Introduction

Since the first chicken embryo explants by Roux in 18851, cultured cells have been used as a tool to study biological processes for more than a century. A vast number of techniques have been developed for the analysis of mammalian cultured cells, including the establishment of stable cell lines that express DNAs, RNAs and proteins of interest. In recent years, approaches using retroviral and lentiviral-derived constructs to generate stable cell lines have been increasingly common2. The most extensive application of insect cell lines is in protein production, in which recombinant baculoviruses are used to infect lepidopteran cell lines to achieve high transient expression levels of proteins of interest3. In Drosophila, cell culture is an often-overlooked complement to the powerful genetic tools available in this research model. The utility of cell-based assays coupled with genome-wide RNAi or chemical library screening4,5 and the promising potential of using Drosophila S2 cells for recombinant protein production6,7 underlines the need for improved vector systems for working with Drosophila cells.

A number of strategies have been used to generate stably transformed Drosophila cell lines8. Cotransfection of the expression plasmid of interest with a second plasmid encoding drug resistance has been used9,10,11,12,13, as well as single-vector strategies with two separate expression cassettes14,15. P-element-mediated transformation has also been used to incorporate selectable markers in cell lines16. In all cases, however, drug resistance can be preferentially selected over expression of the protein of interest, often necessitating single-cell cloning and screening of multiple stable clones. Viral-derived internal ribosome entry sites (IRES) have been used to develop multicistronic vectors in mammalian cells17,18 and although IRES sequences have been described in insect systems19,20,21, a similar approach has not been adopted to Drosophila cell culture vectors. Moreover, the use of IRES sequences to achieve multicistronic expression can also lead to variable levels of the expressed proteins22,23.

“2A-like” sequences, also called CHYSEL (cis-acting hydrolase element) peptides, are found in viral polyproteins and serve as signals to direct apparent self-cleavage of the polyprotein into individual proteins. The sequences act on the ribosome peptidyltransferase center, causing release of the nascent chain while sense codons are still present, described as “ribosome skipping”24,25,26,27. The 2A-like sequences contain a canonical Asp-Val/Ile-Glu-X-Asn-Pro-Gly(2A)Pro(2B) motif, which results in separation between the Gly(2A) and Pro(2B)28. These sequences have been used for the generation of polycistronic vectors expressing multiple genes of interest in mammalian cells and transgenic animals, avoiding the use of internal ribosomal entries (IRES) or multiple promoters29,30,31,32,33.

Here, we show that T2A, a 2A-like sequence from the insect virus Thosea asigna, is correctly processed in Drosophila cells and describe new vectors (pAc5-STABLEs) that use T2A for the efficient production of stable cell lines, opening new possibilities for insect cell culture applications.

Results

New vectors for stable selection in Drosophila cells

To obtain strong constitutive expression, we chose the pAc5.1 vector (Invitrogen), which contains the Drosophila Actin5c promoter, as a backbone for our new vectors. As a negative control for antibiotic selection, we used pAc5–GFP (Fig 1A), which expresses GFP, but does not contain a selectable marker. Several 2A-like peptides have been used successfully in polycistronic vectors for mammalian cells, but we reasoned that for insect cells, the best candidate was the T2A peptide sequence, derived from the insect cell virus Thosea asigna26. We generated pAc5-STABLE1-Neo (Fig. 1B), in which the sequence encoding the T2A peptide was cloned in frame between the GFP and the NeoR genes. The NeoR protein confers cellular resistance to the aminoglycoside antibiotic G418. Cells that survive antibiotic selection should therefore express the protein of interest in an obligate manner.

Figure 1
figure 1

Schematic representation of the p Ac5-STABLE vectors.

All constructs were based on the vector backbone of pAc5.1 (Invitrogen), with expression driven by the Actin5C promoter. (A) In pAc5-GFP, GFP alone served as a negative control for selection experiments. (B) In pAc5-STABLE1-Neo, designed for bicistronic expression, GFP and NeoR are separated by the T2A sequence. Unique sites are shown. (C) pAc5-STABLE2-Neo, designed for tricistronic expression, features FLAG-tagged mCherry, GFP and NeoR, each separated by a T2A peptide. In the case of dT2A, it has identical peptide sequence to T2A, but degenerated sequence at the nucleotide level. Unique sites are shown. (D) pAc5-dTES.GFP-Neo was constructed by cloning Drosophila CG6522 into the EcoRI-XbaI sites of pAc5-STABLE1-Neo, generating an N-terminal fusion to GFP.

Full size image

For cases in which there are two proteins of interest to be expressed at the same time, we generated pAc5-STABLE2-Neo (Fig. 1C). As pAc5-STABLE1-Neo, it contains the T2A peptide in frame between GFP and NeoR. It also contains a sequence encoding a FLAG epitope-tagged version of mCherry fluorescent protein34 fused C-terminal to the FLAG tag. This is separated from GFP by an additional T2A peptide, encoded by degenerate nucleotide sequence to prevent vector recombination. The vectors were designed for flexibility, to allow replacement of FLAG-Cherry, GFP, or NeoR with any gene of interest, or to allow the generation of N-terminal or C-terminal fusions to the fluorescent proteins. To demonstrate the utility of the vector with a novel protein, the coding sequence of Drosophila CG6522 (ortholog of the focal adhesion protein TES,35,36; hereafter called dTES) was cloned at the N-terminus of GFP in pAc5-STABLE1-Neo (Fig. 1D).

Generation of stable transformants using the pAc5–STABLE1-Neo vector

To test the vectors, we chose the cell line S2R+37, an isolate of embryo-derived Schneider's line 238, since it is generally more adherent to cell culture vessels than parental S2 cells and has been used successfully in numerous RNAi screenings. We tested pAc5-STABLE1-Neo for positive selection against G418 compared to pAc5–GFP or non-transfected cells as controls. Equal quantities of the three different vectors were introduced into the cells, which were treated with 0 (Fig. 2A, D, G), 600 (Fig. 2B, E, H), or 2000 μg/ml (Fig. 2C, F, I) of G418 during four weeks. While very few control cells survived G418 selection for this period, cells transfected with pAc5-STABLE1-Neo continued growing even in the highest concentration (Fig. 2I). As anticipated, most of selected cells also expressed GFP, distributed homogenously throughout the nucleus/cytoplasm. Using image quantitation (which may underestimate weak expressors; see Methods), more than 68% of cells surviving in 2000µg/ml G418 also expressed GFP (Fig. 2L). This proportion was higher than non-selected cells or those with intermediate G418 selection (Fig. 2J–K). GFP intensity levels between cells were variable as detected by fluorescence microscopy, perhaps reflecting copy number of integrated vectors. Higher concentrations of G418 were not tested and could potentially lead to more homogenous GFP levels. We conclude that vector pAc5-STABLE1-Neo can effectively confer G418 resistance to the GFP-expressing cells and that it can be used for the generation of stable cell lines.

Figure 2
figure 2

pAc5-STABLE1-Neo confers G418 resistance to S2R+ cells.

(A–I) Bright field images of S2R+ Drosophila cells that were mock-transfected (Ctrl; A–C), transfected with pAc5-GFP (D–F) or with pAc5-STABLE1-Neo (G–I). Three days after transfection, G418 was added to the media at concentrations of 0 (A, D, G), 600 (B, E, H) or 2000 μg/ml (C, F, I). (A'–I') Fluorescent images of the same cells from panels A to I showing GFP expression. (J–L) Percentage of surviving cells that are positive for GFP. Letters correspond to panels/treatments directly above each graph. All images were taken after 30 days of treatment with G418.

Full size image

Positive selection shows that NeoR is correctly expressed and functionally active, but it could be functional even if expressed as an unprocessed GFP-T2A-NeoR fusion protein. To discard this possibility and confirm that the T2A peptide was functioning, transfected cells were analysed by Western blot (Fig. 3). Using anti-GFP antibodies, the proteins show the right size of 27–30 kDa when expressed from both vectors pAc5-GFP and pAc5-STABLE1-Neo. Due to cloning strategy, GFP from pAc5-GFP contains extra amino acids at the C-terminus and migrates slightly slower than the GFP encoded by pAc5-STABLE1-Neo. No upper band corresponding to unprocessed GFP-T2A-NeoR was observed, even in longer exposures (data not shown). These results indicate that the T2A sequence functions as expected and undergoes apparent self-cleavage in Drosophila S2R+ cells.

Figure 3
figure 3

T2A is correctly processed in Drosophila cultured cells.

Western blot analysis of S2R+ cells transfected either with pAc5-STABLE1-Neo or pAc5-GFP vectors, probed with anti-GFP antibodies. Expected sized of non-processed GFP-T2A-neo is shown (58kDa; arrow). The slight difference in size between the GFP proteins derived from the two plasmids is caused by additional polylinker-derived amino acids in the case of pAc5-GFP. Molecular weights in kDa are indicated to the left.

Full size image

To confirm that the effect was not specific to S2R+ cells, we also tested pAc5-STABLE1-Neo in an additional Drosophila cell line, Kc16739. This line is derived from the Kc line, established from 6–12 hr female embryos. As before, we tested pAc5-STABLE1-Neo for positive selection against G418 compared to pAc5–GFP or non-transfected cells as controls. The Kc167 cells were treated with 0 (Fig. 4A, D, G), 600 (Fig. 4B, E, H), or 2000 μg/ml (Fig. 4C, F, I) of G418 during four weeks. Similar to our results with S2R+ cells, cells transfected with pAc5-STABLE1-Neo survived to a greater extent than those transfected with the control plasmids (more than 60%; Fig. 4L). This indicates that both GFP and NeoR are functional and suggesting that T2A apparent self-cleavage is properly achieved in Kc167 cells as well.

Figure 4
figure 4

pAc5-STABLE1-Neo confers G418 resistance in Kc167 cells.

(A–I) Bright field images of Drosophila Kc167 cells that were mock-transfected (A–C), transfected with pAc5-GFP (D–F) or transfected with pAc5-STABLE1-Neo (G–I). Three days after transfection, G418 was added to the media at concentrations of 0 (A, D, G), 600 (B, E, H) or 2000 μg/ml (C, F, I). (A'–I') Fluorescent images of the same cells from panels A to I showing GFP expression. (J–L) Percentage of cells that are positive for GFP. Letters correspond to panels/treatments directly above each graph. All images were taken after 30 days of treatment with G418.

Full size image

Generation of stable cell population using the pAc5-STABLE2-Neo

Multiple self-cleavage peptides can be processed in the same cell, with published examples of up to 4 functional proteins being generated from a single polyprotein containing three 2A-like sequences29,40. To test whether more than two proteins can be processed from a multicistronic vector in insect cells, we generated the vector pAc5-STABLE2-Neo. Again, we used the Thosea asigna T2A peptide to separate the proteins. S2R+ cells were transfected with pAc5-STABLE2-Neo and compared to pAc5-GFP or mock-transfected cells as controls. Cells were treated with 0 (Fig. 5A, D, G), 600 (Fig. 5B, E, H), or 2000 μg/ml (Fig. 5C, F, I) of G418 during four weeks. Cells transfected with pAc5-STABLE2-Neo survived G418 treatment at the highest concentration and all surviving cells had detectable levels of GFP and FLAG-mCherry after four weeks of treatment. As expected, a high degree of overlap of the GFP/mCherry signals was observed (Fig 5I insets), although relative intensities also varied. These results suggested that the three proteins were processed correctly.

Figure 5
figure 5

Expression of two different proteins as well as a selectable marker from a single vector in Drosophila cultured cells.

(A–I) Brightfield images of Drosophila S2R+ cells that were mock-transfected (Ctrl; A–C), transfected with pAc5-GFP (D–F) or transfected with pAc5-STABLE2-Neo (G–I). Three days after transfection, G418 was added to the media at concentrations of 0 (A, D, G), 600 (B, E, H) or 2000 μg/ml (C, F, I). Fluorescent images of the same cells from panels A–I showing GFP expression (A'–I') or FLAG-mCherry expression (A”–I”). White boxes mark zoomed region shown in inset (panel I, overlay). All images were taken after 30 days of treatment with G418.

Full size image

To verify that T2A processing was efficient, we checked the FLAG-mCherry and GFP products by Western blot analysis (Fig. 6). The GFP proteins encoded by both pAc5-STABLE1-Neo and pAc5-STABLE2-Neo were similar (27 kDa). The FLAG-mCherry protein was detected by anti-FLAG antibodies and also exhibited the predicted size (30 kDa). No upper bands corresponding to unprocessed fusion peptides were observed. These results demonstrate that dT2A and T2A are processed correctly in Drosophila cells and that pAc5-STABLE2-Neo can be used to generate stable cell lines that express two independent proteins of choice by selection with G418.

Figure 6
figure 6

T2A and dT2A are correctly processed in Drosophila cultured cells.

Western blot analysis of S2R+ cells transfected either with pAc5-STABLE1-Neo or pAc5-STABLE2-Neo vectors, or only mock-transfected. Observed bands using antibodies against GFP and FLAG (to detect FLAG-Cherry) were consistent with correct processing of the T2A/dT2A sequences. Expected sizes of non-processed FLAG-Cherry-dT2A-GFP-T2A-neo (91kDa; white arrow) or GFP-T2A-neo (58kDa; black arrow) are depicted; none is detected in either case, even in long exposures. Molecular weights in kDa are indicated to the left.

Full size image

Generation of stable transformants expressing a GFP-fusion protein

As an additional example of the utility of these vectors, we generated a stable cell line that expressed a GFP-fusion with the novel protein CG6522, the Drosophila ortholog of the focal adhesion protein Testin. CG6522 (or dTes) contains a PET domain and three zinc-finger containing LIM domains, belonging to the superfamily of PET-LIM proteins. TES associates with cytoskeletal proteins such as Mena, zyxin, talin and actin35,36 and has been shown to have tumor suppressor properties in mice41. By confocal microscopy, cells stably transfected with pAc5-STABLE1-Neo alone expressed GFP, which localised strongly in the nucleus in S2R+ cells, although staining is also visible in the cytoplasm. The cytoplasmic GFP expression did not colocalise extensively with F-actin (Fig. 7A, B). In contrast, stable cells generated with pAc5-dTES.GFP-Neo showed that dTes-GFP is highly enriched in the cytoplasm and showed extensive colocalisation with the F-actin cytoskeleton (Fig. 7C, D). Like human TES, this suggests that Drosophila dTes protein may also associate with actin. To confirm that the fusion protein is correctly processed, we analysed the cell extracts by Western blot (Fig. 7E). Anti-GFP antibodies revealed a band that corresponds to the predicted size of the dTES.GFP fusion protein (90 kDa from dTES plus 27 kDa from GFP), indicating that the T2A peptide is correctly processed.

Figure 7
figure 7

dTes-GFP colocalises with actin cytoskeleton in S2R+ cells.

(A–D) Confocal pictures showing the expression of GFP encoded by pAc5-STABLE1-Neo (GFP-T2A-neo; A, B) or the fusion protein encoded by pAc5–dTes.GFP-Neo (dTES-GFP; C, D). Cells are stained with Alexa568-Phalloidin (red) to show the F-actin cytoskeleton and with DAPI (blue) to show the nuclei. Grayscale images represent single green (A'–D') or single red channels (A”–D”). B and D show the region indicated in A or C, respectively. Arrowheads in B and D indicate dTes-GFP protein accumulation, while arrows show F-actin. Scale bars in A and C = 8 μm. (E) Western blot analysis of S2R+ cells transfected with pAc5-dTES.GFP-Neo and probed with anti-GFP. A single band of expected size is observed. Molecular weights in kDa are indicated to the left.

Full size image

Discussion

Here we have used viral-derived T2A sequences to create new vectors for establishing stable Drosophila cell lines. The vectors pAc5-STABLE1-Neo and pAc5-STABLE2-Neo produce dicistronic or tricistronic transcripts, respectively, that are efficiently processed in S2R+ and Kc167 cells. These vectors avoid the need to use cotransfection with a separate vector expressing an antibiotic-resistance gene or the need for a dual-promoter. Both vectors are versatile. Using pAc5-STABLE1-Neo one can choose between making an N-terminal or C-terminal GFP fusion protein or to substitute GFP by any protein of interest. pAc5-STABLE2-Neo allows the expression of distinct fluorescent protein fusions, or simultaneous expression of a fluorescent marker (GFP or mCherry) and a protein of interest (by substituting one of the fluorescent protein-encoding modules). This allows visual identification of cells expressing the protein of interest, without worry about the fluorescent protein interfering with protein function. Although all drug-selected cells should express the protein of interest, the co-expressed fluorescent protein may also allow discrimination between high- and low-expressors when coupled with fluorescence-activated cell sorting (FACS). Alternatively, ORFs with different tags could be substituted. Tags such as 6×HIS, GST, or others that facilitate protein purification will make these vectors useful for recombinant protein production for biochemical and structural biology purposes, especially where multiprotein complexes are being studied. Although T2A-processing appears efficient in both S2R+ and Kc167, it is highly recommended to confirm processing of polyproteins by Western blot. Actin 5C-based expression and G418 selection may also be compatible with other insect cell types42,43, increasing the utility and potential applications of these vectors. Since we have shown that T2A-mediated multicistronic expression is feasible in Drosophila cultured cells, a similar strategy may be applied in vivo to expand the already powerful molecular genetic toolbox available through fly transgenesis.

G418-based selection is effective and takes 3–4 weeks to establish a stable population. Puromycin selection (mediated by the pac gene) has been used successfully for Drosophila cell selection when using co-transfection or dual promoter approaches13,15. Our earlier attempts to use puromycin selection in combination with the T2A sequence, or with a distinct 2A peptide derived from porcine teschovirus-1 (P2A), were problematic, linked to poor apparent cleavage efficiency (data not shown). Nevertheless, by using extended T2A sequences44, distinct linker sequences between modules, or placing the PuroR cassette in the first (5′), rather than last (3′), position, the use of puromycin selection may be possible. Since our vector design allows other selection cassettes to be substituted for NeoR, using other selection strategies, such as those using hygromycin, or the faster-acting blasticidin45,46, may give further flexibility and reduce the time to achieve stable populations.

In conclusion, the vectors pAc5-STABLE1-Neo and pAc5-STABLE2-Neo described here will allow the creation of stable Drosophila cultured cells in an efficient way and serve as a flexible platform for future insect-cell based applications.

Methods

DNA constructs

pAc5.1-V5-His-A (Invitrogen) and pAc5.1B-EGFP (called pAc5-GFP in text; obtained from Addgene; gift of E. Izaurralde47) were used in the construction of vectors described here. In pAc5-STABLE1-Neo (Fig. 1), GFP is followed by the T2A peptide sequence derived from Thosea asigna (EGRGSLLTCGDVEENPGP) and the neomycin resistance gene (NeoR; confers resistance to G418 in eukaryotic cells; derived from pIRES-Neo) (Invitrogen). mCherry was synthesised (Geneart) and later PCR-amplified to add an N-terminal FLAG epitope and a C-terminal T2A sequence, which has degenerate sequence at the nucleotide level to minimise vector recombination (called dT2A in text). This cassette was cloned into pAc5-STABLE1-Neo as an EcoRI-XbaI fragment, generating pAc5–STABLE2-Neo. Organisation of unique restriction sites, ORFs and T2A sequences in pAc5-STABLEs is summarised in Figure 1. Unique sites allow replacement of FLAG-Cherry, GFP, and/or NeoR, allowing flexibility for new applications. In most cases, PCR amplification and subcloning was used to generate plasmids. Primer sequences and cloning details are available upon request. All vectors were sequence-verified and DNA preparation was done using standard procedures.

Cell culture, transfection, antibiotic treatment and immunofluorescence

Drosophila S2R+ cells37 and Drosophila Kc16748,49 were obtained from the Drosophila Genome Resource Center (DGRC) and were cultured in Drosophila Schneider's medium (Invitrogen) supplemented with 10% of fetal bovine serum (FBS), 1% of penicillin/streptomycin at 25°C in a humidified incubator.

Cell transfection was performed using the non-liposomal reagent Effectene (Qiagen) following the manufacturer's instructions using a DNA:Enhancer ratio of 1∶8 and DNA:Effectene ratio of 1∶3. 25μl of the transfection solution was added to each well in a 96-well plate and 100 μl of medium containing 1×105 cells were dispensed per well. All transfections were performed in quadruplicate. After 72 hours, selective medium was added with different concentrations (0, 600, 2000 μg/ml) of neomycin/G418 (Sigma). Every 5–7 days cells were split if necessary and selective media was replaced. After 3 weeks, cells were transferred to 24-well plates. At 4 weeks, cells were visualised using a CKX41 fluorescence microscope using a 20× objective (NA0.40; Olympus) and images captured using a mounted camera (Olympus E330) and identical exposure times (2 sec.) Percentages of GFP-expressing cells among those detected by bright-field were calculated by analysis of 4 fields, using Photoshop and ImageJ software. Longer exposures confirmed that some weak expressors were not detected when using 2 sec. exposure (data not shown).

For immunofluorescence, after 48–72 hours, transfected cells were transferred to ConA-treated coverslips for attachment50 and fixed with 4% paraformaldehyde, 0.1% Triton X-100 in PBS and treated as previously described51 using DAPI (Sigma) or Alexa568-phalloidin (Invitrogen). Confocal imaging was performed using a TCS-SP2 DM-IRE2 microscope with the 63× objective (HCX-PL-APO, NA1.4; Leica).

Western blot analysis

Cells were harvested in 2× Laemmli buffer and lysates were analysed by SDS-PAGE on 10% gels. Antibodies against GFP (Roche) or FLAG epitope (M2; Sigma) were used, followed by HRP-conjugated anti-mouse secondary antibodies (Jackson Immunoresearch). Signals were detected using Supersignal West Pico (Pierce) and the Chemidoc XRS system (BioRad).