Modular and excisable molecular switch for the induction of gene expression by the yeast FLP recombinase

Site-specific recombinases (SSRs) and their cognate target sequences have become instrumental in biological research (1,2). By varying the number, type, and orientation of SSR recognition sites, one can induce predictable genomic manipulations, like inversions, deletions, and insertions, in cells of virtually any organism. Prominent applications include: (i) conditional gene activation or inactivation to study gene function in cells and in whole organisms; (ii) generation of cell lines that encode cytotoxic products or synthesize constant amounts of proteins after DNA integration into predefined loci; and (iii) selection marker removal.

The FLP/FRT system of Saccharomyces cerevisiae consists of an enzyme (FLP) that mediates high-fidelity recombination between two 48-bp-long recognition sequences (FRT sites) (1,3). Plasmids containing one FRT motif can be inserted into a separate template containing another FRT site. In addition, FLP catalyzes the excision and inversion of DNA flanked by FRT sequences in a direct or inverted repeat configuration, respectively. In the former case, the excision product consists of a circularized molecule. Normally, recombinase-inducible gene expression relies on constructs containing upstream of the open reading frame (ORF), a promoter linked to a direct repeat of FRT sites framing a transcriptional terminator. Recombinase-mediated excision of the latter element leads to the synthesis of the desired product. Here, we generated and tested constructs harboring a FLP-responsive molecular switch with a different arrangement (i.e., with the promoter located downstream of the ORF) (Figure 1A). In this configuration, the transgene lacks a promoter and thus should not be transcribed. Importantly, by flanking both genetic elements with FRT sites in the same orientation, one assures, in a single-step, not only transgene activation but also removal of the prokaryotic plasmid backbone. Considering the well-described phenomenon of gene silencing due to the linkage of transcription units to nonmammalian DNA (see References 4 and 5), the elimination of prokaryotic sequences is especially desirable when the goal is prolonged transgene expression.

The pUC19-based plasmids pGS. DsRed and pGS.pA+.DsRed (Figure 1A) were made according to established methods (6). Using ExGen500 (Fermentas, Vilnius, Lithuania), 8 × 10⁴ HeLa cells in wells of 24-well plates (Greiner Bio-One, Alphen aan de Rijn, The Netherlands) were transfected with 0.2 µg pGS.DsRed or pGS.pA+.DsRed mixed with 0.2 µg pAd.FLPe (7) or the control vector pUC19. pAd.FLPe encodes a catalytically enhanced version of wild-type FLP (8). Two days later, monitoring of DsRed (9) expression in living cells by direct fluorescence microscopy (DFM) revealed widespread DsRed synthesis exclusively in cultures that received pAd.FLPe (Figure 1B). Interestingly, in pAd.FLPe-negative cells, pGS.DsRed, but not pGS.pA+.DsRed, led to a small fraction of DsRed positive cells (Figure 1B). To rule out the possibility that the different levels of DsRed expression were due to significant differences in transfection efficiencies, in another set of experiments we included an enhanced green fluorescent protein (EGFP)-coding construct as an internal control. The same pattern of DsRed expression was observed as before. Importantly, EGFP signal visualization revealed an essentially constant percentage of transfected cells regardless of the DNA mixture applied (not shown). These results demonstrate FLPe-dependent activation of transgene expression. Moreover, gene product synthesis in the absence of recombinase (i.e., leaky expression) was clearly less with pGS.pA+.DsRed than with pGS.DsRed. This most likely reflects the blocking effect of the poly(A) signal of the murine methallothionein gene on transcription arising from cryptic promoters located in the plasmid backbone.

Next, to better detect and accurately quantify transgene expression, we performed flow cytomety on HeLa cells that were, once again, transfected with pGS.DsRed or pGS.pA+.DsRed together with pAd.FLPe or with pUC19. Untransfected cells and cells cotransfected with pAd.FLPe and pUC19 served as controls. Results depicted in Figure 2, A and B are consistent with those obtained by DFM (i.e., pGS.DsRed, but not pGS.pA+.DsRed, yielded detectable amounts of reporter protein in recombinase negative cells). Indeed, pGS.DsRed originated DsRed expression in approximately 1% of these cells, whereas transgene expression driven by pGS.pA+.DsRed was at background levels. Importantly, in the presence of pAd.FLPe, both constructs induced DsRed synthesis in a similarly large fraction of target cells leading to induction factors (i.e., on/off ratios) of 25 and 3325 for pGS.DsRed and pGS.pA+.DsRed, respectively. We conclude that transgene expression is much more stringently controlled in pGS.pA+.DsRed than in pGS.DsRed. Thus, the former construct provides an ideal platform for applications requiring fierce repression and, at will, robust induction of transgene expression.

Subsequently, we investigated whether the observed transgene activation is a result of precise FRT targeted recombination. To this end, we performed Southern blot analysis on MunI-linearized extrachromosomal DNA isolated from cre-expressing PER.tTA.Cre76 cells (10) transfected with pGS.pA+.DsRed plus either pUC19 or pAd.FLPe. The autoradiograph in Figure 2C shows, in addition to the input 5.1 kb substrate, the emergence of a 1.9 kb species exclusively in DNA extracted from FLPe positive cells. The latter DNA corresponds to the excision product predicted from recombination at the FRT sites of pGS.pA+.DsRed. No other DNA fragments were detected indicating precise and FLPe-dependent site-specific recombination. This was confirmed by PCR (Figure 2D) and nucleotide sequence (not shown) analyses of FLPe-generated junctions. Finally, since Cre positive cells were used, our results indicate that the DNA elements in pGS.pA+.DsRed are not prone to rearrangements by this other recombinase. This opens the perspective of combining the molecular switch in pGS.pA+.DsRed with the Cre/loxP system (11) and possibly with other SSRs (1,12,13). In fact, we envision that the circular molecules, freed of prokaryotic DNA and containing exclusively control regions of mammalian origin, if endowed with target sequences for other enzymes such us the bacteriophage ΦC31 recombinase or the adeno-associated virus Rep78/68 proteins, can be utilized to insert DNA into pseudo-att or AAVS1 sites, respectively (1). Another possibility to increase transgene expression durability is by introducing elements that can confer episomal maintenance (e.g., scaffold/nuclear matrix attachment regions or the Epstein-Barr virus origin of replication) (1).

In conclusion, vectors containing a FLP-responsive molecular switch were designed in such a way as to yield, in a single-step, functional expression modules that are devoid of prokaryotic sequences. Importantly, transgene activation from the optimized vector is strictly recombinase-dependent. Finally, FLP/FRT components do not crosstalk with other recombination systems permitting combinatorial approaches.