Synthesis of an Enzymatically Active FLP Recombinase In Vitro: Search for a DNA-Binding Domain

We have used an in vitro transcription and translation system to synthesize an enzymaticaily active FLP protein. The FLP mRNA synthesized in vitro by SP6 polymerase is translated efficiently in a rabbit reticulocyte lysate to produce enzymaticafly active FLP. Using this system, we assessed the effect of deletions and tetrapeptide insertions on the ability of the respective variant proteins synthesized in vitro to bind to the FLP recognition target site and to carry out excisive recombination. Deletions of as few as six amino acids from either the carboxy- or amino-terminal region of FLP resulted in loss of binding activity. Likewise, inertions at amino acid positions 79, 203, and 286 abolished DNA-binding activity. On the other hand, a protein with an insertion at amino acid 364 retained significant DNA-binding activity but had no detectable recombination activity. Also, an insertion at amino acid 115 had no measurable effect on DNA binding, but recombination was reduced by 95%. In addition, an insertion at amino acid 411 had no effect on DNA binding and recombination. On the basis ofthese results, we conclude that this approach fails to define a discrete DNA-binding domain. The possible reasons for this result are discussed.

The 2,um plasmid of Saccharomyces cerevisiae encodes a protein, the FLP recombinase, that promotes an inversion event across a specific site within two 599-base-pair (bp) inverted repeats of the plasmid (8,45). This reaction results in two isomeric forms of the plasmid and appears to be essential for amplification of the plasmid copy number in the cell (14,35,41,46). The FLP recombination system serves as an attractive model to study the molecular events in the pathway of site-specific recombination (6,15,16). The first step in the reaction involves a specific interaction of FLP protomers with the target sequence called the FRT (FLP recognition target) sequence (4,9). The FRT site consists of three 13-bp symmetry elements, two of which are in inverted orientation and separated by an 8-bp core region (4,38). Results obtained by using a gel mobility shift assay and various footprinting techniques (3,7) suggest that a single symmetry element of 13 bp is the basic unit to which a FLP protomer binds. Results from these same experiments indicate that FLP protomers are assembled in an ordered manner onto the FRT site.
Crystallographic studies of several sequence-specific DNA-binding proteins have already provided considerable insight into the molecular interactions between some proteins and their target sequences (reviewed in references 2, 32, 36, and 48). The DNA-binding domains of proteins for which there is no crystallographic information have been identified by limited proteolysis and purification of functional polypeptides (1,12,19,20,43) or by assays of truncated proteins whose synthesis has been directed from deletion variants of their genes in vitro (18,42) and in vivo (24,34).
To understand the interactions important for the FLP binding reaction, we were interested in identifying elements of the protein that are required for site-specific DNA binding. To this end, we used in vitro transcription and translation of mutant FLP genes to assess the enzymatic activity of the respective FLP proteins. We found that the DNA-* Corresponding author. binding activity of the FLP protein is remarkably sensitive to perturbation of its structure.

MATERIALS AND METHODS
Bacterial straim, DNA modification enzymes, and vectors.
Escherichia coli HB101 [F-hsdS20 rB-mB-recA13 ara-14 proA2 lacY) galK2 rpsL20(Smr) xyl-5 mtl-l supE44] and K802 (hsdR hsdM+ gal met supE) were used to propagate plasmids. Cells were made competent for transformation with calcium chloride as described by Mandel and Higa (29). Plasmid vector pSP64 was obtained from Promega Biotec. A derivative of pSP64 that has the alfalfa mosaic virus coat protein leader sequence adjacent to the SP6 promoter was obtained from Lee Gehrke (23). This leader sequence effects efficient translation of heterologous mRNA in both the rabbit reticulocyte lysate and wheat germ extracts (23). This plasmid was designated pPS672. Plasmids pBA104 and pBA112 are described by Andrews et al. (4). pBA104 contains two FRT sites in direct orientation and was used as an excisive recombination substrate for the FLP protein (16). A 100-bp fragnent containing the FRT site used in gel mobility shift assays was obtained from plasmid pBA112. DNA restriction and modification enzymes were obtained from Boehringer Mannheim Biochemicals (Indianapolis, Ind.), Pharmacia, Inc. (Piscataway, N.J.), and Bethesda Research Laboratories, Inc. (Gaithersburg, Md.) and used according to the specifications of the manufacturers. Preparation of plasmids from 1-ml cultures was done as described previously (21). Cesium chloride density gradient purification of plasmids was performed from 1-liter cultures as described elsewhere (11,40). The following duplex oligonucleotide linkers were used in this study: BamHI-5'-CCGGATCCGG-3' (Bethesda Research Laboratories), BamHI-5'-CCGGATCC TGG-3' (New England BioLabs, Inc., Beverly, Mass.), SalI-5'-GGTCGACC-3' (New England BioLabs), NcoI-5'-CATGCCATGGCATG-3' (New England BioLabs), and XhoI-5'-CTCGAGCTCGAG-3' (kindly provided by Keith Schappert). Plasmid construction. Two plasmids were used to direct the synthesis of FLP mRNA by SP6 Fig. 1A. pPS645 DNA was linearized with SmaI and treated with BAL 31 nuclease according to the instructions of the manufacturer (Bethesda Research Laboratories). The termini were repaired with reverse transcriptase and four deoxynucleoside triphosphates, and the blunt ends were ligated to a BamHI linker (5'-CCGGATCCTGG-3'). The circularized DNA was used to transform competent HB101 by selection for ampicillinresistant colonies. Deletion endpoints were determined precisely by sequencing from the BamHI site into the gene by the technique of Maxam and Gilbert (31).
Amino-terminal deletions of the FLP gene. A schematic illustration of the strategy used to construct amino-terminal deletions in plasmid pTA39 is shown in Fig. 1B. Plasmid DNA was digested with PstI and XbaI and treated with exonuclease III (New England BioLabs), followed by Si nuclease (Sigma Chemical Co., St. Louis, Mo.). Since the PstI site is refractory to digestion with exonuclease III, unidirectional deletions were generated into the FLP gene sequences from the 5'-terminal end while leaving the SP6 promoter intact. The fragments were then digested with NcoI, the ends were repaired with reverse transcriptase (Life Sciences, Inc., St. Petersburg, Fla.) and the plasmids were recircularized by using T4 DNA ligase. Ampicillinresistant colonies were obtained after transformation of HB101. The extent of deletion was determined by restriction digestion, followed by the double-stranded sequencing technique (47) with the SP6 promoter primer (5'-CATACGATT TAGGTGACACTATAG-3') (New England BioLabs). Constructs that had the remaining FLP gene sequences in frame with the ATG supplied by the NcoI site were chosen for further analysis.
Linker insertion construction. All in-phase oligonucleotide linker insertions in the FLP gene contained in plasmid pTA43 were done according to the linker-tailing technique described by Lathe et al. (25). Plasmid pTA43 was linearized with EcoRV, Hindlll, or NdeI, position 5917, 106, or 341, respectively, of the 2,um plasmid coordinates (17). The HindlIl and NdeI ends were repaired with reverse transcriptase and ligated to Sail (5'-GGTCGACC-3') and BamHI (5'-CCGGATCCGG-3') linkers, respectively. The EcoRV end was ligated directly to the BamHI linker. Plasmids containing the appropriate linkers in the EcoRV, HindIII, and NdeI sites are referred to as pTA209, pTA211, and pTA213, respectively. Insertion of an XhoI linker (5'-CTC GAGCTCGAG-3') into one of two RsaI sites (position 384) and into the two DraI sites (positions 5807 and 6179) of the gene was achieved by partial digestion with the respective enzyme. Plasmids designated here as pTA251, pTA243, and pTA250 have the linkers inserted into the DraI sites (positions 5807 and 6179) and RsaI site (position 384), respectively.
In vitro transcription and translation assays. In vitro transcription of templates by using SP6 polymerase and in vitro translation of capped mRNA in a rabbit reticulocyte lysate were carried out with kits and specifications of Promega Biotec. Plasmid DNA templates used for transcription were linearized with BamHI or SmaI, extracted with phenolchloroform, ethanol precipitated, and suspended in water at Demonstration that FLP proteins without the carboxy-terminal six amino acids are not functional in the mobility shift and recombination assays. (A) Autoradiograph of mobility shift assay gel. Translation products were incubated with 0.02 pmol of a 32P-end-labeled 100-bp EcoRI-HindIII fragment containing the FRT sequence for 20 min at 30°C. The reactions were subjected to electrophoresis on a 5% polyacrylamide gel. The gel was dried and exposed to film (see text for details). S, 100-bp EcoRI-HindIII fragment; CI, CII, and CIII, FLP-FRT complexes I, II, and Ill, respectively. Lanes: 1, no extract; 2 and 3, 8 and 16 pLl of mock translation extract minus mRNA; 4 and 5, 8 and 16 pd of pTA23 translation product; 6 and 7, 8 and 16 p.1 of PTA225 translation product; 8 and 9, 8 and 16 ,ul of pPS645 translation product; 10, 0.2 U of a partially purified FLP protein (5). (B) Autoradiograph of recombination assay gel. Translation products were incubated with 0.01 pmol of 32P-end-labeled 8.9-kbp EcoRI-EcoRI fragment of pBA104 (an excisive recombination substrate) for 20 min at 30°C. The reactions were subjected to electrophoresis on an 0.8% agarose gel. The gel was dried and exposed to film (see text for details). S, 8.9-kbp linear EcoRI-EcoRI substrate; P, 7.0-kbp EcoRI-EcoRI recombinant product. Lanes; 1, no extract; 2 and 3, 2.5 and 5.0 ,ul of mock translation extract minus mRNA; 4 and 5, 2.5 and 5.0 p.l of pTA23 translation product; 6 and 7, 2.5 and 5.0 pJ of pTA225 translation product; 8 and 9, 2.5 and 5.0 pul of pPS645 translation product; 10, 0.1 U of a partially purified FLP protein (5). (C) Autoradiograph of SDS-polyacrylamide gel. Translation products (40,000 cpm) were subjected to electrophoresis on a 15% polyacrylamide-SDS gel. The gel was impregnated with En3 Hance, dried, and exposed to film. Lanes: M, molecular size markers (indicated in kilodaltons on the left); 1, 10 ,ul of mock translation extract minus mRNA; 2, pTA23 translation product; 3, pTA225 translation product; 4, pPS645 translation product. a concentration of 1 ,ug , The transcription reaction was extracted with phenol-chloroform, and the RNA was precipitated with ethanol and suspended in 25 pJl of water. A 250-ng (0.5-pmol) sample of this RNA was then translated in the rabbit reticulocyte lysate with 450 pmol of [2,3,4,5-3H]leucine (110 Ci mmol-1; ICN Radiochemicals, Irvine, Calif.) and incubated at 30°C for 1 h. The translation mixtures were used immediately for FLP-FRT complex formation and for recombination assays (see below). In all instances, the yields of trichloroacetic acid-precipitable counts were approximately the same.
Gel mobility shift and recombination assays. The 100-bp ii. Reactions were analyzed by agarose gel electrophoresis (5).

RESULTS
Synthesis of enzymatically active FLP in vitro. To have a convenient means to assay the FLP protein encoded by various mutants, we developed an in vitro system for synthesis of enzymatically active FLP. First, FLP mRNA was synthesized in vitro, using SP6 polymerase to transcribe the SmaI-linearized template pPS645. The transcripts were then translated in vitro in a nuclease-treated rabbit reticulocyte lysate in the presence of [3H]leucine. The results from SDS-polyacrylamide gel electrophoresis indicated that there was synthesis of full-length FLP protein (45 kilodaltons) in the extract (Fig. 2C, lane 4) importance of the C-terminal residues in FRT recognition. We constructed deletions of the 3' end of the FLP gene that SITE AMINO ACID NSERTION encoded proteins lacking less than 60 amino acids from the carboxyl terminus (Fig. 1A).

7-ht-Lys
The tide. In the case of the protein encoded by pTA23, the hese proteins.
addition was Ser-Arg-Ile (total length, 420 amino acids). Two plasmids encoding FLP proteins of 415 and 399 amino acids, respectively, likewise directed the synthesis of enzymati-'LP mRNA was omitted, there was very little cally inactive FLP protein. translation in the extract (Fig. 2C, lane 1).
The experiment with plasmid pTA23 used a template that ity of the FLP protein synthesized in vitro was was linearized with BamHI. The 3' end of the message it by testing for the formation of specific FLP-synthesized in vitro would thus be formed by runoff tranimplexes by using a gel retardation assay. We scription from the linearized template. To corroborate this usly shown that the FLP protein, expressed in result, a plasmid was constructed in which a termination from E. coli, forms three distinct complexes codon was placed near the C terminus of the gene. The ated with a DNA fragment containing an FRT plasmid was then linearized at a site about 100 bp past the P synthesized in vitro formed the same three termination codon. Fig. 2A, lanes 8 and 9)

as did FLP made in vivo
This plasmid (pTA225) was constructed by ligation of an ke complexes formed from the FLP made in vitro XbaI linker (5'-CTAGTCTAGACTAG-3'; New England Biigrated with a slightly slower mobility than did oLabs) into the filled-in BamHI site of pTA23. The XbaI d with the Biorex II fraction of FLP. The reason linker contains stop codons in all three reading frames, and y be that for each reaction, the wells in the the second one would be used to terminate synthesis of the gel were markedly overloaded with protein (1.6 pTA225 protein. This plasmid was linearized with PvuII, erived from the translation lysate. which cuts about 100 bases from the termination codon in :ested the ability of FLP synthesized in vitro to the XbaI linker. For this template, the mRNA is predicted to :isive recombination across two FRT sites that allow synthesis of a 422-amino-acid protein. This protein orientation. The substrate for this reaction was does not contain the terminal six amino acids present in the -nd-labeled fragment (pBA104; 16). Recombina-wild-type protein but has Ser-Arg-Ile-Leu-Val at its carboxy-,d by purified FLP protein was evidenced by the terminal end. of a 7-kbp labeled fragment (Fig. 2B, lane 10).
This gene likewise was found to encode an enzymatically rotein synthesized in vitro generated the same inactive FLP protein ( Fig. 2A and B, lanes 6 and 7). The ig. 2B, lanes 8 and 9). The control extract from autoradiograph of the SDS-polyacrylamide gel (Fig. 2C) NA was omitted did not support the formation of indicates that the radiolabeled translation products encoded Fig. 2A, lanes 2 and 3), nor was it able to carry by pTA23 (lane 2), pTA225 (lane 3), and pPS645 (lane 4) nation (Fig. 2B, lanes 2 and 3).
were of the predicted sizes. tro system has several advantages. There is no Deletion of six amino acids from the amino-terminal region fy the FLP protein synthesized in the extract to of FLP diminishes FRT recognition. We next assessed the : DNA-binding and recombination assays. Also, importance of the amino-terminal region of FLP in FRT site protein makes possible an assessment of the recognition. Deletions of the 5' end of the FLP gene were LP protein added to the reactions. Finally, the constructed as described in Materials and Methods and the e in vitro is not subject to proteolytic degrada-legend to Fig. 1B. One plasmid (designated pTA155) was y affect aberrant proteins synthesized in E. coli.
chosen for further analysis. The mRNA synthesized from f six amino acids from the carboxy-terminal SmaI-linearized pTA155 would direct the synthesis of a LP diminishes FRT recognition. -We first deter-polypeptide of 417 amino acids. It should be noted that the her the carboxy-terminal region of the 423-first two amino-terminal residues (Met-Pro) encoded by the FLP protein was required for recognition of the wild-type FLP gene are preserved in this deletion mutant runcated polypeptides that lacked 60, 137, 173, since these amino acids are encoded by the polylinker o acids from the carboxy terminus were made sequences. transcripts of linearized pPS645 DNA cut with The in vitro translation products of mRNA from pTA155 III, EcoRI, or EcoRV, respectively. None of and pPS645 were assessed as described above for the ability :ptides was able to form complexes with a 100-bp to bind to the FRT site and to carry out excisive recombiiing fragment, and no recombination was ob-nation. Results of the mobility shift assay showed that again, ,ul of mock translation extract minus mRNA; 4 and 5, 2.5 and 5.0 R1 of pTA43 translation product; 6 and 7, 2.5 and 5.0 ,ul of pTA209 translation product (the recombinant product is barely visible here, but with extended exposure of the autoradiograph a distinct band is clearly seen); 8 and 9, 2.5 and 5.0 ,u1 of pTA211 translation product; 10 and 11, 2.5 and 5.0 ,ul of pTA213 translation product; 12 and 13, 2.5 and 5.0 ,u1 of pTA243 translation product; 14 and 15, 2.5 and 5.0 p.1 of pTA250 translation product; 16 and 17, 2.5 and 5.0 ,ul of pTA251 translation product; 18, 0.1 U of FLP. (C) Autoradiograph of SDS-polyacrylamide gel. Translation products (40,000 cpm) were subjected to electrophoresis on a 15% polyacrylamide-SDS gel. Lanes: M, molecular size markers (indicated in kilodaltons on the left); 1, 10 p.l of mock translation extract minus mRNA; 2, pTA43 translation product; 3, pTA209 translation product (see below); 4, pTA211 translation product; 5, pTA213 translation product; 6, pTA243 translation product; 7, pTA250 translation product; 8, pTA251 translation product. (Note: the pTA209 translation product migrated to a lower position than did the other translation products. An explanation of this aberrant migration may be that the SDS-bound polypeptide assumes a conformation different from those of the other polypeptides. Anomalous migration of proteins has previously been reported [18].)   (25). This technique has been used in other studies to define the protein kinase domain of the epidermal growth factor receptor (28) and certain functional domains of the oncogenic protein of Fujinami sarcoma virus (44). We therefore examined the effect of the tetrapeptide insertions on the ability of the modified FLP proteins to bind to the FRT site and to carry out recombination. The plasmids used that carry in-frame l insertions in the FLP gene are listed in Fig. 4. Also shown are the sequences of the tetrapeptide insertions as well as their relative positions in the FLP protein. The mRNA was synthesized from the SmaW-linearized templates of these plasmids and used in the in vitro translation reaction.
The mobility shift and recombination assays were done with the in vitro-synthesized proteins encoded by the plasmids mentioned above. Results of the mobility shift assay (Fig. 5A) show that none of the three protein-nucleic acid complexes was formed with the proteins encoded by pTA251 (insertion at amino acid 79; lanes 16 and 17), pTA243 (insertion at amino acid 203; lanes 12 and 13), and pTA211 (insertion at amino acid 286; lanes 8 and 9). However, the protein encoded by pTA213 (insertion at amino acid 364) yielded the three complexes but at a reduced level (lanes 10 and 11) that we estimate to be 20 to 40%o that of the wild-type FLP level (lanes 4 and 5). The proteins encoded by pTA209 (insertion at amino acid 115; lanes 6 and 7) and pTA250 (insertion at amino acid 411; lanes 14 and 15) produced about the same amount of all three FLP-FRT complexes as did the wild-type FLP protein encoded by plasmid pTA43 (lanes 4 and 5). The results of the recombination assay (Fig. 5B) show that the proteins encoded by pTA251 (lanes 16 and 17), pTA243 (lanes 12 and 13), and pTA211 did not carry out excisive recombination with the pBA104 substrate. However, the protein encoded by pTA250 gave yields of excision product (lanes 6 and 7) comparable to those of wild-type FLP made from pTA43 (lanes 4 and 5). The protein encoded by pTA209 produced an extremely low amount of excision product (lanes 14 and 15) that we estimate to be less than 5% of the level produced by native FLP. (The excision product was clearly noticeable only when the autoradiograph was exposed four times as long as usual.) As seen in the autoradiograph of the SDS-polyacrylamide gel (Fig. SC), these translation products were of the predicted lengths.
(The protein encoded by plasmid pTA209 had an anomalously fast migration, perhaps because of an aberrant conformation during electrophoresis.) DISCUSSION We have constructed a number of deletions and in-phase linker insertions in an attempt to determine whether the DNA-binding domain of the FLP recombinase is localized to a specific region of the polypeptide. An in vitro transcription-translation system that directs the synthesis of enzymatically active FLP protein was used to assess the effects of these mutations. Hope and Struhl (18) have used this basic approach successfully to localize the DNA-binding domain of the GCN4 protein to the carboxy-terminal 60 amino acids of the polypeptide. In a similar way, the DNA-binding domain of the glucocorticoid receptor has been delineated (42).
Unlike the situation for GCN4 and the glucocoritcoid receptor, we find that the DNA-binding activity of FLP, as assayed in a mobility shift assay, is diminished upon removal of as few as six amino acids from the aminoand carboxyterminal ends of the FLP protein. DNA-binding activity is also diminished in FLP proteins with tetrapeptide insertions at amino acids 79, 203, and 286. We estimate that the mobility shift assay is sensitive enough to have detected at least a 10-fold reduction in binding affinity for the FLP recombinase.
The results may suggest that the FLP protein is sensitive to changes in its structure at certain regions such that the small deletions and insertions affect the folding or thermostability of the protein. However, this is not generally true, since three insertions, at amino acids 115, 364, and 411, preserved DNA-binding activity and one (amino acid 411) was fully recombination competent. It is possible that the amino and carboxyl termini play a critical role in the thermodynamic stability of the FLP protein, as has recently been shown for the ends of T4 lysozyme (30).
The other possibility is that the DNA-binding domain of the FLP protein responsible for FRT recognition is not on May 3, 2019 by guest http://mcb.asm.org/ Downloaded from constituted from contiguous amino acids as for the helixturn-helix motif (2,10,36,48) or zinc finger domain (13,33,43). Rather, the domain in FLP may consist of residues from several diverse regions of the polypeptide, as in the case of the EcoRI enzyme (22,32). If this is true, then the insertiontolerant regions of FLP could identify regions between segments that form the DNA-binding domain. A genetic study (27) identified changes at amino acids 51, 258, 298, and 343 that resulted in an elevated affinity of the FLP protein for the FRT site. These results would tend to support the hypothesis that the DNA-binding domain of FLP consists of diverse regions of the polypeptide. Figure 6 summanzes all known mutations from this study and others (27,37,39) on the FLP gene and their effects on the sequence-specific DNA-binding function of the protein. A better understanding of the organization of the DNA-binding domain of the FLP protein will have to await further detailed physical analysis of the protein.