediating cleavage, producing the uncleaved polyprotein product (GFP2AGUS) (~5% of radiolabel incorporated) and the cleavage products (GFP2A) and [GUS], together accounting for ~95% incorporation of radiolabel. This approach gave the surprising result that the GFP reporter protein upstream of 2A accumulated to a higher level than that downstream of 2A (GUS) (Figure 1). Reversing the gene order (GUS2AGFP) produced the same type of imbalance. Having eliminated “trivial” causes for this imbalance (e.g. protein/RNA degradation, premature termination of transcription/translation) [3], these data were at variance with a proteolytic mechanism which predicts a unitary stoichiometry of the products. Inhibition of protein synthesis by the antibiotic puromycin produced a distinct product with a size corresponding to the upstream protein, indicating a pause on the mRNA at the site where the nascent chain is released [3]. The ribosome pause seen by ribosome toeprinting occurred at the end of the 2A coding sequence (-NPG^↓P-), with glycine and proline in the P- and A- sites, respectively [17]. This imbalance was due to different levels of biosynthesis of each portion of the ORF and represents a novel type of recoding [5, 18].

Figure 1.

Analysis of 2A-mediated “cleavage”. Artificial reporter polyproteins (boxed areas) used to programme in vitro translation systems are shown together with translation profiles obtained from rabbit reticulocyte lysates (right). The control pGFPGUS and pGUSGFP constructs produce only a single translation product—the [GFPGUS] and [GUSGFP] fusion proteins respectively. The translation profile from the pGFP2AGUS construct shows three major products: uncleaved [GFP2AGUS] and the cleavage products [GFP2A] and [GUS]. The profile from pGUS2AGFP also shows three major products: uncleaved [GUS2AGFP] and the cleavage products [GUS2A] and [GFP].

3. Breaking the code

Owing to the degeneracy of the genetic code, a protein sequence can be encoded by many different synonymous mRNA coding sequences in protein biogenesis. Furthermore, many organisms, including viruses, tend to have biases towards certain synonymous codons (the “codon bias”) in their genes. As described by Gao et al. [19] a marked codon bias within the FMDV genome is evident—the amino acids E, S, N, P, G, P tend to use GAG, TCC, AAC, CCT, GGG and CCC respectively. However, in the context of a synthetic reporter polyprotein, synonymous codon usage of this conserved motif does not affect apparent cleavage efficiencies whereas introducing non-synonymous changes impairs cleavage activity. Frameshifting the 2A oligopeptide with respect to the reporter proteins completely abolishes cleavage activity [3, 10, 15, 20]. Further, synonymous codon usage patterns for G_2A at the 2A/2B cleavage site (GGG, GGC, GGA, GGT) have no effect on the cleavage efficiency [19]. This supports the view that it is the amino acid residue rather than the nucleotide sequence which is critical for the activity of 2A. In contrast to these studies, Kjær and Belsham found that in FMDV genomes codon bias operates in encoding the NPGP motif, raising the possibility that the RNA sequence could also contribute to the recoding event [21]. It’s noteworthy that conservative mutation of all potential nucleophilic amino acids within the highly conserved D¹²(V/I)E(S/T)NPG_2A^↓P¹⁹_2B motif are active—further proof that 2A-mediated cleavage is not a proteolytic mechanism [15]. The take home message from mutagenesis experiments is that 2A is largely intolerant to sequence alteration over most of its length—2A peptides are fine tuned to function as a whole [5, 9, 10, 15, 22]. Our own extensive mutagenesis indicates that several variations of the conserved motif still display some level of activity [10, 15]. However, certain 2A sequences have not evolved to produce a simple binary outcome. “Sub-optimal” 2A sequences usually result in either “standard” translation (generating a fusion protein), or, a molar excess of products upstream of 2A—due to dissociation of ribosomes after peptide release, rather than recommencing elongation. It is not difficult to imagine clear advantages associated with incomplete StopGo activity (e.g. dual protein targeting, described in Roulston et al. [23]), but also the possibility of “regulating” the ratio of up- to down-stream products (e.g. “sensor” of translational stress, described in [24]).

Various studies have demonstrated that synonymous codon usage bias plays an important role in the translation of certain mRNAs [25, 26]. While codon optimisation is routinely used to increase translational efficiency, with regard to 2A peptides, wild type sequences may be evolutionary optimised to enhance the StopGo mechanism. The length of the FMDV 2A used is also important for cleavage in vitro/in vivo—N-terminally truncated forms of 2A in our reporter system showed that the minimal length required for activity was 12aa (12 N-terminal and 1 C-terminal to the G^↓P site) [9, 15]. In the case of shorter 2As, cleavage efficiency has been improved by insertion of various spacer sequences immediately upstream of the 2A sequence such as a glycine-serine linker (e.g. -GSG- or -SGSG-, [27, 28, 29]; a 3XFLAG epitope tag -DYKDHDG-DYKDHDI-DYKDDDDK-, [30]; or a V5 epitope tag -GKPUPNPLLGLDST- [31]. These “flexible” linkers create a space between the N-terminal protein and the 2A peptide, favouring a conformation which facilitates efficient cleavage [32]. By contrast, an optimised Kozak sequence from the silkworm (Bombyx mori) placed immediately downstream of the 2A peptide, improved the expression of the sequence downstream of 2A [33]. Longer versions with extra sequences (>5aa) derived from the FMDV capsid protein (“1D”) immediately upstream of 2A produce higher levels of cleavage [12, 15, 16]. After “tweaking” of the FMDV 2A sequence, 2A (+11aa 1D) proved to be the most favourable in terms of both length and cleavage efficiency [34, 35]. GSG linkers and longer versions of 2A sequences to improve “skipping” efficiencies are not fail-safe solutions and should be assessed empirically for every protein in context. Previously described position effects have also shown that expression of genes flanking 2A is highly dependent on their arrangement [33, 36, 37, 38]. For example, the yield of β-carotene by engineered yeasts could be altered by exchanging the order of codon-optimised carotenogenic genes GGPPS, CARB and CARRP [39].

4. A translational model of 2A

4.1 Reprogramming the ribosome

Based on dynamic molecular modelling and results from experiments with in vitro translation systems, the model of 2A-mediated translational recoding [3, 20, 40] proposes the nascent 2A peptide (~25aa) interacts with the exit tunnel of the ribosome such that the C-terminal portion is sterically constrained within a region of the peptidyl transferase centre (PTC) of the ribosome that cannot be accessed by prolyl-tRNA^Pro. The ribosomal PTC resides in the large ribosomal subunit and catalyses the two principle chemical reactions of protein synthesis: peptide bond formation and peptide release. It should be noted that the unique case of proline the nucleophile (nitrogen) is also sterically constrained since it is part of a ring structure. It has been shown that prolyl-tRNA^Pro is the poorest nucleophile among all aminoacyl tRNAs—polypeptide elongation is paused at this specific site. Our StopGo model predicts that (i) prolyl-tRNA^Pro dissociates from the A-site, (ii) the nascent polypeptide is released at this site by termination factors eRF1/3 and (iii) prolyl-tRNA^Pro (re)enters the ribosome a the A-site and must by translocated to the P-site to allow the next aminoacyl-tRNA to enter for resumption of translation and the synthesis of the downstream product (Figure 2) [17, 41]. Machida and colleagues showed however that processing of encephalomyocarditis virus proteins 2A and 2B in a reconstituted in vitro translation system requires neither eukaryotic initiation or termination (release) factors [42].

Figure 2.

Model of 2A translational “recoding” activity. In ribosome “skipping”, the 2A sequence is thought to interact with the ribosome exit such that the stereochemistry in the peptidyl transferase centre is altered: the peptide bond is not formed. The model proposes that prolyl-tRNA exits the A-site, eRF1/3 enters the A-site (at a sense codon) and terminates translation. After exit of eRF1, prolyl-tRNA then re-enters the A-site and is then (pseudo) translocated such that the A-site is now vacant and elongation can resume.

Our proposed model of this StopGo translational recoding event predicts that three alternative outcomes arise; having synthesised sequences upstream of 2A, ribosomes either (i) release the nascent peptide and then resume translation of the downstream sequences, (ii) release the nascent peptide and translation is terminated at that point, or, (iii) that no translational recoding occurs: the glycyl-prolyl peptide bond is formed and the protein is synthesised in the normal manner. The ratio of these translation products is dependent upon the specific 2A-like sequence in question—a view supported by the observations from many of the laboratories that have used StopGo protein co-expression applications. Initially studied in mammalian cell cultures (HeLa, [14]; HEK293, [43]; CHO, [44]) 2A-mediated ribosome skipping functions in all eukaryotic expression systems tested to date: amoeba [45], fungi [38, 46, 47, 48, 49], algae [50, 51], plants [52, 53] and animals [33, 54, 55, 56] but not prokaryotic systems: the structure of the ribosome and the exit tunnel is highly conserved among eukaryotes but differs between eu-and prokaryotes.

4.2 Targeted expression of recombinant proteins

Our model of the 2A mechanism predicted this “cleavage” would occur within the ribosome: that any protein downstream of 2A would emerge from the ribosome with a nascent N-terminus. Therefore, if one included a co-translational signal sequence immediately downstream of 2A, then this would be recognised by the signal recognition particle (SRP) as if it were generated during “normal” translation. In a proof-of-concept study this was the case: polyproteins were constructed comprising internal signal sequences and we were able to show that such proteins could be both co-expressed and independently targeted to sub-cellular sites (or secreted), greatly increasing the utility of the 2A co-expression system [57, 58]. To verify the applicability of this method to developmental studies, Trichas et al. designed a 2A bicistronic reporter that correctly processed a membrane-localised TdTomato gene (Myr-TdTomato) and nuclear-localised enhanced green fluorescent protein (EGFP) gene (H2B-GFP) in cell culture and transgenic mice [59]. Additionally, this approach produced functional expression of Myr-mCherry and H2B-GFP in the branchipod crustacean Daphnia magna [60]. Meanwhile, we demonstrated the (partially) active 2A-like sequences present at the N-terminus of a number of NOD-like receptor proteins (NLRs) within the purple sea urchin (Strongylocentrotus purpuratus) also play a dual role as exocytic pathway signal peptides. If 2A mediates translational recoding, the 2A signal sequence is synthesised as a discrete translation product separate from the downstream product which is localised to the cytoplasm. If 2A does not mediate translational recoding, the 2A signal sequence is fused to the downstream translation product, functions as a signal sequence and targets the fusion protein to the exocytic pathway—a novel form of dual protein targeting (Figure 3) [23].

Figure 3.

Cleavage and signal sequence functions of 2A. The wild-type and point mutant version was cloned upstream of sequences encoding cherry FP and GFP linked via Thosea asigna virus (TaV) 2A into a single ORF. Image analysis shows the wild-type 2A produces a high proportion of cherry localised in the cytoplasm (and diffuses into the nucleus). The mutant inactive 2A acts as a signal targeting cherry to the exocytic pathway (bars ~5μm. Nuclei stained with DAPI are shown in blue).

4.3 The unwanted “tags”

The addition of partial 2A peptide sequences to the C-terminus of proteins may interfere with enzymatic activity [61]. For secreted proteins, inclusion of a furin recognition sequence (e.g. ^↓-RRR-, ^↓-RKRR-, ^↓-RRKR-) between the first gene and 2A results in efficient removal of 2A residues—furin is a ubiquitous serine protease localised on the trans-Golgi networks of virtually all cell types [62]. Proteins expressed in plants/fungi could have this tag removed by endogenous proteinases acting on similar hybrid linker peptides, LP4-2A in plants (SN^↓AADEVAT) [63], TEV-2A in fungi (ENLYFQ^↓S) [49]. Although not a full-proof solution, a GSG linker used in conjunction with a furin recognition site (e.g -furin-GSG-2A-, -GSG-furin-GSG-2A-) reportedly enhanced cleavage efficiency via increased exposure of the -RRKR- site [32, 44, 64]. The “unwanted” tag may remain—recognised by monoclonal antibodies, peptide epitopes can characterise, purify, and localise proteins of interest in vitro and in vivo [65, 66]. Commercial antibodies have been raised against the consensus 2A sequence, which thus serves as a useful target for identifying 2A-tagged proteins in biochemical assays—validated research applications include ICC, immunofluorescence, IP, and Western Blot [67, 68]. The presence of a proline amino acid residue (or often several additional amino acids, derived from restriction site sequences introduced during cloning) at the N-terminus of co-expressed proteins may affect protein stability within polycistronic constructs [69, 70, 71]. In this case, proteins that require authentic termini can be introduced at the first gene position.

5. 2A and 2A-like sequences

5.1 Viral and cellular 2As

The presence of 2A-like sequences has been reported in a number of viral genomes within different genera of the Picornaviridae, other positive strand viruses such as the Dicistroviridae and Iflaviridae [16, 72], double-strand RNA viruses belonging to the Totiviridae/Reoviridae families [73], and surprisingly in a tentatively assigned negative-sense single-stranded RNA virus of the Bunyaviridae family [72]. The positive-strand RNA viruses typically possess one 2A/2A-like sequence, but some viruses have two, three or more motifs—the picornavirus duck egg-reducing syndrome virus (DERSV) was shown to have a total of seven 2A proteins, the first six separated by the “DxExNPGP” sequence motif (Table 1) [74]. Translation in vitro with our pSTA1 dual reporter system [GFP-2A-GUS] shows some of these multiple 2A^NPGP sequences perform just as well as the commonly used FMDV 2A peptide: IMNV 2A1 and 2A2 [16]; Aalivirus A1/B1 2A1-2A4; Grusopivirus A1/C 2A1-2A3; Limnipivirus A1 2A1, B1 2A2, C1 2A1, D1 2A1-3; Mosavirus B1 2A1 and 2A2 (unpublished data). The advantage of releasing more than one protein from the same ORF suggests that host infection mechanisms in viruses with multiple 2A peptides in their genomes may be more complex [6]. Probing databases with this conserved motif revealed cellular 2A-like sequences within non-LTR retrotransposons (non-LTRs) of Trypanosoma brucei, T. cruzi, T. vivax and T. congolense, and, a range of simple aquatic species: Xenopus tropicalis (African claw-toed frog, vertebrate), Branchiostoma floridae (Amphioxus, Florida lancelet, cephalochordate), Aplysia californica (California sea slug, mollusc), Crassostrea gigas (Pacific oyster, mollusc), Lottia gigantean (Owl limpet, mollusc), Nematostella vectensis (sea anemone, cnidarian) and S. purpuratus (echinoderm) [75, 76, 77]. To allay public fears and bypass the use of animal-virus sequences, non-viral 2A variants (e.g. those found in the California sea slug and purple sea urchin) have been used just as effectively for coordinated multi-protein co-expression [78].

Genus	Species	2A^NPGP	Amino acid sequence	Accession no
Positive-stranded RNA viruses: Picornaviridae
Aalivirus	AalV-A1	2A1	LLTSEGATNSSLLKLAGDVEENPGP	KJ000606
		2A2	FEMPYDDPEWDRLLQAGDIEQNPGP
		2A3	PIPARPDPQWNNLQQAGDVEMNPGP
		2A4	EHFNQTGGWVPDLTQCGDVESNPGP
	AalV-B1	2A1	ATTLQVSEYLKDLTIDGDVESNPGP	MH453803
		2A2	LKVKKLEGDYVRDLTQEGVEPNPGP
		2A3	SVRVTDAGWVRDLTVDGDVESNPGP
		2A4	VFKCHDKCWVDDLTNCGDVESNPGP
		2A5	IFKCHEGCWVEDLTVDGDVESNPGP
	DERSV	2A1	TSTAQATSYVKDLTIDGDVESNPGP	UYL81882
		2A2	KTCREVEGSYVKDLTEEGIEPNPGP
		2A3	LLKIGNAAWVRDLTEDGDVEENPGP
		2A4	VYNCHESCWNRDLTIDGDVELNPGP
		2A5	VFKCHEKCWQKDPTQDGDVEQNPGP
		2A6	EFKCHEHCWVRDLTMDGDVEENPGP
Avisivirus	AsV-A1	2A1	EVGAYDEVDHRDILMGGDIEENPGP	KC465954
		2A2	EMGVFDETDHRDILLGGDIEENPGP
	AsV-B1	2A1	PQFEKERSAHEDVLLGGDVESNPGP	KF979333
		2A2	SESVQYLEPQIDICVCGDVERNPGP
Grusopivirus	GrV-A1	2A1	FEKHVKPWRSQEDLSKEGIEPNPGP	KY312544
		2A2	ITDNRYKETDAKWLSRYGVEMNPGP
		2A3	VTQDLYAATNQDQLSNQGIESNPGP
	GrV-C	2A1	YFEERSPHPTQKELGQFGVETNPGP	MK443503
		2A2	ENNSNYDERDAKHLSRYGIEMNPGP
		2A3	CVCTRWSPTMQSELGKYGIEKNPGP
Kunsagivirus	KuV-C1	2A1	IAAASAQGWQRDLTQDGDVESNPGP	KY670597
		2A2	LGIVISDSVWQRDLPREGVEENPGP
		2A3	SYDPLAPSQWCRDLTCEGIEPNPGP
Limnipivirus	A1	2A1	CKEFVRESDNQELLKCGDVESNPGP	JX134222
		2A2	WDLSTGWFHFFRLLRSGDVEQNPGP
	B1	2A1	MDVVDDYPFKRDLTRDGDVESNPGP	KF306267
		2A2	IDLVQAAYSRMRLLLSGDVEQNPGP
	C1	2A1	KLLEQILAYKRDLTACGDVESNPGP	KF874490
		2A2	SRWIHARFARLRLLLSGDVEQNPGP
	D1	2A1	EEEVDWGVGRMRLKMSGDVEENPGP	MG600094
		2A2	AVHLLVTWMRRRLTLSGDIESNPGP
		2A3	DLRAVKSFIESQLMRAGDVERNPGP
Mosavirus	B1	2A1	ESRGTGNCDATTISQCGDVETNPGP	KY855435
		2A2	YVRRSANRTAADISQDGDVETNPGP
Parechovirus	E	2A1	WFDARTGFKTPLMNPCGDVEENPGP	KY645497
		2A2	QIEKRYGYRFWLLMLCGDVELNPGP
	RtPV	2A1	MLDRRMGYRSRILCQCGDVEENPGP	MF352429
		2A2	WFNKRSGYRSRLLSQCGDVEENPGP
Potamipivirus	B1	2A1	LMEKTEEAGWLRDLTREGVEENPGP	MK189163
		2A2	FDDYHQEGGWIRDLTAEGVEPNPGP
Unassigned	WCP	2A1	MKEDEAGGWKEDLTEDGDVESNPGP	MG00066
		2A2	EQAIPETTWRRDLTQSGDVESNPGP
		2A3	PGAIPASVWVHDLTTDGDVESNPGP
Unassigned	WP-LV 48	2A1	GPSCYDRNNHCNILLSGDIEENPGP	NC_032820
		2A2	VFNASYLDCFISLLSCGDIESNPGP
		2A3	PIQGLTQRFESTLLLGGDIEENPGP
Positive-stranded RNA viruses: Iflaviridae
Iflavirus	EoPV	2A1	EQIVTAQGWAPDLTQDGDVESNPGP	NC_005092.1
		2A2	QRQNIIGGGQRDLTQDGDIESNPGP
	PnPV	2A1	EQIVTAQGWVPDLTVDGDVESPNPG	NC_003113.1
		2A2	RRQNIIGGGQKDLTQDGDIESNPGP
	DBMIV	2A1	EQIVTAQGWVADLTQDGDVESNPGP	NC_034384.1
		2A2	LRQNKILGGERDLTRDGDVESNPGP
Non-segmented double-stranded RNA viruses: Totiviridae
Unassigned	IMNV	2A1	IEISDCMLPPPDLTSCGDVESNPGP
		2A2	IEKPFDKEEHTDILLSGDVESNPGP

Table 1.

Multiple 2A/2A-like sequences in viruses.

5.2 Comparison of active 2A sequences

Of the many 2A peptides identified to date, four viral 2As have been widely used in biotechnology and biomedicine: FMDV (“F2A”), equine rhinitis A virus (ERAV, “E2A”), porcine teschovirus-1 (PTV-1, “P2A”), and Thosea asigna virus (TaV, “T2A”) (see Table 2) [27, 54, 59, 79]. Comparing the in vitro activity of different 2As inserted between GFP and GUS, T2A₂₀ has the highest cleavage efficiency followed by E2A₂₀, P2A₂₀ and F2A₂₀ [15]. In human cell lines, mice and zebrafish, cleavage and targeting of NLS-EGFP-2A-mCherry-CAAX was most efficient in P2A₁₉ linked constructs followed by T2A₁₈, E2A₂₀ and F2A₂₂ [43]. To test the performance of selected 2A peptides in insect cell cultures (B.mori) the cleavage efficiency of six types of 2A with a GSG linker were analysed: P2A, T2A, E2A, F2A, BmCPV2A and BmIFV2A. For the most used sequences P2A₂₂ exhibited the highest cleavage efficiency in all insect cell lines tested followed by E2A₂₃, F2A₂₅ and T2A₂₁ [32]. Interestingly, P2A₁₉ and T2A₁₈ functions in Drosophila are approximately equivalent both in cultured cells and in vivo [80]—given their poor efficiency at polypeptide separation, E2A and F2A may not be useful in Drosophila [80, 81]. A study to characterise the 2A system for metabolic engineering applications in Saccharomyces cerevisiae showed that the F2A₁₉, T2A₁₈, and P2A₁₉ sequences are functional in S. cerevisiae cells [82]—earlier analysis in S. cerevisiae indicated that ERBV-1 (Equine rhinitis B virus 1) 2A had the highest cleaving efficiency among 22 viral 2A sequences tested [48]. For potential upcoming amoeba-based bioprocesses codon optimised P2A₁₉, T2A₁₈, E2A₂₀ and F2A₂₂ were screened for activity in Dictyostelium discoideum [45]. In this work P2A and T2A performed the best with cleavage rates of 99.6% and 97.4% respectively, followed by E2A (95.5%), and F2A (66.9%). The availability of a range of 2A sequences can be useful—to minimise the risk of homologous recombination, it is wise to use different 2A peptide sequences if more than two genes are being linked.

Table 2.

2A/2A-like sequences most commonly used for protein co-expression.

6. IRES-dependent translation

Current strategies for multi-gene co-expression include: (i) pre-mRNA splicing; (ii) proteolytic cleavage sites; (iii) fusion proteins; (iv) IRESs (internal ribosomal entry sites); (v) self-cleaving 2A peptides and (vi) a new-kid-on-the-block termed HACKing (Highly efficient and Accessible system by CracKing genes into the genome) [83]. Among the six systems listed, the most popular strategies are 2A/2A-like sequences and viral IRES elements. Like the 2A peptide, the IRES was identified first among picornaviruses but in a different group, the Enteroviruses, typified by poliovirus (PV) [84]. Since then, IRES elements have been identified in other positive-sense RNA viruses such as encephalomyocarditis virus (EMCV) [85], DNA viruses [86, 87] and a growing number of cellular mRNAs involved in growth control, cell cycle progression and apoptosis [88]. In the majority of eukaryotic cells the mRNA 5′ 7meG “cap” is a signal for eukaryotic initiation factors (eIFs) to recruit ribosomes and initiate translation [89]. IRES elements allow the initiation of translation in a cap-independent manner, i.e., ribosomes bind internally without scanning the 5′ UTR of the transcript ensuring co-expression of genes before and after the IRES [90]. Most IRESs require IRES trans-acting factors (ITAFs) to function in addition to several canonical IFs [91]. These cis-acting RNA regulatory elements “hijack” ribosomes of the host cell, redirecting them towards the production of viral proteins [92, 93]. Viral IRESs have been classified into four different types based on their secondary structure and, in turn, their mechanism of initiation: Type I (enteroviruses/rhinoviruses), Type II (cardio-/aphthoviruses), Type III (flaviviruses e.g. HCV) and Type IV (dicistroviruses e.g. CrPV) [94, 95]. It has been shown that the efficiency of translation initiated from different IRES elements differs significantly [96, 97, 98].

Most IRES-based vectors developed up to now use picornavirus IRESs (e.g. EMCV and PV), based on the strong efficiency of such IRESs in transient transfection, compared to cellular IRESs [99, 100]. In mRNAs transcribed from bicistronic constructs, the first cistron is translated by a cap-dependent scanning mechanism while translation of the second cistron depends on the presence of IRES in the intercistronic space. The advantages of IRES-mediated systems include: complete separation of the coupled genes; lack of fusion proteins which may adversely affect the activity of the proteins; the relative expression of different genes can be “fine-tuned” by varying the strength of the IRES applied to each gene. IRES sequences, however, have limitations: (i) the large size (~500 bp) can reduce the packaging capacity in viral vectors [101], (ii) expression of the downstream gene can be as much as 10 fold lower than the upstream gene [102, 103, 104], (iii) recognition of different IRES elements varies with cell type, species, and even cell stress state [91, 97] and lastly (iv) the risk of homologous recombination and competition among different IRESs for translation factors [105].

Nevertheless, different IRES-based vectors including plasmids, adeno-associated virus (AAV)-derived and lentiviral vectors have been used successfully in pre-clinical as well as in clinical gene therapy protocols (reviewed in [106]). In plants, IRES-based constructs have been used to engineer pest resistance in tobacco [107], enhance salt and/or drought tolerance in potato [108] and produce carotenoids in rice endosperm [104]. However, compactness in length and their ability to allow the concatenated genes to be translated at equivalent levels both in vivo and in vitro has seen 2A peptides gain popularity over traditional IRES elements. Adenovirus [109], AAV [110], retrovirus [111], lentivirus [112] and plasmid [64] expression vectors incorporating 2A peptide sequences have been used in a variety of biotechnology and biomedical applications (reviewed in [113, 114, 115]).

7. Biotechnology/biomedical applications

7.1 Heterologous expression of multi-gene biosynthetic pathways in yeast

A particular advantage of using 2A co-expression technology is the ability to consistently express all the required genes in the same cell. For example, in biosynthetic pathways involving a number of separate enzymes it is critical they are all expressed in the same cell. If this is not the case, instead of progressing from reactants to products a breakdown in the pathway could lead to a build-up of volatile intermediates. Enzymatic systems requiring several cofactors have been reconstructed, by means of 2A peptides, both in Pichia pastoris and S. cerevisiae [48, 116, 117]. Significant advantages of the P. pastoris expression system include post-translational modifications such as glycosylation, proper protein folding and secretion (by Kex2 as signal peptidase) of recombinant proteins [118]. In the early days, environmental stress conditions imposed on the host presented a potential bottleneck in terms of recombinant protein production and secretion. Particular attention has been paid to the role of glycine betaine (GB) synthesis in abiotic stress resistance [119]. The GB pathway is a two-step oxidation from choline and betaine aldehyde, catalysed by choline monooxygenase (CMO) and betaine aldehyde dehydrogenase (BADH), respectively. The genes encoding the two enzymes of GB synthesis in the halophyte Suaeda salsa were cloned and fused with the F2A in a single ORF. The fused genes “CMO-2A-BADH” transformed in P. pastoris were expressed successfully and the polyprotein was cleaved to each functional protein [CMO-2A] and [BADH]. The recombinant yeasts were more tolerant to salt, methanol and high temperature stresses [120].

Biobased manufacturing for the production of fuels has become increasingly popular as the serious effects of global climate change become apparent. Using a viral 2A sequence, P. pastoris was engineered to produce the biofuels isobutanol and 3-methyl-1-butanol (3M1B, isoamyl alcohol). Upregulation of the endogenous L-valine biosynthetic pathway (PpIlv5, PpIlv3, PpIlv6, PpIlv2) and heterologous expression of the keto-acid degradation pathway (LlkivD, ScADH7) led to an engineered strain capable of producing isobutanol [121]. Specifically, the strain overexpressing all six genes produced a 100-fold improvement over the levels observed in the strains overexpressing only the two keto acid degradation pathway genes. In a follow on study, engineered strains overexpressing the keto-acid degradation pathway as well as the upper portion of the amino acid L-valine and L-leucine (PpLeu1, PpLeu2, PpLeu6) biosynthetic pathways were able to produce the platform chemical 3M1B [122]. Compared to the predominant biofuel ethanol, the advantages of 3M1B can be summarised as higher energy density, better miscibility with fossil fuels, lower water solubility, and greater applicability to conventional engines. The potential of this system is vividly demonstrated by the expression of nine genes from a single polycistronic transcript based on T2A peptides—presently, the construct with the highest number of genes expressed in a coordinated fashion. This expression strategy resulted in P. pastoris strains producing the target products of both the carotenoid (CrtE, CrtB, CrtI,CrtY) and violacein (vioC, vioB, vioE, vioD, vioA) biosynthesis pathways [47].

To date, a variety of studies have examined the use of multicomponent enzymatic systems based on the 2A approach in the baker’s yeast S. cerevisiae [123, 124, 125, 126, 127, 128]. The utility of 2A peptides has been demonstrated in several metabolic engineering applications, such as the production of C-glucosylflavones or β-carotene. Flavonoids are important phytochemical compounds with a range of medical benefits including powerful activities as antioxidants. In rice plants, two enzymes are responsible for the biosynthesis of flavone-C-glycosides, the cytochrome P450 flavanone 2-hydroxylase (F2H) and C-glucosyltransferase (CGT). While pathway engineering in tobacco resulted in minor C-glycoside formation, yeast transformed with the [F2H-F2A-CGT] polyprotein produced high concentrations of 2-hydroxynaringenin which could be converted chemically to the “nature-identical” plant product [129]. “Carotenoids” is a generic term used to designate a diverse group of pigments widely distributed in nature. Chemical synthesis of the flavour and fragrance (apo-) carotenoid β-ionone is economically impractical. For de novo production of apocarotenoids in yeast, three β-carotene biosynthesis genes from the carotenoid-producing ascomycete Xanthophyllomyces dendrorhous (crtI, crtE and crtYB) were co-expressed in S. cerevisiaefrom a single T2A polycistronic construct. To enable β-ionone production, a carotenoid-cleavage dioxygenase from raspberry (RiCCD1) was co-expressed in the β-carotene producing strain - a novel microbial production system for a fruit flavour compound [130].

7.2 Engineering carotenoid biosynthesis in plants

Conventional methods to “stack” transgenes in plants include sexual crossing, sequential re-transformation and single/multiple plasmid co-transformation [131]. These strategies, however suffer from the inherent weakness that co-expression of the heterologous proteins is unreliable. 2A/2A-like constructs have been successfully used to co-express β- and δ-zein proteins to improve the nutritional value of tobacco plants [132]; mpi (maize proteinase inhibitor) and pci (potato carboxypeptidase inhibitor) for insect and pathogen resistance in rice [133] and Baccillus thuringiensis (Bt) crystal proteins, Cry1Ab and Cry2Ab to delay insect resistance to Bt toxins in rice [134]. In a recent study, 11 genes were successfully introduced into transgenic maize to rebuild the anthocyanin biosynthesis pathway (ZmBz1, ZmBz2, ZmC1 and ZmR2) by combining the advantages of a bidirectional promoter (BDP) and 2A linker peptides [135]. Carotenoids have attracted interest not only as a source of pigmentation but also for their beneficial effects on human health—since most animals are unable to synthesise carotenoids, they need to obtain them from food and dietary supplements. To alleviate vitamin A deficiency (VAD) through food intake, biofortified “Golden Rice” (Oryza sativa, GR) was engineered via two gene cassettes for the expression of the phytoene synthase (Psy) and carotene desaturase (CrtI) genes to produce β-carotene (provitamin A) (GR1 [136], GR2 [137]). Among GR varieties, “Korean Golden Rice” was developed using the recombinant PAC gene that linked the Capsicum Psy and Pantoea CrtI genes (PAC, Psy-F2A-CrtI) [104]. To increase the levels of carotenoids, the DNA sequences of these two carotenoid biosynthetic genes originating from heterologous sources, Capsicum annum for Psy and Pantoea ananatis for CrtI, were codon optimised for rice plants (pstPAC) [138]. In a separate study, Kim et al. genetically manipulated the carotenoid biosynthetic pathway in Korean soybean using an efficient Agrobacterium-mediated transformation method and the PAC recombinant gene to produce transformants whose β-carotene levels exceeded those reported in transgenic rice plants [139].

Zeaxanthin, capsanthin, canthaxanthin and astaxanthin are high-value carotenoids used industrially as colourants and feed supplements. Their antioxidant properties orchestrate important pathways in animals, including immuno/photo protection and oxidative stress modulation [140]. In both prokaryotes and eukaryotes, Astaxanthin is derived from β-carotene by hydroxylation and ketolation reactions catalysed by β-carotene hydroxylase (BCH) and β-carotene ketolase (BKT), respectively. In early studies of ketocarotenoid production, both genes from a marine bacterium (Paracoccus sp) were expressed simultaneously as a F2A polyprotein in both tobacco and tomato plants [141]. β-Carotene can typically be increased by the introduction of one or two transgenes, however producing diverse carotenoids beyond β-carotene (e.g. zeaxanthin and ketocarotenoids) typically requires the simultaneous engineering of a greater number of transgenes [142]. To produce downstream carotenoids, Ha and colleagues adopted three steps of the pathway to produce Zeaxanthin—a BCH expression cassette, under the control of the rice globulin promoter, combined with a PAC gene giving rise to a single T-DNA vector (pB-PAC; Bch and Psy-2A-CrtI). Astaxanthin was produced by the introduction of four steps (PSY, CRTI, BCH and BKT) using a single T-DNA vector (pBAK-PAC) with two cassettes for two bicistronic genes (BAK, Bch-F2A-Bkt and PAC, Psy-F2A-CrtI). The single T-DNA vectors resulted in the accumulation of zeaxanthin and astaxanthin in the endosperm of the transgenic rice seeds [143]. In a later study, bi-, tri and quadcistronic 2A systems for co-expression of these four genes were tested in transgenic rice seeds. To appease public concerns, three highly active 2A variants not found in mammalian viruses, T2A₂₀, IMNV2A1₃₀ and IMNV2A2₃₀ were used for the coordinated expression of multiple transgenes. This study demonstrated successful expression of two, three and four genes resulting in stepwise biosynthesis of β-carotene, zeaxanthin, adonixanthin and astaxanthin [144].

7.3 The broad utility of 2A and 2A-like sequences

Previous and more recent works employed the 2A approach to assemble biosynthetic pathways for heterologous expression in filamentous fungi. For bioproduct synthesis, the P2A peptide was employed to reconstitute synthesis of β-lactam antibiotics by moving the genes for the complete penicillin biosynthesis pathway from Penicillium chrysogenum (pcbAB, pcbC, and penDE) to the fungal host Aspergillus nidulans [145]. A similar approach was also used by Hoefgen and colleagues to produce the psychotropic mushroom alkaloid psilocybin by heterologous expression of the entire biosynthetic gene cluster (psiH-psiD-psiK-psiM) in A. nidulans [49]. A 2A peptide-based bicistronic protein expressing platform was used to monitor transgene expression (Cel7A, cellobiohydrolase) in the cellulase producing fungus Trichoderma reesei using eGFP as a marker (Cel7A-F2A-eGFP or eGFP-F2A-Cel7A) [38]. To date, 2A peptides have been used in various artificial reporter polyprotein systems comprising chloramphenicol acetyl-transferase (CAT), GUS and fluorescent proteins (FPs, e.g. GFP, RFP, YFP) in various cell types, as well as targeting to various subcellular localisations in plant [52, 146] and animal cells [56, 59]. Targeted endonucleases including zinc finger nucleases (ZFN), transcription activator-like effector nucleases (TALENs) and clustered regularly interspaced short palindromic repeats (CRISPRs) are increasingly being used for genome editing (GE) in numerous cell types and species. GE by using 2A peptide coupled co-expression of nuclease and fluorescent proteins combined with fluorescence-activated cell sorting (FACS) can aid selective enrichment of transfected cells. This approach to achieve higher genome editing efficiencies works well for these three major nucleases [147, 148]. To expand the experimental toolkit, antibiotic (hygromycin) and magnetic H-2K^K cell selection methods have been developed to enrich transfected cells with ZFN/TALEN-induced mutations (2A-Hygro^R-eGFP and eGFP-2A-H-2K^k) [149]. In GE, the two colour “Traffic Light Reporter” (TLR; eGFP-T2A-mCherry) developed by Scharenberg and colleagues has proved invaluable for evaluating the two major DNA repair pathways, nonhomologous end-joining (NHEJ) and homologous recombination (HDR) [150]. Developing new methodologies governing CRISPR-Cas activity using 2A sequences include genetic switches [151], logic gates [152], and gene regulation circuits [153].

New types of cancer treatment via gene therapy promise to revolutionise management of the disease. Here, 2A is used to co-express the T-cell receptor TCRα and TCRβ chains (in transformed patient T-cells: ex vivo gene therapy) targeted against cancer antigens. Use of 2A sequences ensures both subunits of the receptor can be co-expressed at similar levels, minimising side effects. A second major impact in human health has been in the rapidly expanding field of regenerative medicine and transplantation. 2A has played a pivotal role in the co-expression of the multiple transcription factors required to produce induced pluripotent stem cells (iPSCs)—one can now make patient-specific stem cells relatively simply to avoid immune response problems and/or organ rejection (reviewed in [115]). Many other biotechnological applications that depend on the co-expression of genes use 2A/2A-like sequences e.g. the production of antibodies and antigens that can be used in vaccine production [64, 154, 155]. Bovine ephemeral fever virus (BEFV) is an economically important pathogen that causes an acute febrile illness of cattle and water buffalo in many parts of the world (Africa, the Middle East, Australia and Asia). Various studies have been conducted to develop an efficient vaccine for BEFV, including live attenuated, subunit and recombinant vaccines [156]. Although experimental and commercial BEFV vaccines have been developed, these suffer from either genetic instability, low efficacy, the need for multiple doses, or, are simply too expensive under field conditions. The 14.9 kb (−ve) ssRNA genome of BEFV contains 10 genes in the order 3′-N-P-M-G-G_NS-α1-α2-β-γ-L-5′ (Figure 4) [157]. G protein, as a class I transmembrane surface glycoprotein, is the target of neutralising antibodies and a prime candidate for subunit vaccine design [158]. In a recent study, we developed an antigen expression system to express recombinant G/G_NS glycoproteins with the transmembrane anchor domain deleted, enabling truncated proteins to be secreted into tissue culture media after mammalian cell transfection [65]. The V5 epitope tag was genetically fused to the C-termini of the proteins for protein detection and purification of expressed proteins from the cell media (Figure 4). Furthermore, a genetic fusion of each of the virus glycoproteins with GFP via T2A was used for live-cell fluorescence microscopy. In this “proof-of-principle” study, truncated G glycoprotein was detected throughout the exocytic pathway and secreted efficiently from BHK cells into the cell media. In a follow-on study we evaluated the immunogenicity of a secreted, C-terminally truncated form of bovine viral diarrhoea virus (BVDV) E2 glycoprotein in mice [66]. BVDV is the causative agent of one of the most widespread and economically important virus infections in cattle [159]. Glycoprotein E2 plays a key role in host cell immune responses to viral infection [160]. Here, truncated E2 glycoprotein purified from cell media was found to induce both humoral and cellular immune responses in BALB/c mice. This antigen expression system provides both a simple purification protocol along with a feasible strategy for further, large-scale, production of vaccines.

Figure 4.

Genome structure of BEFV and synthesis of the regions encoding the G and G_NS glycoproteins. Structural organisation of the BEFV genome shown as arranged in negative sense. The BEFV genome encodes five structural proteins including a nucleoprotein (N), a polymerase-associated protein (P), a matrix protein (M), a large RNA-dependent RNA polymerase (L) and a glycoprotein (G) spanning the viral envelope and a non-structural glycoprotein (G_NS) followed by a series of “accessory” proteins. Regions encoding the full-length G and G_NS proteins (boxed areas) are shown, together with the signal peptide (grey shaded areas) and the transmembrane anchoring domains (blue shaded areas). Regions encoding the C-terminally truncated (transmembrane anchoring domains deleted) forms of G and GNS proteins (boxed areas) are shown, together with the signal peptide (grey shaded areas) and the C-terminal V5 epitope tag (white boxes).

A particularly useful aspect of the ribosome skipping mechanism is the possibility of differentially targeting the proteins produced from a single gene. For example, Park and colleagues used three 2A peptides (F2A, T2A and E2A) to simultaneously express four distinct fluorescent protein variants targeted to subcellular compartments such as mitochondria, nuclei, endosomes, and membranes for real-time monitoring of cellular dynamics in HeLa cells [161]. To facilitate studies of subcellular Ca²⁺ signalling in insulin-secreting pancreatic beta cells, particularly, signalling between the ER and mitochondria, Jeyarajan et al. [162] designed a plasmid that contained the sequences of two different Ca²⁺ reporter molecules, G-CEPIA-er and R-CEPIA3-mt, separated by T2A. Successful segregation of the two probes to their respective organelles allowed simultaneous measurement of changes in free Ca²⁺ to be made using live cell imaging in real time. 2A peptides were also used for the co-expression and differential subcellular targeting of two fluorescent marker proteins in insect cell culture systems (Drosophila melanogaster S2 cells) and Drosophila suzukii cell lines [163]. All four 2A peptides tested (F2A₃₁, F2A₃₂, T2A, DCV2A) showed comparable activity in cell lines, leading to the production of independent upstream and downstream proteins that were directed to the nucleus or membrane by a C-terminal nuclear localisation signal (NLS) on the upstream protein and a poly-lysine/CAAX membrane anchor on the downstream protein.

8. Conclusions

In prokaryotes proteins with linked functions (e.g. the formation of macromolecular structures, biochemical pathways) can be expressed in a co-ordinated manner by encoding multiple proteins as polycistronic mRNAs: a single RNA is transcribed from different genes organised—concatenated—into operons. This is only possible since prokaryotic ribosomes initiate translation by binding individual coding regions within the (polycistronic) mRNA. In contrast, eukaryotes initiate translation only at a single site—the 7meG “cap” structure at the 5′end of mRNA. Here, expression of proteins is achieved by the co-ordinated transcription of the individual genes located at different sites throughout the genome. The model of one gene one enzyme/one gene one polypeptide developed in the 1940s was substantially modified in the late 1960s from studies on the biogenesis of poliovirus proteins and insulin. In both these cases it was proposed proteins were formed by post-translational processing of a single precursor (poly)protein [164, 165, 166, 167]. Encoding multiple proteins in the form of a polyprotein is a common, if not ubiquitous, strategy of viruses with positive stranded (mRNA sense) genomes. Such polyproteins are proteolytically “processed” by virus-encoded proteinases, although host-cell proteinases may also be utilised in processing virus polyproteins.

Viruses have evolved many different mechanisms to both maximise the coding capacity of their genomes and to generate multiple, different, proteins from a single ORF—many involving manipulating or modifying the host-cell translational apparatus [2]. In the case of 2A-mediated ribosome skipping, the strategy has been very widely used in biotechnology and biomedicine in the generation of transgenic cell lines/organisms as outlined above. Being able to concatenate genes in this manner has simplified and accelerated the whole process of producing transgenics: reducing, or even eliminating, the need for re-iterative rounds of transformation/selection required to introduce “traits” based upon multiple gene expression. The range of applications is immense, spanning humans, animals, plants and fungi. To conclude, the study of FMDV polyprotein processing has provided a valuable tool that has allowed researchers to create “artificial” polyproteins such that multiple, different, proteins can be generated from a single (poly)cistronic construct—driven by a single promoter. Furthermore, since 2A-mediated recoding occurs co-translationally, products can be targeted to different combinations of sub-cellular sites or secreted from the cell. Over 1500 research publications (https://www.st-andrews.ac.uk/ryanlab/Index.html) and many patent/patent applications (https://www.freepatentonline.com/login.html) stand testament to the influence 2As have had on co-expression technology—this is not the end, this is not even the beginning of the end, this is just perhaps the end of the beginning.

Acknowledgments

The long term support of the Wellcome Trust and the Biotechnology and Biological Sciences Research Council (BBSRC) is gratefully acknowledged. The University of St Andrews is a charity registered in Scotland no. SCO13532.

Abbreviations

AAV	adeno-associated virus
BDP	bidirectional promoter
BEFV	bovine ephemeral fever virus
BHK	baby hamster kidney
BmNPV	bombyx mori nuclear polyhedrosis virus
BmCPV	bombyx mori cytoplasmic polyhedrosis virus
Bt	Baccillus thuringiensis
CAT	chloramphenicol acetyl-transferase
CHYSEL	cis-acting hydrolase element
CRISPR-Cas9	clustered regularly interspaced palindromic repeats – associated protein 9
CrPV	cricket paralysis virus
DBM	diamond back moth
DBMIV	diamondback moth iflavirus
DCV	Drosophila C virus
DERSV	duck egg-reducing syndrome virus
DNA	deoxyribonucleic acid
EoPV	ectropis oblique picorna-like virus
EMCV	encephalomyocarditis virus
eIFs	eukaryotic initiation factors
ER	endoplasmic reticulum
ERAV	equine rhinitis A virus
ERBV-1	equine rhinitis B virus 1
FACS	fluorescence-activated cell sorting
FMDV	foot-and-mouth disease virus
FPs	fluorescent proteins
GE	genome editing
GFP	green fluorescent protein
GR	golden rice
GUS	glucuronidase
HACKing	highly efficient and accessible system by cracking genes into the genome
HCV	hepatitis C virus
ICC	immunocytochemistry
IMNV	infectious myonecrosis virus
IP	immunoprecipitation
iPSCs	induced pluripotent stem cells
IRES	internal ribosome entry site
ITAFs	IRES trans-acting factors
mRNA	messenger RNA
NLS	nuclear localisation signal
OMRV	Omono River virus
ORF	open reading frame
OpbuCPV18	Operophtera brumata cypovirus-18
PnPV	Perina nuda picorna-like virus
PTV-1	porcine teschovirus-1
PV	poliovirus
RFP	red fluorescent protein
RtPV	Rattus tanezumi parechovirus
RNA	ribonucleic acid
SRP	signal recognition particle
TALEN	transcription activator-like effector nucleases
TaV	Thosea asigna virus
TCR	T-cell receptor
TEV	tobacco etch virus
TLR	Traffic Light Reporter
ToV-TJ	Tianjin totivirus
tRNA	transfer RNA
VAD	vitamin A deficiency
WP-LV	Wenzhou picorna-like virus
WCP	Wuhan carp picornavirus
YFP	yellow fluorescent protein
ZFN	zinc finger nucleases