Site-specific recombination

The great promise of clonal analysis for research studies was nonetheless hindered by the inefficient and nonspecific method of using ionizing radiation to induce genetic recombination. While irradiation of mutant strains coupled with identifiable markers generated clones that could be recognized and studied, it was also mutagenic itself, often causing tissue damage throughout the organism.

A solution to the problems associated with irradiation came from the development of P-element-mediated germ line transformation in Drosophila (Rubin and Spradling 1982), which permitted researchers to insert genes from other organisms into flies. This advance, along with the identification of the bacteriophage Cre/loxP (Hamilton and Abremski 1984) and the yeast FLP/FRT (Andrews et al. 1985; Sadowski 1995) site-specific recombinase systems, spurred experiments that demonstrated that these recombinases function in flies. FLP/FRT-mediated recombination was shown to be highly efficient in flies (Golic and Lindquist 1989; Golic 1991), but the Cre/loxP system proved to be less efficient and somewhat toxic in flies (Siegal and Hartl 1996; Heidmann and Lehner 2001). The FLP/FRT recombination system thus quickly became the method of choice in Drosophila, making site-specific recombination between homologous chromosomes much more user-friendly and significantly modernizing the generation of mosaics.

The beauty of the FLP/FRT system is that recombination in flies is now inducible, site-specific, and highly efficient. FLP is a sequence-specific recombinase that mediates recombination by binding two conserved, short FRT sequences, and catalyzing cleavage and chromosomal exchange (Senecoff and Cox 1986; Ma et al. 2007). Due to its ease and utility, the FLP/FRT system has undergone many innovations since it was first introduced into flies and remains the premier tool for generating mosaics in Drosophila research.

Depending upon the position and orientation of FRT sites in DNA, FLP/FRT-directed activity leads to recombination or to DNA excision and rejoining (Golic and Lindquist 1989; Golic 1991). When positioned at the same location on each sister chromatid, the FRT site will catalyze trans-chromosomal exchange (Box 1 and Figure 1A). Recombination occurs between the two chromosomes specifically during the G2 phase of the cell cycle when sister chromatids are paired. Conversely, when two FRT sites are positioned in the same orientation on the same chromosome, FLP catalyzes cis-recombination and excision of the sequences between the two FRTs (Figure 1B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

trans-chromosomal and cis-chromosomal recombination. (A) trans-chromosomal recombination is mediated by the presence of FRT sites on distinct chromosomes. Expression of FLP promotes recombination between the FRT sites specifically during the G2 phase of the cell cycle, generating two individual cells with distinct genotypes after the mitotic event. Mitotic descendants of the two original cells are generated, producing sibling clones (“twin spots”) when X segregation occurs (see text for details). In epithelia, sibling clones generally stay in close proximity. In the scheme, distinct chromosomes carry two independent markers (red and green boxes). After recombination occurs, the two original cells will give rise to two differently-labeled clones (red and green twin-spot clones). Because it relies on sister chromatid exchange, trans-chromosomal recombination can only occur in mitotically active cells and is also known as Mitotic Recombination. (B) cis-chromosomal recombination occurs when FRT sites are present on the same chromosome arm and oriented in the same direction. FLP recognizes FRTs and causes recombination with the consequent excision of the DNA sequence (generally called an FRT cassette) contained within the two FRTs. This cassette contains a cell marker (red box). Thus, a cell with a different genotype is generated in a background of cells with the original genotype. The excision–recombination event is heritable, thus all of the progeny of the recombined cell will inherit the “FLP-out” genotype that loses the red marker (white clone in a background of red cells). Notably, this type of recombination can occur in any stage of the cell cycle.

Control of FLP expression

Since the inauguration of FLP/FRT as a tool in Drosophila, a variety of FLP transgenes have been created to precisely control the expression of FLP both temporally and spatially (Box 3). The heat shock promoter hsp70^P is commonly used to control pulses of FLP expression in response to modulations in temperature. Under this promoter, FLP is not expressed at 18° and remains low at 25°, but is highly expressed above 30° (Golic and Lindquist 1989; Solomon et al. 1991). Although the heat shock promoter induces FLP throughout the entire organism, recombination frequency and clone induction can be controlled easily by manipulating the duration of heat shock, the temperature (e.g., 30 or 37°), or the stage of development. A heat shock early in development, when there are few cells, will yield a few clones that will be able to grow fairly large; a heat shock at a late stage, when tissues can have thousands of cells, will result in numerous, small clones (Struhl and Basler 1993). Recombination efficiency can also be varied by the method of heat delivery: for instance, a water bath tends to be more efficient at heat transmission than an air incubator, and therefore a shorter heat shock is required to achieve the same result (Golic 1993). Moreover, the expression of FLP can be restricted spatially by coupling it to tissue-specific promoters. For example, eyeless-FLP, which allows clones to be generated specifically in cells where the eyeless promoter is normally active, has been particularly useful because mutant eyes can be generated while the rest of the animal remains wild-type and thus viable (Newsome et al. 2000). Expressing FLP under GAL4/UAS (upstream activating sequence) control (see Figure 2) is another way to increase the flexibility of FLP-mediated recombination.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

The Gal4/UAS system. Flies carry an enhancer-Gal4, UAS-GFP, and UAS-GOI (gene of interest) transgenes. When the enhancer is active, Gal4 is expressed and binds the Upstream Activating Sequence (UAS), leading to the expression of both GFP and the gene of interest (GOI).

Below, we discuss FLP/FRT-mediated cis- vs. trans-recombination (also respectively called intra- or interchromosomal recombination), in separate sections for clarity.

Twin-spot clones

The generation of twin-spot clones remains an important example of a method that involves trans-chromosomal recombination. A widely used refinement of the twin-spot method is to use Drosophila strains in which a fluorescent gene under control of a ubiquitously expressed promoter (e.g., ubi^P-GFP) is combined with an FRT chromosome and used in trans to the FRT labeled with (or without) another marker. Such lines are widely available in the community, making it possible to study clonal behavior in all stages of development (Box 3). As an example, crossing flies that carry a mutation of gene X on an FRT 40A chromosome to flies expressing hsp70^P-FLP and carrying ubi^P-GFP on an FRT40A chromosome that is wild-type for X, will yield progeny that are trans-heterozygous for the X mutation and for ubi^P-GFP. In mitotically active cells, when a heat shock is applied, FLP is expressed, resulting in recombination between the two FRTs (Figure 1A). After chromosome segregation, two genotypically distinct daughter cells are generated. One cell will be homozygous mutant and lack GFP expression, distinguishing it from its sister cell that possesses two copies of ubi^P-GFP (in Figure 1A, illustrated as red vs. green cells). Cell proliferation then generates “twin” marked clones of cells of each genotype that are easily identified by GFP expression. As illustrated in Figure 1A, the recombinant, double ubi^P-GFP-positive cells (green) can also be distinguished from nonrecombinant cells that retain one copy of ubi^P-GFP by the different intensity of the fluorescent signal (yellow).

The power of the twin-spot method lies in the fact that an internal control clone is always generated, in the same event, next to the mutant clone. Twin spots allow a direct and reliable comparison between mutant and wild-type clones, and the size of the wild-type twin can serve as a reference for comparison with the homozygous mutant clones. Studies of growth regulation have long relied on the twin-spot method, as clone size provides insight into how a mutation affects cell and tissue growth (Neufeld et al. 1998; Johnston et al. 1999). When mutant clones are smaller than their wild-type twin clones the gene likely functions as a positive growth regulator. Conversely, the gene is probably a negative growth regulator when the mutant clones are larger than their wild-type twins. The use of mosaic analysis as the basis of genetic screens was critical for the identification of new growth regulators that ultimately elucidated the Hippo growth regulatory pathway (Moberg et al. 2001; Tapon et al. 2001, 2002; Wu et al. 2003; Huang et al. 2005).

Twin-spot generator

Clone-marking methods such as ubi^P-GFP were improvements over adult cuticle markers, and allowed analysis of positively marked clones in situ during growth of imaginal discs and other tissues. However, a disadvantage was that one of the twin clones contained no marker—was negatively marked—after recombination. Such negatively-marked clones can be difficult to distinguish, especially in highly folded tissues. An ingenious FLP/FRT-mediated strategy that independently labels each sibling clone was developed by Griffin et al. who named it “Twin Spot Generator” (TSG) (Griffin et al. 2009). The TSG method utilizes chromosomes carrying hybrid sequences encoding partial, complementary fluorescent proteins [e.g., the C-terminus of enhanced GFP (EGFP) and the N-terminus of monomeric red fluorescent protein (RFP)1], flanking an FRT site placed where an intron is normally located. Mitotic recombination is required for the hybrid coding sequences to be reconstructed, transcribed, and spliced to produce fully functional proteins, thus generating sibling clones marked with either GFP or RFP. Since chromosomal crossover is required to link the hybrid coding sequences, all nonrecombinant cells remain unlabeled (Griffin et al. 2009). An advantage of this method is that due to the fluorescent proteins, it permits immediate detection of clones, thus allowing clones to be followed from the first cell division after clone induction.

Mitotic clones in the female germ line

The generation of mitotic clones in the female germ line allows the genetic analysis of the maternal role of genes in oogenesis and embryogenesis, but recessive lethal and dominant female-sterile (DFS) mutations cause arrest of early embryogenesis in females, rendering them unable to lay eggs (Wieschaus 1980; Garcia-Bellido and Robbins 1983; Perrimon and Gans 1983). Chou and Perrimon took advantage of the X-linked DFS ovo^D1 mutant to devise a way to study the role of maternal gene products (Chou and Perrimon 1992). Since males are unaffected by ovo^D1 mutations, they are used as carriers of the mutation, thus by coupling the ovo^D1 mutation to an FRT-bearing X chromosome germ line clones could be generated for studying the maternal effect of homozygous lethal mutations. In the “FLP-DFS” technique, males carrying the ovo^D1 FRT¹⁰¹ chromosome are crossed with females carrying the FRT¹⁰¹ linked to a recessive lethal zygotic mutation (m). Once hsp70^P-FLP expression is induced by a brief heat shock, recombination between the FRT sites generates homozygous ovo^D1 daughter cells, which die, and also homozygous m mutant daughter cells that survive and produce germ line clones. Nonrecombinant heterozygous ovo^D1 mutant cells also die because of the mutation. This powerful tool has been used to study the effect of essential maternal genes for the development of the oocyte and embryo, and allowed genetic screens that would identify X-linked genes required for these early stages of development (Chou and Perrimon 1992). Later, the FLP-DFS technique was extended to the autosomes through insertion of a P-element containing the ovo^D1 mutation on each autosomal chromosome arm, greatly expanding its versatility (Chou and Perrimon 1996).

The Minute technique

Minute mutants constitute a class of ∼50 mutations that affect ribosomal protein genes (Marygold et al. 2007). Most Minutes (M) lead to developmental delay and thin bristle defects in heterozygous mutants (M^+/−), whereas M^−/− is cell lethal. Clones of M^+/+ (i.e., wild-type) cells generated in heterozygous M^+/− animals have a significant growth advantage and can take over large territories within the tissue.

This property was exploited by Antonio Garcia-Bellido and his students in their studies of Engrailed function, and led to the seminal discovery of developmental compartments, the fundamental units of development (Garcia-Bellido et al. 1973, 1976). Using markers such as y and f to mark M^+/− cells in mosaic animals, Gines Morata and Pedro Ripoll, then PhD students in the Garcia-Bellido laboratory, observed that the M^+/− cells were eliminated from the tissue due to active competition with M^+/+ cells (Morata and Ripoll 1975). They named this phenomenon “cell competition” and experiments from several groups ensued to investigate the properties of the competitive elimination of cells (Morata and Ripoll 1975; Simpson 1979) [discussed in detail in Alpar and Johnston (2017)].

M^+/− mutants have also been used in what is known as the “Minute technique.” In cases where a mutation results in very small or no clones, an elegant trick is to put it in a M^+/− background. This gives the mutant cells a relative growth advantage, and can allow weak but otherwise viable cells to survive and form a clone. Generation of Notch (N) mutant clones serves as an illustrative example. Cells homozygous for N mutations do not survive. However when a N mutation is combined with M^+/+ in a M^+/− background, the N+ M^+/− cells are at a disadvantage, allowing the N mutant cells to survive and form clones for analysis (Portin 1980).

The mosaic analysis with a repressible cell marker (MARCM) system

The need for a positive cell label to trace cell lineages and at the same time provide a means for manipulating gene expression in those cells led to the invention of MARCM (Lee and Luo 1999, 2001). MARCM employs the Gal4/UAS system (Figure 2) and also the Gal4 repressor Gal80 (Lue et al. 1987; Suster et al. 2004), to specifically mark one of the sibling clones of a recombination event while leaving the remaining cells unmarked (Figure 3). Gal4, a yeast transcriptional activator whose activity is inhibited by Gal80, functions in Drosophila as a transcriptional activator of any gene positioned downstream of UAS elements (Fischer et al. 1988) and is extensively used for conditional gene expression (Brand and Perrimon 1993) (Figure 2). The advantage of MARCM over a typical twin-spot analysis is that it allows expression of a Gal4-regulated (UAS) transgene only within the marked clone, while the transgene is repressed in all other cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The MARCM system. (A) The mother cell carries hsp70FLP and Gal4 coding sequences under the control of two different promoters, a UAS-GFP to mark cells and FRT sites on the two homologous chromosomes that carry either the mutation of interest (depicted by a blue X) or Gal80. Gal4 cannot drive GFP expression because of the inhibitory activity of Gal80. Two daughter cells are generated after heat shock induction and FLP-mediated recombination. (A’) One cell contains two copies of Gal80 and consequently has no GFP expression. (A”) The sibling cell does not inherit Gal80 and consequently Gal4 is transcriptional active, driving the expression of the marker UAS-GFP. Thus, homozygous mutant cells are labeled, whereas wild-type and heterozygous mutant cells remain unlabeled. MARCM, mosaic analysis with a repressible cell marker; UAS, upstream activating sequence.

A typical genetic set up for a MARCM experiment consists of flies carrying an inducible FLP, a Gal4 transgene driven by a promoter, such as actin, a UAS-GFP (or any other fluorescent marker), and FRT chromosomes heterozygous for a mutation in a gene of interest. The presence of a constitutively expressed Gal80 distal to one of the FRTs will determine which sibling clone will express UAS-GFP (Figure 3A). The FLP, Gal4, and UAS-GFP are generally located on other chromosomes, to remain unlinked to the FRTs. Upon induction of FLP expression, the two FRT chromosomes recombine. As mitotic recombination requires exchange between sister chromatids during the G2 phase of the cell cycle, MARCM is limited to proliferating cells. Chromosomal segregation then generates two different daughter cells: one cell expresses two copies of Gal80, thereby inhibiting Gal4 activity and keeping the cell unlabeled (Figure 3A’). The other daughter cell does not express Gal80 and consequently Gal4 is active and drives expression of the marker. Placement of Gal80 on the FRT chromosome that carries the mutation of interest prevents UAS-GFP expression in homozygous mutant clones, but allows GFP expression in the homozygous wild-type clones. Likewise, placement of Gal80 on the wild-type FRT chromosome yields GFP-expressing homozygous mutant clones (Figure 3A’’).

The MARCM system lends itself well to powerful variations. It is easily adapted for clonal expression of a UAS-regulated transgene, either as a way to overexpress a particular gene or to knock it down with a UAS-RNAi (RNA interference). In addition, by adding a constitutively expressed or inducible marker, such as another fluorescent protein or CD2, to the Gal80-carrying FRT chromosome, both sibling clones resulting from a recombination can be positively marked (de la Cova et al. 2004). Positive marking of both sibling clones allows direct comparison of differences in their size and shape, for example, which has been illuminating in studies of growth and cell competition (de la Cova et al. 2004, 2014); a double repressor method, called twin-spot MARCM, was also developed to allow developmental tracking of single neurons and distinct neuronal populations (Yu et al. 2009). Here, loss of either one or the other of the two repressors by a daughter cell allows for differential marking. These methods have proven to be effective tools for a variety of different research questions in larval and adult tissues.

Notes on the use of Gal80 in mosaics

The combined use of Gal80 and the Gal4/UAS system is an extremely useful addition to the Drosophila genetic toolkit, especially when temporal control of gene expression is an important aspect of mosaic generation. In the context of the MARCM technique, temporal control of Gal80 (and thus Gal4-regulated gene expression) can be achieved by regulating FLP induction. However, in this case, functional repression by Gal80 can be only released—but not induced—in a temporally controlled manner, and only in the daughter cells in which copies of the Gal80 gene have not been inherited. Gal80 is also useful in FLP-out experiments, since control of its induction is possible. In one variation of the FLP-out method, Gal80 can be constitutively turned on or off in FLP-expressing cells using tub^P > stop > Gal80 (FLP-in) or tub^P > Gal80 > (FLP-out) (Zecca and Struhl 2002; Gordon and Scott 2009; Bohm et al. 2010). Temporal regulation can also be achieved by linking Gal80 to promoters that are active at a specific stage of development. Still, inherent in these uses of Gal80 is its perdurance, which prevents rapid release of Gal4 inhibition. Gal80 can perdure as long as 40 hr after recombination in MARCM experiments in imaginal discs (de la Cova et al. 2004), so experiments need to be designed with this in mind. A widely used method for rapid temporal modulation of Gal4 activity is through a temperature-sensitive Gal80 allele (Gal80^ts) that is active and represses Gal4 at 18° but is inactive at 29° (McGuire et al. 2004). This feature allows fairly tight and reversible temporal control over overexpression or knockdown experiments by a simple shift of the incubation temperature from 18 to 29° at any stage of development.

The FLP-out technique

In the 1990s, a significant innovation arose from the use of the FLP/FRT system in the fly to recombine DNA sequences in cis. When FRT sites are oriented in the same direction and located at two distinct locations within the same chromosome or segment of DNA (Figure 1B), FLP catalyzes the excision of the DNA in between the FRT sites. This technique, known as “FLP-out,” is widely used to generate clones of cells—without a requirement for a specific cell cycle phase—that express a gene or RNAi at a specific time or in a specific tissue. The FLP-out method was developed by Gary Struhl and Konrad Basler as a way to test the idea that Wingless (Wg) functions as a pattern-organizing morphogen in the wing disc (Struhl and Basler 1993). They took advantage of the knowledge that when two FRTs are placed as direct repeats on the same DNA segment, FLP will catalyze the excision of the region between them and joining of the resulting DNA ends (Golic and Lindquist 1989) (Figure 1B). The original constructs were designed so that the actin5C promoter (act^P) was separated from the wg coding sequence by a FRT-flanked “STOP cassette,” consisting of the Draf gene (used as an inert spacer) followed by transcriptional termination sequences to prevent expression of wg (Struhl and Basler 1993). A short, heat shock-induced pulse of hsp70^P-driven FLP expression led to excision of the STOP cassette and allowed expression of the wg gene. Since the FLP-out event is heritable, after cell division each of the cell’s progeny also expressed wg. This technique allowed Struhl and Basler to demonstrate that when expressed in a clone, wg “organized” the nearby tissue, nonautonomously patterning cell fates (Struhl and Basler 1993). Later, the method was adapted for similar studies of Hedgehog (Hh), and the FRT sites were shortened to make them slightly less efficient, giving additional control to the FLP-out process (Senecoff et al. 1988; Basler and Struhl 1994).

The FLP-out method was subsequently combined with the Gal4/UAS system to make it broadly useful (Pignoni and Zipursky 1997). The FLP-out Gal4 system (Pignoni and Zipursky 1997) allows the induced clonal expression of Gal4 and, accordingly, expression of any UAS-regulated transgene (Figure 4). The typical FLP-out cassette consists of a ubiquitously expressed promoter (e.g., act^P), a cell marker [a commonly used one is the rat CD2 coding sequence (Dunin-Borkowski and Brown 1995)] followed by a transcription termination (STOP) sequence; this is then flanked by FRT sequences (annotated act^P > CD2^STOP > Gal4, where > denotes FRT)] (Box 2). In the absence of FLP activity, CD2 is expressed under the control of act^P, but the STOP sequence prevents expression of Gal4 (Figure 4). In the presence of FLP, stochastic FRT-mediated recombination and excision of the CD2^STOP cassette occurs, allowing Gal4 and UAS-GFP expression (Figure 4). The Gal4/UAS-GFP-expressing cell and its progeny are permanently labeled by GFP expression, providing a positive marker of the clone. These clones also lack expression of CD2, providing a negative clonal marker, whereas all cells that have not undergone recombination still express it (Figure 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

The FLP-out Gal4 technique. Schematics of the genetics of the offspring of a cross: the animal carries a hsp70-FLP, a transgene containing the actin promoter followed by a FRT-flanked cassette containing the coding sequence of the marker protein CD2; downstream of the FRT cassette is the Gal4 coding sequence. UAS-GFP and UAS-GOI are also present in the genome. After heat shock, FLP expression is induced and it causes recombination between FRT sequences and excision of the CD2 cassette. Consequently the Gal4 coding sequence comes under the control of the actin promoter, and Gal4 is expressed. Gal4 binds UAS driving the expression of both GFP and the GOI. This generates clones of cells overexpressing the GOI that are marked by GFP and by the absence of CD2. GOI, gene of interest; UAS, upstream activating sequence.

The FLP-out Gal4 system has become an extremely powerful and avidly used tool for conditional, heritable regulation of gene expression in clones in Drosophila. In addition to UAS-GFP, the presence of other UAS-regulated transgenes in the animal allows gene expression to be manipulated in the cells within the clones. Control of FLP expression by the hsp70^P, which can be expressed in all cells, allows clone number to be easily optimized for different tissues, as described above. However, it is possible to restrict clone generation to certain tissues by putting FLP expression under the control of tissue-specific enhancers. Gene expression can also be restricted from specific tissues by including the Gal4 inhibitor Gal80 under the control of a tissue-specific enhancer. A further useful innovation of Gal4-regulated expression is Split-Gal4 (Luan et al. 2006), in which the Gal4 coding sequence is split into two modules, with each fused to a leucine zipper (Zip+ or Zip−) motif, which will create strong heterodimers with each other [for example, Gal4 DNA-binding domain (DBD)-Zip- and Gal4 transcription activation (TA) domain-Zip+ fusions]. Each Gal4 domain-Zip fusion is expressed under the control of a different promoter/enhancer, thus fully functional Gal4 is reconstituted only when the modules’ expression overlaps and they heterodimerize via their Zip− and Zip+ motifs. The Split-Gal4 system can be a powerful way to restrict gene expression to exquisitely refined regions. However, a disadvantage of this system is that existing Gal4 driver strains cannot be used. Making Split-Gal4 strains is potentially a time consuming endeavor as each must be derived de novo. Moreover, with the addition of motifs to enhance transcriptional activation (e.g., VP16 and p65), the Gal80-binding region was eliminated in many Split-Gal4 TA-Zip fusions, making these unresponsive to Gal80 suppression and limiting their versatility (Luan et al. 2006; Pfeiffer et al. 2010). A way around the latter problem was introduced with the “Killer Zipper” (KZip+) system (Dolan et al. 2017), which consists of a dominant-negative repressor of Split-Gal4 activity that is expressed under the control of a third promoter/enhancer . Expression of KZip+ (Gal4DBD-Zip+) in cells containing Split-Gal4 components effectively prevents reconstitution of functional Split-Gal4 due to dominant homodimerization of KZip+ with the Gal4DBD-Zip- module; the Gal4DBD homodimers can also swamp out UAS binding by functional Split-Gal4 heterodimers. The former caveat was addressed with a Split-LexA system called G-MARET (Gal4-based Mosaic-inducible And Reporter-exchangeable Enhancer Trap), which makes use of existing Gal4 driver lines (Yagi et al. 2010). In G-MARET, any existing Gal4 reporter can be switched to a LexA TA, and as a consequence, genetic mosaics can be generated that are composed of cells expressing either Gal4 or LexA TA in a tissue-specific manner. Further details of the Split-Gal4 and Split-LexA methods can be found in the original publications.

FLP-out modifications and alternative recombinase enzymes

Over the years, the FLP-out method has been modified in several ways that extend its uses. Prominent examples are the integration of FLP-out technology with the Q and LexA/lexO systems. The Q system is a binary expression system from the Neurospora qa (Q) gene cluster (Potter et al. 2010), in which a DNA element (e.g., gene-X) is transcriptionally regulated by the activity of a QUAS sequence. When a promoter drives the expression of QF, a transcription factor that binds QUAS, the transcription of a QUAS-responder gene is promoted. QS, a repressor of QF activity, can also be expressed, via the activity of a different promoter. The Q system can be used with or instead of the Gal4/UAS system. Similarly, the LexA transcriptional system (Yagi et al. 2010) takes advantage of the bacterial transcriptional activator LexA, which binds the LexA operator sequence (lexO) to drive the transcription of a gene of interest. The Gal4/UAS, QF/QUAS, and LexA/lexO systems can be combined to generate, mark, and manipulate different populations of cells within the same animal, greatly expanding the repertoire of uses for the FLP-out system.

Construction of mutant variants of the FRT sequence (e.g., FRT3, in which minor alterations in the spacer element eliminate cross-reactivity with FRT) have provided increased control over FLP-mediated recombination (Schlake and Bode 1994). Furthermore, in addition to FLP, several other site-specific yeast recombinases have been adapted for targeted recombination in Drosophila (Nern et al. 2011). These recombinases include KD from Kluyveromyces drosophilarum, R from Zygosaccharomyces rouxii, B2 from Z. bailii, and B3 from Z. bisporus (Araki et al. 1985; Toh-e and Utatsu 1985; Chen et al. 1986; Utatsu et al. 1987). Each of these recombinases recognizes a unique and specific sequence (KDRT, RSRT, B2RT, and B3RT, respectively) and catalyzes recombination at high frequency. They have low toxicity in flies and, importantly, do not cross-react with FLP or each other (Nern et al. 2011). The recombinases can thus be coexpressed, permitting orthogonal manipulation of multiple recombinases and recognition target sites. For example, FLP/FRT and KD/KDRT pairs can be combined in the animal and used with tissue- or cell-specific Gal4 drivers to sequentially refine expression in particular cell lineages (Awasaki et al. 2014; Nern et al. 2015). The principles of such toolkit intersection underlies some of the most powerful multicolor cell-labeling techniques.

“Memory” experiments and G-TRACE

Drosophilists can take advantage of hundreds of Gal4 lines for targeted gene expression, where Gal4 is expressed under the control of either known or unknown tissue- or region-specific enhancer sequences (Brand and Perrimon 1993; Duffy 2002) (see, for example, the Gal4 section of the Bloomington Drosophila Stock Center, http://flystocks.bio.indiana.edu/). However, a major caveat to their use is that gene expression regulated by these enhancers can vary both temporally and spatially depending on the tissue or stage of development. This can be problematic if gene expression in a specific tissue or at a specific time in development is desired. After the FLP-out system was developed, it was quickly appreciated that it could be used to track enhancer expression dynamics and identify cell lineages during development (Ito et al. 1997; Weigmann and Cohen 1999; Jung et al. 2005; Wu and Johnston 2010). As an example, combining ptc-Gal4, an enhancer trap insertion that is expressed in all cells that express the patched gene, with a UAS-FLP and the act^P > CD2^STOP > Gal4 FLP-out cassette, will permanently label every cell that ever expresses ptc, past or present, effectively preserving a memory map of ptc-Gal4 spatial expression. Memory experiments provide a visual record of where different enhancers are expressed during development.

Despite the utility of such memory experiments in providing a snapshot of ongoing and transient expression patterns from a given Gal4 line, they do not differentiate past Gal4 expression from present, real-time Gal4 expression. To address the problem of temporal information, the University of California, Los Angeles Undergraduate Research Consortium in Functional Genomics developed G-TRACE (Evans et al. 2009), a labeling method that combines real-time analysis with the memory marking system to provide information about temporal and spatial expression of a Gal4 driver (Figure 5). G-TRACE is based on the original memory methods using enhancer-Gal4 lines, but adds a second level of control. First, Gal4 binds UAS-RFP and UAS-FLP and drives their expression simultaneously. FLP then excises the FRT-flanked transcriptional STOP cassette that separates the ubi promoter from the GFP open reading frame (Figure 5). This event allows GFP expression to be maintained in all progeny of the cell that underwent FLP-mediated recombination. Conversely, RFP is only expressed in those cells that express Gal4 at the moment of visualization. G-TRACE analysis therefore reveals both the current and previous expression patterns of individual enhancers, with the fluorescent “color code” defined as follows: red marks cells in which the promoter is active in real-time; yellow identifies both real-time expression and lineage expression, meaning that the cell expresses the gene of interest at the moment of the analysis and in an antecedent time; and green identifies cells that previously expressed the gene of interest, providing lineage-tracing information (Evans et al. 2009).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

G-TRACE. This method establishes a “memory” of expression from the enhancer-Gal4 throughout the life history of the animal, and is useful for determining where an enhancer is expressed through all stages of development. The genetic toolbox of the G-TRACE method is shown. Gal4 is expressed under the control of an enhancer of choice. When the enhancer is active, Gal4 is expressed and drives the expression of both UAS-RFP and UAS-FLP. FLP then mediates recombination between the FRTs and excision of the STOP cassette in a FLP-out reaction. Consequently, UAS-GFP is constitutively expressed under the control of a ubiquitously active promoter (e.g., the ubiquitin promoter) in that cell and all of its progeny. RFP, red fluorescent protein; UAS, upstream activating sequence.

CoinFLP

The need to conduct genetic mosaic screens in which a reliable ratio of mutant to nonmutant cells would be generated drove the development of the CoinFLP method (Bosch et al. 2015). This method makes use of a FLP-out cassette containing alternating canonical FRT and FRT3 sites, which are chosen randomly by the FLP recombinase to yield one of two outcomes (Bosch et al. 2015). Since each FRT type is capable of recombining with its own type but not a different type, recombination is mutually exclusive: excision of sequences will only occur between canonical FRT sites or between FRT3 sites.

In CoinFLP, FLP recombinase is expressed under the control of a tissue-specific enhancer (e.g., ey-FLP or Ubx-FLP). Within the FLP-out module, a STOP cassette is placed between a FRT site and a FRT3 site, followed by another FRT and another FRT3, and finally by the Gal4 coding sequence. When FLP is expressed, recombination that occurs between the FRT sites generates a cell that expresses a Gal4 and any UAS-regulated gene of interest, while recombination that occurs between the FRT3 sites generates a cell that does not express Gal4 and thus no transgenes. Due to the arrangement of FRT and FRT3 sites, when FRT3 sites recombine the intervening FRT site is excised and subsequent FLP/FRT-mediated events are prevented. The combination of the stochastic and mutually exclusive nature of the FLP recognition of FRT sites makes recombination similar to the 50–50 chance of a “coin toss” (Bosch et al. 2015). The tissue-specific expression of FLP allows clones to be restricted to certain tissues and, depending upon the onset of expression, can lead to large clones amenable to screens using gene overexpression or knockdown.

An even more powerful tool, based on the CoinFLP system, is CoinFLP-LexGAD/Gal4 (Bosch et al. 2015). Here, the STOP cassette is followed downstream by one canonical FRT site, the LexGAD coding sequence, one FRT3 site, and the Gal4 coding sequence. After differential recombination, daughter cells will express either LexGAD or Gal4: the former will bind to LexAop sequences driving the expression of downstream sequences, similarly to the Gal4/UAS system. The ladder binds the UAS driving downstream expression. One daughter cell will express a fluorescent marker whereas the other cell will express a different fluorescent marker. Furthermore, the combination of the CoinFLP-LexGAD/Gal4 system with the GFP Reconstitution Across Synaptic Partners system (Feinberg et al. 2008; Gordon and Scott 2009) allows the visualization of the boundary between two clonal populations: in addition to the specific fluorescent marker, both LexAop and UAS drive the expression of a different portion of a “split” GFP, where the two components reconstitute at the interface between the two populations (Bosch et al. 2015).

Colorful mosaics: Brainbow, Raeppli, TIE-DYE, and MultiColor FlpOut (MCFO)

Tracking cell lineages over long periods of time can be extremely challenging. MARCM permits a variety of strategies to label individual groups of cells with different markers; however, only a portion of cells of a given tissue can be labeled with these techniques, and information about the unlabeled cells is not easily deduced. Several new strategies have been devised to surmount this problem.

The first such labeling tool reported, Brainbow, is a stochastic site-specific recombination-based system in which different fluorescent proteins are deployed depending on selection of a series of loxP sites (Livet et al. 2007). This technique and variants of it have been extensively used in mammalian models to differentially label multiple cell lineages simultaneously in a variety of tissues. Drosophila versions of Brainbow were subsequently developed, including Flybow (Hadjieconomou et al. 2011) and dBrainbow (Hampel et al. 2011). Both methods were created using modifications of the original Brainbow strategy, and are compatible with intersectional strategies that use traditional FLP/FRT and Gal4-, Q-, and LexA-based systems for even greater sensitivity and control (Hadjieconomou et al. 2011).

TIE-DYE (Worley et al. 2013) can be considered the first whole-tissue-labeling tool, since it will successfully mark most cells in a target tissue or organ (Figure 6). TIE-DYE takes advantage of FLP-out methodology and is comprised of an inducible FLP and three constructs, each with an independent FRT-flanked STOP cassette that is preceded by a ubiquitously active promoter (ubi^P, tub^P, and act^P are commonly used) and followed by the coding sequence of GFP, lacZ, or Gal4. FLP-induced recombination leads to the excision of the STOP cassette, allowing GFP, lacZ, and Gal4 to be expressed. These constructs can be combined with UAS-RFP and UAS responder transgenes of interest, leading to generation of clones of cells expressing GFP, lacZ, Gal4, or combinations of these cell markers (Figure 6, B and C). Gal4 activates expression of a cell marker (RFP) and also potentially of a gene of interest (GOI). Overexpression or downregulation of a GOI is therefore marked in red. As with traditional FLP/FRT technology, the longer the heat shock, the more excisions are generated and therefore more combinations will be possible. A benefit of this technique is that it permits estimation of growth rates in different regions of tissues in conjunction with genetically-induced perturbations (Figure 6, B and C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

TIE-DYE. (A) In TIE-DYE, heat shock-driven expression of FLP mediates excision of three different STOP cassettes, allowing the expression of Gal4, GFP, lacZ, or a combination of those, depending on the severity and duration of the heat shock. Gal4 expression enables the activation of the UAS transgenes: UAS-RFP and UAS-GOI. Clones overexpressing the GOI are marked with RFP. If the transgene is an RNAi construct then its expression will result in the downregulation of the GOI. (B and C) The images show examples of clones in wing discs (B) and ovaries (C), taken from Worley et al. (2013). GOI, gene of interest; RFP, red fluorescent protein; RNAi, RNA interference; UAS, upstream activating sequence.

Another appealing whole-tissue-labeling tool is Raeppli (Kanca et al. 2014), which generates a combination of fluorescent markers (Figure 7). The combinatorial complexity depends on the number of used copies of the Raeppli construct present in the genome. Similar to TIE-DYE, Raeppli allows simultaneous analysis of perturbed and nonperturbed regions of the tissue. Whereas TIE-DYE requires fixation and thus is not suitable for live imaging, visualization of Raeppli clones does not require fixation and staining. The efficiency of recombination is also higher than that observed with TIE-DYE.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

The Raeppli system. (A) A simplified version of the Raeppli construct is depicted. The UAS is followed by a minimal hsp70 promoter, an attB site, a full hsp70 promoter, and one FRT-flanked STOP cassette. Downstream, the integrase coding sequence precedes a series of four different fluorescent protein genes (E2-Orange, mKate, TFP1, and TagBFP) interspersed by attP sites. cassette. Downstream, the integrase coding sequence precedes a series of four different fluorescent protein genes (E2-Orange, mKate, TFP1, and TagBFP) interspersed by attP sites. (B) Upon heat shock, FLP-mediated recombination excises the STOP cassette, leaving integrase under the control of the (full) hsp70 promoter. Integrase protein actively recombines attB and attP sites. Different fluorescent protein genes become expressed when Gal4 is present and are governed by the specificity of the genomic enhancer used. In the example, E2-orange and mKate are excised and TFP1 is expressed, becoming the cell marker. TagBFP is not expressed because of the presence of a stop signal downstream of the TFP1 coding sequence. The coding sequence of each fluorescent gene is followed by a stop codon. (C) An image of a Raeppli wing disc labeled with four markers, taken from Kanca et al. (2014). UAS, upstream activating sequence.

Figure 7A depicts a simplified scheme of the Raeppli construct and its activation. Activation requires several sequential steps. First, FLP activity within a cell mediates excision of a stop cassette that prevents integrase expression (Figure 7B). The expression of integrase, driven by the activity of the full hsp70^P, then causes recombination between the attB and one of the attP sites. Following this event, the integrase is excised together with the coding sequence of one or more fluorescent proteins (Figure 7B). When Gal4 is expressed, UAS sequences allow expression of one of the fluorescent protein genes, each of which is followed by a stop codon. The progeny of each individual cell inherits and maintains the ability to express this same fluorescent protein, creating several individual clones marked with different colors (Figure 7C). LexA can also be used to induce the expression of one fluorescent protein gene, since it binds the LexO sequence that promotes downstream transcription (Yagi et al. 2010). In addition, because either Gal4 or LexA can be used, Raeppli works well in combination with other genetic manipulations. Finally, Raeppli’s real power is that it is a combinatorial method: the use of multiple copies increases the number of color combinations that can be generated. For example, two copies of the Raeppli constructs produce 10 different color combinations, and it is also possible to use three copies.

Another tool for stochastic multicolor labeling of neurons, MCFO, uses smGFP (“Spaghetti Monster GFP”) as a labeling tool that, like Brainbow, Flybow, and dBrainbow, takes advantage of the FLP/FRT system. SmGFP is a mutant, nonfluorescent form of GFP that retains the solubility, stability, and low toxicity of small fluorescent proteins while eliminating their problematic broad excitation and emission spectra (Viswanathan et al. 2015). smGFP essentially operates as a brightly fluorescent hyper-antigenic-labeling tool, by serving as a scaffold for multiple copies of a single-epitope tag that can be detected with commercially available antibodies (e.g., HA, FLAG, MYC, V5, or OLLAS). The MCFO system is comprised of a FLP-out cassette controlling expression of smGFP (>STOP > smGFP) under control of 10× UAS, combined with a cell-type or tissue specific Gal4 driver. In the presence of FLP, the FLP-out cassette is excised, and Gal4/UAS activates expression of smGFP (which is directed to the membrane by a myristoylation signal). The combination of differentially-tagged smGFPs can be used in combination to obtain multicolor FLP-out to reveal cell shape and position, and allows tracking of neurons during development (Nern et al. 2015), but in principle could be useful in a variety of cellular contexts.

Mutant Analysis by Rescue Gene Excision (MARGE)

The MARGE technique is a newly developed tool for mosaic analysis that avoids these problems (Zhou et al. 2016). MARGE is an intrachromosomal recombination-based method in which an FRT-flanked FLP-out transgenic cassette rescues a homozygous mutation. The FRT-delimited cassette, which consists of the DNA sequence that rescues the homozygous mutation and a ubiquitously expressed fluorescent marker, is excised upon induction of FLP. Since MARGE does not require cell proliferation, this technique allows the study of the effect of mutations on already differentiated cells. In principle, the effect of mutations in any gene in the genome could be studied, even those located on the highly heterochromatic fourth chromosome, simply by inserting a rescuing cassette into the fly genome.

FlpStop

An equally useful method takes advantage of the almost 3000 intron Minos-mediated integration cassette (MiMIC) insertions in >1800 distinct genes produced by the Gene Disruption Project (http://flypush.imgen.bcm.tmc.edu/pscreen/index.php) to engineer conditional mutant alleles. “FlpStop” is a small construct designed to integrate into MiMIC insertions via ϕC31 integration and disrupt gene function in a cell type-specific manner with transcriptional and translational blocks (Fisher et al. 2017). The gene disruptive construct consists of a “FLEx-switch” convertible FRT-flanked cassette (Schnutgen et al. 2003) that contains transcriptional terminator sequences and a splice acceptor and stop codon (SA-STOP). Cell-type specificity is achieved with UAS sequences in front of the cassette, subjecting it to Gal4 regulation.

In contrast to FLP-out cassettes, in which FRT sites are in the same orientation (>STOP>) and embedded sequences are excised, the FlpStop cassette contains FRT sites that are in the opposite orientation (>STOP<), taking advantage of the ability of FLP to catalyze inversion of DNA sequences between inverted FRT sites (Golic and Lindquist 1989). Once the FlpStop cassette is integrated into a MiMIC insertion in the proper orientation, upon expression of FLP it is conditionally inverted (“flexibly switched”) via FLP-FRT-mediated recombination, locking in the disruptive element and creating a loss-of-function allele (Fisher et al. 2017). Cells that also express Gal4 will express UAS-tdTomato, thus positively labeling the cells in which the cassette has been inverted. A real advantage of FlpStop is that both disrupting (D) and nondisrupting (ND) alleles can be made. When FLP is under control of a promoter such as hsp70^P, this inventive method can also be used to generate conditional alleles: when put in trans to a null allele of the gene of interest and heat shocked at appropriate times, D alleles can be reverted to ND to rescue the mutant and yield information about genetic sufficiency, and ND can be reverted to D, inducing a mutant and allowing tests for genetic necessity.

Flip-Flop

Like FlpStop, the Flip-Flop strategy (Nagarkar-Jaiswal et al. 2017) allows mosaic analysis in postmitotic cells via the inversion of a Flip-Flop cassette that is induced by FLP activity. As described for FlpStop, the Flip-Flop cassette is integrated into a MiMIC insertion site and, when the cassette is in the disruptive orientation, generates mutant cells marked with mCherry. An advantageous feature of Flip-Flop is that it also marks wild-type cells with EGFP. The module containing the EGFP coding sequence has the opposite orientation of the module containing the mCherry coding sequence. The inversion of the Flip-Flop cassette is therefore marked with the loss of EGFP and the expression of mCherry. This method thus provides the added advantage of a broadly useful way to tag endogenous proteins in combination with the generation of mutant cells, independently marked, for clonal analysis (Nagarkar-Jaiswal et al. 2017).