Protein Coexpression Using FMDV 2A: Effect of “Linker” Residues

Many biomedical applications absolutely require, or are substantially enhanced by, coexpression of multiple proteins from a single vector. Foot-and-mouth disease virus 2A (F2A) and “2A-like” sequences (e.g., Thosea asigna virus 2A; T2A) are used widely for this purpose since multiple proteins can be coexpressed by linking open reading frames (ORFs) to form a single cistron. The activity of F2A “cleavage” may, however, be compromised by both the use of shorter versions of F2A and the sequences (derived from multiple-purpose cloning sites) used to link F2A to the upstream protein. To characterise these effects, different lengths of F2A and T2A were inserted between green and cherry fluorescent proteins. Mutations were introduced in the linker region immediately upstream of both F2A- and T2A-based constructs and activities determined using both cell-free translation systems and transfected cells. In shorter versions of F2A, activity may be affected by both the C-terminal sequence of the protein upstream and, equally strikingly, the residues immediately upstream introduced during cloning. Mutations significantly improved activity for shorter versions of F2A but could decrease activity in the case of T2A. These data will aid the design of cloning strategies for the co-expression of multiple proteins in biomedical/biotechnological applications.


Introduction
Many biomedical applications require vectors that can direct the expression of multiple proteins; subunits of hetero-multimeric proteins, multiple therapeutic genes (combined and/or synergistic effects), or, simply, coexpression of a therapeutic protein along with proteins that act as (selectable) markers of transformed cells [1,2]. A number of approaches are used to coexpress multiple genes, including fusion proteins (which may include proteinase cleavage sites), alternative mRNA splicing, multiple promoters, reinitiation of translation, and internal ribosome entry sites (IRESes). Each, however, has associated disadvantages: fusion proteins localise to only a single subcellular site, while steric hindrance may alter their function. If a proteinase cleavage site is incorporated, this requires colocalisation of the substrate and processing enzyme in the same subcellular site. Internal promoters frequently show interference or are downregulated, while expression from IRESes (dependent on various cellular binding factors) varies between different cell types. Although derived from a single bicistronic mRNA, expression of the downstream ORF (IRES-driven cap-independent translation) is typically ∼10% of that of the upstream ORF (capdependent translation). IRES elements, identified both in viral and cellular eukaryotic mRNAs, differ in nucleotide length (from 130 bp to 1 kb). However, the most efficient viral IRESes successfully utilized in vectors used for biomedical purposes are about 500 bp in length. Their comparatively large size can be a limiting factor when using virus-based vectors which have limited coding capacity: adeno-associated vectors cannot package more than ∼5 kb efficiently, whilst retroviral vectors can package only ∼7-8 kb [1][2][3][4][5][6][7].
Foot-and-mouth disease virus 2A (F2A) and "2Alike" sequences have become a useful alternative to these approaches since multiple proteins can be coexpressed at equimolar amounts from a single transcript mRNA under the control of a single promoter. 2A mediates a cotranslational "ribosome skipping" event (for simplicity referred to as "cleavage"), to produce the C-terminus of 2A. Interestingly, the length of 2A in the FMDV polyprotein (18aa) is defined by the site of the skipping event (forming the C-terminus of F2A), plus the N-terminus delineated by the site where 2 BioMed Research International a virus-encoded proteinase (3C pro ) trims "2A" from the upstream capsid protein 1D at a later stage in virus replication. We have shown, however, that the functional length of "2A" actually incorporates (capsid protein 1D) sequences upstream of 2A. The longer versions of 2A described below are therefore 2A plus N-terminal extensions of the upstream capsid protein 1D, but for simplicity referred to as F2A [8][9][10][11].
The major advantages of using the 2A system in the construction of multicistronic vectors are (i) its small size (54-174 bp) compared to IRESes, (ii) that coexpression of proteins linked by 2A is independent of the cell type (since cleavage activity is only dependent on eukaryotic ribosomes, structurally highly conserved amongst the eukaryota), and (iii) that multiple 2As may be used, the activity of each being completely independent. In the case of F2A, it was demonstrated that although good cleavage was observed using the 19aa version, the use of longer versions of 2A was reported to produce higher levels of cleavage [8][9][10][11].
To date, F2A sequences of various lengths have been utilised in biotechnology, from 18 to 58aa [24][25][26][27][28][29][30][31]. In heterologous (artificial polyprotein) contexts, the 2A sequence remains as a C-terminal extension of the upstream gene: potentially a consideration when using longer, albeit more efficient, F2A sequences. C-terminal extensions may affect protein conformation, whilst an "authentic" C-terminus may be crucial for activity or for posttranslational modification. Shorter versions of 2A have been used, although often at the expense of cleavage efficiency [9,12,24,27,28]. In the case of proteins entering the exocytic pathway, C-terminal extensions of 2A may be "trimmed" away by the incorporation of a furin cleavage site between the upstream protein and 2A [24,32]. Furthermore, it has been reported that in some cases the efficiency of 2A cleavage may be greatly inhibited when coexpressing certain proteins targeted to the exocytic pathway. Cterminal sequences of the upstream gene were implicated to be responsible for these lower levels of cleavage [33].
To further optimise the 2A coexpression system, we used T2A and shorter versions of F2A inserted between GFP and CherryFP fluorescent "reporter" proteins. Mutations were introduced into the region immediately upstream of both F2A-and T2A-based constructs. The effects on activity were determined using both cell-free coupled transcription/translation systems in vitro and in transfected HeLa cells, a human cell line most commonly used in biomedical applications. Activity can be affected in two ways: (i) by the C-terminal region of the protein upstream of F2A (where shorter F2As are utilised) and, significantly, (ii) by the short amino acid tract linking the protein upstream with F2A introduced by the cloning strategy. In the case of shorter versions of F2A, in some cases mutation significantly increased activity. Conversely, in the case of the highly active T2A, some mutations decreased activity. Taken together, these data show that sequences within 30-40aa upstream of the cleavage site may be highly influential in determining 2A activity: particular care should be taken in this aspect in the design of constructs for coexpression of multiple proteins prior to their further development in biomedical applications.

Construction of Plasmids
pGFP-F2As-CherryFP Plasmid Construct Series. Here, F2As of various lengths (58, 30, 25-20, and 18aa) were inserted between sequences encoding GFP and CherryFP (Table 1). To make the pGFP-F2A 58 -CherryFP construct, the sequence encoding the 58aa version of F2A was obtained by digesting pSV7 plasmid (kind gift of Sandra Vater) with NsiI and ApaI. This insert was then used to replace the TaV 2A in pJC3, similarly restricted. Plasmid pGFP-F2A 58 -CherryFP was then used as a template to produce further truncated forms of F2A. Briefly, different lengths of F2A were amplified by PCR, creating an XhoI site at the N-terminus and a NotI site present downstream of the CherryFP in both plasmids.   Table 1. These forms were produced by recombination in competent JM109 cells of two PCR products. The first was amplified with forward primer encoding the mutated sequences and a reverse primer binding to the vector backbone at nt 4500 (5 -GTTCCACTGAGCGTCAGACCCCGTAG-3 ). The second was amplified using a reverse primer encoding the mutated sequence and a forward primer binding to the vector backbone at nt 4500 (5 -CTACGGGGTCTGACGCTC-AGTGGAAC-3 ). Forward and reverse primer sets for each construct are shown in Table S1 (see Table S1 in the species list in the Supplementary Material available online at http://dx.doi.org/10.1155/2013/291730).

Transcription and Translation.
Plasmid constructs were used to programme Quick TnT transcription/translation reticulocyte lysate systems (Promega). Translation reactions (10 L) were performed according to the manufacturer's instructions. Briefly, 20 ng of plasmid DNA was used to programme rabbit reticulocyte lysates containing 35 S methionine (10 Ci) and incubated at 30 ∘ C for 1 h. Translation reactions were stopped by the addition of 2x protein sample buffer. Protein products were analysed by 12% SDS-PAGE.

Expression in HeLa Cells.
HeLa cells were maintained in DMEM medium supplemented with 10% FCS. The cells were transfected in 60 mm dishes (preplated 20 h earlier to 60% confluency) with 1.5 g of plasmid DNA and 7 L of Lipofectamine2000 (Invitrogen) in a final volume of 400 L OptiMEM. Transfection mix was added to 2 mL of antibioticfree serum-containing medium and incubated for 5 h. Cells were further incubated for a total of 30 h after-transfection after addition of 2 mL of media containing 10% FCS.  Figure 1: Sequence comparison between pGFP-F2A 20 -CherryFP and pGFP2AGUS constructs. While the downstream genes were different (CherryFP or GUS, resp.), both constructs had the same upstream GFP gene and F2A of 20aa. The only difference was in the sequence immediately upstream of F2A due to different restriction sites introduced for the specific cloning purposes.

Western
Triton X-100, 1% Na deoxycholate, 0.1% SDS), and freshly added 1/20 volume of EDTA-free protease inhibitor cocktail (Sigma) by incubation on ice for 30 min and centrifugation at 12000 rpm for 20 min at 4 ∘ C. Cellular debris was discarded and supernatants containing proteins were analysed by 12% SDS-PAGE. The proteins were then transferred to a nitrocellulose membrane, which was probed with anti-GFP or anti-CherryFP primary antibodies overnight, after blocking in PBST containing 5% milk for 1 h. Following overnight incubation, the membranes were washed three times in PBST. Detection of bound primary antibodies was performed with respective HRP-conjugated secondary antibodies (Dako), in PBST containing 5% milk for 1 h. The membranes were washed three times with PBST, rinsed in deionised water, and subjected to enhanced chemiluminescence by incubating the membranes in freshly prepared visualisation solution for 2 min. The membranes were exposed to an autoradiograph film for 3 to 45 seconds.

Substitutions Immediately Upstream of F2A (20aa Version;
F2A 20 ): Effect on Cleavage Efficiency. Previously we showed that the 20aa version of F2A encoded by pGFP2AGUS was highly active [8,27]. Indeed, this version has been used widely in other laboratories. To our surprise, the cleavage activity observed for the same length of F2A used to link GFP and CherryFP (pGFP-F2A 20 -CherryFP) was noticeably lower (Figures 1 and 2(c)). The upstream GFP and the F2A (20aa) were identical in both constructs: the only difference was the short "linker" between GFP and F2A created by the cloning strategy: -RAKRSLE-(pGFP-F2A 20 -CherryFP; linker derived from furin and XhoI restriction site) and -SGSRGAC-(pGFP2AGUS; linker derived from XbaI and SphI restriction sites), with only the arginine (shown in bold type face) in common ( Figure 1).
To investigate the reason underlying the different cleavage activities, we produced a set of three pGFP-F2A 20 -CherryFP mutant constructs in which the sequence of this linker was converted, stepwise (changed residues underlined), to the pGFP2AGUS linker sequence: mut1 (-RAKRSLE-substituted to -SGSRSLE-), mut2 (-RAKRSLEto -RAKRGAC-), and mut3 (-RAKRSLE-to -SGSRGAC-) with the mut3 sequence being identical to that of the more active pGFP2AGUS sequence (Figures 1 and 2(a)). Coupled transcription/translation analyses using in vitro systems produced five translation products. The major translation products corresponded to proteins of predicted molecular weights: uncleaved [GFP-F2A 20 -CherryFP] and the two cleavage products, [GFP-F2A 20 ] and CherryFP, the latter two products comigrating on the gel due to similar mass ( Figure 2(b)). The ratio of the cleaved : uncleaved products was higher in mutants 2 and 3, demonstrating higher cleavage activity in these constructs. Translation profiles derived from all four constructs also showed a certain level of internal initiation within GFP, producing the N-terminally truncated cleavage product [ΔGFP-F2A]. On the basis of the molecular mass of the product and distribution of methionine codons within GFP, we assume that this low-level internal initiation occurred at Met88-in a favourable Kozak context (cCAUGc). There was also a slower migrating band, corresponding to a protein of ∼45 kDa, which was also present in western blots (Figure 2 Data obtained from both in vitro translation and in vivo (transfected HeLa cells) using these four constructs showed that in the case of pGFP-F2A 20 -CherryFP, introduction of the -SGS-substitution (mut1) somewhat improved cleavage activity. Introduction of either the -GAC-substitution adjacent to the N-terminus of F2A (mut2) or the -SGSRGACsubstitution (mut3) resulted, however, in considerably higher cleavage activity, demonstrated by the increased ratio of cleaved : uncleaved products (more evident in western blots with extended exposures; Figures 2(b) and 2(c)).  The effects of amino acid substitutions were determined using translation in vitro and in HeLa cells transfected with plasmid DNAs and western blotting (Figures 3(b) and 3(c)). The cleavage activity (ratio of cleaved : uncleaved forms) of all mutant constructs was higher (more apparent in western blots with extended exposures (see Figure S1)). This difference was not as significant in longer versions (24/25aa) compared to shorter versions (18/20/21aa): the cleavage activity of the shortest version (18aa) was the most "improved. " Overall, the improved mutant constructs with resulting high cleavage efficiency followed the previously described "cleavage pattern" [27] and Minskaia et al., manuscript in preparation, where stepwise amino acid deletions from the N-terminus of F2A resulted in linear increase in accumulation of uncleaved polyprotein and a decrease in the amounts of cleaved products ( Figure 4). This pattern, however, was changed in In order to more finely map the effect of these mutations, the -RAKR-motif within the 18 and 20aa versions of pGFP-F2A-CherryFP (encoding the least active F2As) was mutated to -AAKA-, producing pGFP-F2A 20Fm -CherryFP and pGFP-F2A 18Fm -CherryFP (Figures 3(a) and Table 1). Interestingly, the cleavage activity for the pGFP-F2A 18Fm -CherryFP construct was as low as that of the parental pGFP-F2A 18 -CherryFP construct, demonstrating that the improved cleavage efficiency of pGFP-F2A 18mut -CherryFP with -SGSRGAC-substitution was, therefore, attributable to the nature of the linker sequence (Figures 3(c) and Table 1). On the contrary, introduction of the -AAKA-substitution in the F2A 20 construct resulted in an increase in cleavage efficiency comparable to that of the mutant construct pGFP-F2A 20mut -CherryFP with the -SGSRGAC-substitution. These results demonstrate that the difference in the 2 residues at the N-terminus of F2A (pGFP-F2A 20Fm -CherryFP versus pGFP-F2A 18Fm -CherryFP), together with the difference in the 3 residues immediately upstream of F2A (-SGSRGACversus  25mut  24mut  23mut  21mut  18mut  20mut3  HeLa  25mut  24mut  23mut  21mut  18mut  20mut3  HeLa   Tubulin Anti-GFP Anti-CherryFP Extended exposure Figure 4: Efficiency of F2A cleavage in pGFP-F2A-CherryFP "mutants" with -SGSRGAC-substitution immediately upstream of F2A of 25-18aa. pGFP-F2A-CherryFP "mutant" constructs were used to coexpress GFP and CherryFP proteins from a single ORF in transfected HeLa cells. The cells were transfected with 1.5 g of plasmid DNA and harvested 30 h after transfection. Cells were lysed in RIPA buffer, and equal amounts of total protein for each transfection were loaded onto 12% SDS-PAGE gel. The proteins were transferred onto a nitrocellulose membrane, blocked in PBS containing 5% milk, and probed with anti-GFP (upper blot) and anti-CherryFP (middle blot) antibodies overnight at 4 ∘ C. Detection of bound primary antibody was achieved by using respective secondary antibodies, followed by ECL detection. All experiments were done in triplicate.

Substitutions Immediately
-AAKASLE-in the F2A 18 construct), can have a profound effect on cleavage efficiency in the case of these short F2As (Figures 3(c) and Table 1).

Introduction of Mutations Immediately Upstream of T2A
May Decrease Activity. The T2A sequence (18aa) is highly active and has not been reported to be susceptible to the nature of the C-terminus of the upstream gene. To test if T2A was susceptible to substitutions immediately upstream, we introduced "reverse" substitutions (mut1: MHSRGSGto -MHSRSLE-and mut2: -MHSRGSG to -RAKRSLE-) in the same [GFP-2A-CherryFP] context (Figures 5(a) and Table 1). The effect on cleavage activity was determined using reticulocyte lysate in vitro translation systems ( Figure 5(b)) and transfected HeLa cells ( Figure 5(c)). These substitutions resulted in decreased cleavage activity in both mutant constructs, more apparent in HeLa cells than in reticulocyte lysates, as detected by both anti-GFP ( Figure 5(c), left blot) and anti-CherryFP ( Figure 5(c), right blot) antibodies. Introduction of the -SLE-substitution immediately upstream of T2A (mut1; Table 1) had even greater effect on activity than the -RAKRSLE-substitution (mut2; Table 1; Figure 5(c)). Again, this observation reveals the importance of the 3 residues immediately upstream of such short F2As. It is worth noting that in our parental construct, the -MHSRGSGsequence comprises 3 restriction sites used for cloning (NsiI, XbaI, and BspEI). Coincidentally, a -GSG-linker motif was recently shown to improve cleavage efficiency ( [12,34], discussed below). , was comparable to that of the T2A wt -and T2A mut2versions, with virtually no uncleaved polyprotein present in HeLa cells. Substitutions immediately upstream of T2A (T2A mut2 ) resulted in decreased coordinate expression of both genes. This construct was still highly active compared to F2A 20wt -and F2A 18wt -based versions with identical amino acids immediately upstream of F2A. Their substitution to the more efficient F2A 20mut3 -and F2A 18mut -based constructs resulted in significant increase in the proportion of cleaved products. Some uncleaved product was, however, observed in HeLa cells transfected with these constructs.

Discussion
The model for 2A-mediated cleavage predicts that the nascent 2A interacts with the exit tunnel of the ribosome to affect the conformational space occupied by the ester bond linking the nascent protein and tRNA gly in the ribosome P-site. The model predicts, therefore, that residues that may influence activity reside within the exit tunnel of the ribosome [35][36][37].  amino acids [38]. The activity of shorter (25-18aa) forms of F2A may, therefore, be affected by the nature of the Cterminal sequence of the protein upstream since this region would now lie within the exit tunnel of the ribosome and affect its interaction with F2A. The aim of this study was to further characterise and improve the efficiency of the 2A coexpression system by introducing various mutations immediately upstream of shorter versions of F2A and T2A linking GFP and CherryFP proteins. Based on the results obtained in vitro in reticulocyte lysates and in HeLa cells transfected with plasmid DNAs, we demonstrate that introduction of the -SGS-substitution (mut1) 4 amino acids upstream of 20aa-long F2A slightly improved cleavage efficiency, while introduction of either -GAC-substitution adjacent to the N-terminus of F2A (mut2, Table 1) or -SGSRGAC-substitution (mut3, Table 1) immediately upstream of F2A resulted in considerably higher cleavage efficiencies (Figures 2(b) and 2(c)).
We further demonstrate that in the case of the shortest F2As (21/20/18aa), substitution of the parental -RAKRSLE-sequence with -SGSRGAC-greatly improved cleavage efficiencies (Figures 3(a)-3(c), 4, and Table 1), an effect somewhat less evident in the case of longer F2As (25/23aa). The 18aa version, which showed relatively low activity, was dramatically improved by the introduction of these mutations. This increase in activity may be attributed to the -SLEto -GAC-exchange since the (furin linker) -RAKR-to -AAKA-exchange had no detectable effect. However, in the case of the 20aa version, identical substitutions within either portion of the linker (-RAKR/SLE-) only slightly increased activity (Figure 3(c)). Introduction of either -SGSRGAC-or -AAKASLE-substitutions in the 20aa F2A version resulted in comparable increase in the cleavage activity. These results show that just 2aa at the N-terminus of "suboptimal" 2As (F2A 20 , F2A 18 ), and the 3 residues immediately upstream of 2A (F2A 18mut versus F2A 18Fm ), can have a profound effect on F2A cleavage efficiency (Figure 7). We further showed that introduction of the "reverse" substitutions of -MHSRGSGto -MHSRSLE-(mut1, Table 1) and to -RAKRSLE-(mut2, Table 1) in the highly active T2A-based construct resulted in  Figure 6: Efficiency of cleavage in constructs with short 2A sequences. pGFP-2A-CherryFP constructs with F2As 30 wt , 20 wt , 20 mut3 , 18 wt , 18 mut , and T2A wt and T2A mut2 were used to coexpress GFP and CherryFP proteins from a single ORF in transfected HeLa cells. The cells were transfected with 1.5 g of plasmid DNA, and harvested 30 h after transfection. Cells were lysed in RIPA buffer, and equal amounts of total protein for each transfection were loaded onto 12% SDS-PAGE gel. The proteins were transferred onto a nitrocellulose membrane, blocked in PBS containing 5% milk, and probed with anti-GFP (upper blot) and anti-Cherry (middle blot) antibodies overnight at 4 ∘ C. Detection of bound primary antibody was achieved by using respective secondary antibodies, followed by ECL detection. All experiments were done in triplicate.
decreased cleavage activity in both mutant constructs (Figures 5(a)-5(c)). Again, these data highlight the importance of the 3aa immediately upstream of T2A (in this instance). Finally, we show that cleavage efficiencies of F2A 30 -, T2A wt -, and T2A mut2 -based constructs were similar, with virtually no uncleaved form present in HeLa cells ( Figure 6). While introduction of the "reverse" substitutions upstream of T2A resulted in decreased coordinate expression of both genes (T2A mut2 ), it was still highly active compared to F2A 20wt and F2A 18wt with identical upstream residues. Their substitution to more "optimal, " F2A 20mut3 and F2A 18mut , resulted in significant increase in the amounts of cleaved products. The presence of some uncleaved translation product does not, however, compare favourably with the "high-cleavers" F2A 30wt and T2A wt .
The data presented here is in agreement with previously published studies that successfully used F2A sequences of various lengths in in vitro and in vivo heterologous systems [24,25,[27][28][29]31]. To avoid using longer F2A sequences that remain attached at the C-terminus of the upstream protein, researchers have used shorter F2As but have reported a range of cleavage activities for F2As of the same length [9,12,24,27,28,39]. Klump and colleagues [40] coexpressed the homeobox transcription factor along with GFP [GFP-F2A 24 -HoxB4]. The 24aa version of F2A produced small amounts of uncleaved translation product in cells-similar to the data we present here. Based on this work, Milsom and colleagues [41] first tested two constructs, pHoxB4-2A-GFP and pMGMT-2A-GFP, coexpressing HoxB4 and MGMT (O 6methylguanine-DNA-methyl-transferase, a transgene that conveys chemoresistance) with GFP and noted that in cells transfected with these constructs the amount of uncleaved [MGMT-F2A 24 -GFP] polyprotein was far greater than that of [HoxB4-F2A 24 -GFP]. Noting that the C-terminal sequence of the upstream protein could be the reason for decreased cleavage efficiency, a tricistronic retroviral vector pHoxB4-F2A 24 -MGMT-IRES-GFP was developed to simultaneously coexpress the three proteins in transfected bone marrow. With HoxB4 as the upstream gene, this tricistronic vector was concluded to be the best for coexpression of both genes, with insignificant amount of uncleaved polyprotein [41]. Similarly, using a 24aa version of F2A, de Felipe and Ryan [39] produced a series of constructs for coexpression of various fluorescent proteins (some including internal cotranslational signal sequences). Uncleaved polyprotein was detected, particularly for constructs encoding proteins targeted to the exocytic pathway [39]. This lowering of activity was attributed to interaction of the nascent peptide with the translocon pore (Sec61 complex) also affecting the interaction of F2A with the ribosome exit tunnel [33]. In the case of proteins targeted to enter the exocytic pathway, successful utilization of a 24aa version F2A with a furin cleavage site immediately upstream was reported by several groups [24,32].
This approach was used to link the antibody heavy and light chain sequences to engineer a mAb expression cassette that, in the context of AAV-mediated gene transfer, resulted in high levels of full-length, functional monoclonal antibodies in vitro and in vivo [28]. Using the same strategy, significant antitumour responses have been observed in the clinic using monoclonal antibodies that block immune checkpoints by coexpressing CTLA-4 heavy and light chains [42], while Camper and colleagues [32] designed a retroviral vector for the expression of a target Fab' fragment as a single polyprotein and showed equimolar expression and processing of the light and truncated heavy chain domains [32].
As for shorter F2As, different cleavage activities (depending on protein arrangement) were observed when GFP, EGFP, and hrGFP were coexpressed with cytochrome P450 2B1 (CYP2B1) using a 20aa version of F2A [43]. With a 19aa version of F2A, cleavage was reported to be less efficient (70%) for [CFP-F2A 19 -PAC], with some uncleaved polyprotein still present [44]. Similar results were achieved when an 18aa version of F2A was used for the generation of retroviral vectors [45]. Ma and Mitra [46] used even shorter F2A of 17aa to coexpress GUS and CAT proteins (in alternative gene orders) in planta and demonstrated that [GUS-F2A 17 -CAT] and [GUS-F2A 17 -GUS] were cleaved to completion, while [CAT-F2A 17 -GUS] was not. They concluded that the difference in cleavage efficiency was due to different upstream genes. In light of our experiments, however, it is noteworthy to mention that, based on the data provided, these constructs also differed in the sequence immediately upstream of 2A: -QQGGKQVD-for GUS-F2A 17 and -EWQGGAVD-for CAT-F2A 17 [46].
Interestingly, cleavage efficiency of shorter F2As in different contexts was shown to be improved by insertion of various spacer sequences, such as -SGS-or -GSG- [12,34,47,48], the V5 epitope tag (-GKPUPNPLLGLDST-) [24], or a 3xFlag epitope tag [31] immediately upstream of F2A. It has been reported that in -retroviral constructs, expression of multiple genes linked with F2A peptides was facilitated by a spacer sequence (-SGSG-or -GSG-) immediately upstream of F2A 22 [12,34], while Yang and colleagues [24] tested both these spacers plus the V5 tag which acted, in this context, as a spacer. In these studies, lentiviral vectors expressing two-gene T-cell receptors directed against the melanoma differentiation antigens gp100 and MART-1 were constructed using F2A 24 . The addition of spacer sequences was, however, shown to be a prerequisite for efficient synthesis and assembly of biologically active T-cell receptor complexes. Interestingly, when the spacer sequence in the form of V5 tag was inserted between furin and F2A 24 , this increased activity and yielded optimal TCR expression in transduced lymphocytes. At the same time, attempts to substitute the V5 peptide with other sequences were unsuccessful, suggesting that the specific amino acid sequence of the V5 tag was responsible for enhanced ribosomal skipping in this context [24]. Recently, Tan and colleagues [31] successfully used a similar strategy by providing a spacer in the form of 3xFlag tag repeats (incorporated for other purposes) upstream of an 18aa version of F2A to express double or triple copies of the rabies virus glycoprotein from a single ORF in human adenovirus 5 (linked via F2As and followed by hrGFP as a marker after IRES).

Conclusions
2A sequences have enabled more complex strategies of coexpression and transgenesis and have been used in many laboratories with highly successful results. Based on the data presented here and from previous studies, the variations in cleavage activity of the various lengths of F2As are attributed to the nature of the region immediately upstream of F2A (18-30aa upstream of the cleavage site), altered by residues introduced by the cloning strategy, or, by the introduction of spacer sequences. Indeed, in the case of the highly active T2A, changes in this region can result in a decrease of cleavage efficiency. Such variations in cleavage activity can, however, be manipulated to great purpose to produce "molecular rheostats. " In a very elegant series of experiments, Yu and colleagues produced a range of truncated and point-mutated forms of F2A and I2A (from the insect infectious flacherie virus) to regulate the ratio of surface-anchored to secreted forms of the immunoglobulin IgG [50]. The data presented in this paper (together with others), therefore, aids the design (and troubleshooting) of coexpression strategies but also serves to show how manipulation of these sequences can be used in expanding the range of such molecular rheostats.