Chimeras of the Flp and Cre Recombinases: Tests of the Mode of Cleavage by Flp and Cre A

doi:10.1006/jmbi.2000.3967 available online at http://www.idealibrary.com on J. Mol. Biol. (2000) 302, 27±48

Chimeras of the Flp and Cre Recombinases: Tests of the Mode of Cleavage by Flp and Cre A. C. Shaikh and Paul D. Sadowski*

Department of Molecular and The Flp and Cre recombinases are members of the integrase family of

Medical Genetics, University of tyrosine recombinases. Each protein consists of a 13 kDa NH2-terminal Toronto, Toronto, M5S 1A8 domain and a larger COOH-terminal domain that contains the active site Canada of the enzyme. The COOH-terminal domain also contains the major determinants for the binding speci®city of the recombinase to its cognate DNA binding site. All family members cleave the DNA by the attachment of a conserved nucleophilic tyrosine residue to the 30-phosphate group at the sites of cleavage. In order to gain further insights into the determinants of the binding speci®city and modes of cleavage of Flp and Cre, we have made chimeric

proteins in which we have fused the NH2-terminal domain of Flp to the COOH-terminal domain of Cre (``Fre'') and the NH2-terminal domain of Cre to the COOH-terminal domain of Flp (``Clp''). These chimeras have novel binding speci®cities in that they bind strongly to hybrid sites containing elements from both the Flp and Cre DNA targets but poorly to the native target sites. In this study we have taken advantage of the unique binding speci®cities of Fre and Clp to examine the mode of cleavage by Cre, Flp, Fre and Clp. We ®nd that the COOH-terminal domain of the recombinases deter- mines their mode of cleavage. Thus Flp and Clp cleave in trans whereas Cre and Fre cleave in cis. These results agree with the studies of Flp and with the cocrystal structure of Cre bound to its DNA target site. They disagree with our previous ®ndings that Cre could carry out trans cleavage. We discuss the variations in the experimental approaches in order to reconcile the different results. # 2000 Academic Press Keywords: Flp; Cre; site-speci®c recombination; DNA binding; chimeric *Corresponding author proteins

Introduction et al., 1987; Lee et al., 1994; Qian et al., 1990), recombination takes place in a synaptic structure consist- The integrase family of conservative site-speci®c ing of two target sites each bound by two recombinases is comprised of over 100 members molecules of recombinase (Amin et al., 1991; Guo (Abremski & Hoess, 1992; Argos et al., 1986; et al., 1997, 1999; Hamilton & Abremski, 1984; Esposito & Scocca, 1997; Nunes-Duby et al., 1998). Hoess et al., 1990b). The crystal structure of the Cre These proteins all use a conserved catalytic tryo- synapse reveals an extensive series of cyclic NH2- sine residue to break and covalently attach to their terminal and COOH-terminal interactions that target sequences at speci®c phosphodiester bonds establish both ``cross-core'' interactions between (Craig, 1988; Evans et al., 1990; Gronostajski & the two protomers bound to the same lox site and Sadowski, 1985; Landy, 1989; Sadowski, 1993, ``synaptic''interactions between the Cre molecules 1995). The target sequences contain two inverted bound to the two lox sites in the synapse (Guo binding sites each of which binds a molecule of the et al., 1997). recombinase. Although cleavage can take place in Two of the most extensively characterized mem- a dimeric structure consisting of two recombinase bers of the integrase family are the Flp protein molecules bound to one target sequence (Andrews encoded by the 2 m plasmid of yeast (Sadowski, 1995) and the Cre protein of the bacteriophage P1 E-mail address of the corresponding author: (Hoess & Abremski, 1990). The ®rst step in the [email protected] recombination reaction is the site-speci®c binding

0022-2836/00/010027±22 $35.00/0 # 2000 Academic Press 28 Chimeras of the Flp and Cre Recombinases of the recombinase to its cognate recognition site, bond to be cleaved activates that bond to receive the FRT site for the Flp protein and the lox site for the catalytic tyrosine from another Flp protomer the Cre protein (Figure 1(a) and (b)). Although the bound to the target FRT sequence. While initial sequences of the sites differ, they share an identical experiments on the l integrase were compatible organization consisting of two inverted 13 bp with a trans mode of cleavage (Han et al., 1993), sequences (``symmetry elements'') surrounding an subsequent experiments using Holliday junctions 8 bp core region (Hoess et al., 1984, 1986; Lee & gave unequivocal evidence of cis cleavage (Nunes- Saito, 1998; Vetter et al., 1983). Duby et al., 1994). The XerC/XerD recombinase Early gel mobility shift and footprinting exper- also gave clear evidence for cis cleavage iments established the 13 bp symmetry elements as (Arciszewska & Sherratt, 1995). Our initial comple- the recognition elements for Flp and Cre (Andrews mentation experiments also indicated that the Cre et al., 1987; Mack et al., 1992). Hoess et al. (1990a) recombinase could cleave in trans (Shaikh & showed that the NH2-terminal domain of Cre pro- Sadowski, 1997). However, the cocrystal structure tected the core region of the lox site whereas the of Cre that was covalently attached to its DNA tar- COOH-terminal domain bound to the symmetry get showed that the cleavage had occurred in cis element. Missing contact probing of a single sym- (Guo et al., 1997). metry element showed that the core-proximal 4 bp The studies showing trans cleavage by Flp and of the symmetry were important for the binding of Cre made use of complementation experiments in intact Cre but not Cre25, the COOH-terminal which a given molecule of recombinase was posi- domain (Hoess et al., 1990a). The core-distal 9 bp tioned upon a full or half-target site (Chen et al., were needed for binding of both intact Cre and 1992; Shaikh & Sadowski, 1997). Trans cleavage Cre25 (Shaikh, 1997). Crosslinking studies using was demonstrated when one molecule of recombi- the NH2 and COOH-terminal domains of Flp nase donated its catalytic tyrosine to another showed a similar bipartite structure of the FRT recombinase protomer that was unable to cleave symmetry element. The core-proximal 4 bp were due to a mutation and was bound to a different important for the binding of the NH2-terminal site. domain, P13 and the COOH-terminal domain, P32 Such complementation experiments are subject bound to the core-distal 9 bp (Panigrahi & to the possible artifact that trans cleavage occurs Sadowski, 1994). because the design of the experiment precludes cis In order to gain further insight into the mechan- cleavage. The experiments with l integrase and ism of binding and cleavage by the Flp and Cre Holliday junctions gave unequivocal results recombinases, we have constructed chimeras because they made use of two integrases (l and between the two recombinases. The Fre protein HK022) of different binding speci®cities and the contains the NH2-terminal 13 kDa of Flp fused to authors were thereby able to position the proteins the COOH-terminal 25 kDa of Cre. The Clp protein at known sites in the Holliday substrate (Nunes- contains the NH2-terminal 13 kDa of Cre fused to Duby et al., 1994). Likewise the XerC and XerD the 32 kDa COOH-terminal domain of Flp. Each of proteins have different binding speci®cities and the chimeras binds site-speci®cally to hybrid sites therefore one can be certain of their position composed of sequence elements from the target during the cleavage experiment (Arciszewska & site of Flp and Cre, the FRT and lox sites. These Sherratt, 1995). novel binding speci®cities enabled us to examine In the present study we have made use of the the mode of cleavage by Flp, Cre, Fre and Clp. chimeric recombinases Clp and Fre to re-examine The members of the integrase family of recombi- the mode of cleavage by Cre and Flp. We exploited nases all employ a conserved nucleophilic tyrosine the altered binding speci®cities of the Clp and Fre that breaks the scissile phosphodiester bond and proteins to position them on hybrid binding covalently attaches the protein to the 30-phosphoryl elements called lrt and fox. We isolated heterodi- end at the site of the break (Argos et al., 1986; meric complexes containing one molecule of each Esposito & Scocca, 1997; Nunes-Duby et al., 1998). recombinase and assayed for covalent attachment However the family members display a diversity of the recombinases to the target site. We ®nd that of mechanisms of cleavage (Jayaram, 1997; the Cre and Fre proteins cleave in cis whereas the Jayaram & Lee, 1995). Cis cleavage means that the Flp and Clp proteins cleave in trans. We conclude molecule of the recombinase that donates the cata- that the COOH termini of the Cre and Flp proteins lytic tyrosine is bound immediately adjacent to the determine their mode of cleavage. scissile bond. Trans cleavage implies that the recombinase molecule that donates the tyrosine is not bound next to the scissile bond but resides else- Results where in the synaptic structure. Production of chimeric proteins Fre and Clp The Flp recombinase was the ®rst to be studied in this respect and all tests have showed that it Proteolysis of the Cre and Flp proteins had cleaves in trans (Chen et al., 1992; Lee et al., 1994, suggested a bidomainal structure for each protein 1999; Pan et al., 1993). This implies the active site (Hoess et al., 1990a; Pan & Sadowski, 1993; Pan of Flp is formed by contributions from two Flp et al., 1991). Furthermore, previous studies (Hoess protomers: the Flp molecule bound next to the et al., 1990a; Panigrahi & Sadowski, 1994; Shaikh, Chimeras of the Flp and Cre Recombinases 29

1997) as well as the recent cocrystal structure of distal 9 bp of the lox symmetry element is called Cre (Guo et al., 1997), suggested that each domain fox. The hybrid symmetry element consisting of the had a unique role in DNA binding. Therefore we core-proximal 4 bp of the lox symmetry element decided to attempt to construct novel recombinases and the core-distal 9 bp of the FRT symmetry by fusing the NH2-terminal domains of Flp and element is called lrt. Cre to the COOH-terminal domain of Cre and Flp, respectively. We made a plasmid construct that fused the DNA encoding the NH -terminal domain 2 Binding of proteins to the ArkZ substrate of Flp (P13) to the COOH-terminal domain of Cre (Cre25) to give the Fre chimera. Another plasmid The substrate ArkZ (Figure 1(c)) contains a encoded Clp, a chimera in which the NH2-terminal single Cre binding element and a Flp binding domain of Cre (Cre13) was fused to the COOH- element ¯anking a lox core. We incubated ArkZ terminal domain of Flp (P32, see Figure 2(a) and with Fre, Clp, Cre, Flp, Cre25 and P32, and ana- (b)). lyzed the complexes using gel mobility shift As can be seen in Figure 2(c) and (d), both chi- assays. As expected, Cre and Flp each formed a meras were produced in large amounts when single complex (cI Cre and cI Flp) caused by bind- expressed as His-tagged proteins transcribed from ing of a single protein molecule to its cognate sym- the phage T7 promoter. Each protein was con- metry element (Figure 4, lanes 2 and 3, 6 and 7). veniently puri®ed using nickel-af®nity chromatog- Interestingly, incubation of ArkZ with both pro- raphy (see Materials and Methods). teins produced the expected cI complexes of Cre and Flp, but very little of the mixed complex II Substrates used to characterize binding of (cII, lanes 12 and 13). The amount of complex for- native and chimeric recombinases mation can be compared with the binding of the same amounts of Cre and Flp to the full-lox and Before attempting complementations between full-FRT sites (Figure 5(a) and (b)), sites shown in Cre or Flp and Fre or Clp, we used gel mobility Figure 1(a) and (b), where both proteins readily shift assays to characterize the binding activities of generate dimeric complexes (cII) and where Cre the chimeras on a variety of substrates. These also generates higher order (HO) complexes that experiments provided an estimate of the relative are most likely the result of protein-dependent binding af®nity of each hybrid protein for the lox synaptic assemblies (Shaikh, 1997; Wierzbicki et al., and FRT sites and allowed us to design composite 1987). binding substrates that were used later in the tests The His-Fre and His-Clp proteins bound weakly of the mode of cleavage. to the ArkZ substrate but, like their parental The target sequence of Cre is called lox recombinases, generated only complex Is (Figure 4, (Figure 1(a)) and consists of two inverted 13 bp lanes 4 and 5, 8 and 9). Note that the presence of symmetry elements (inverted arrows) surrounding the His-tag on both chimeras allowed us to resolve an 8 bp core region (open box). The sequences of the migration of the four different protein-depen- the lox symmetry elements are shown in green, dent complex Is in the polyacrylamide gel. Since with the core-proximal 4 bp underlined. The lox only the COOH-terminal domains of both Cre and core region is shown in magenta. The two cleavage Flp are able to form protein:DNA complexes in sites for Cre are within the core and are 6 bp apart vitro (Hoess et al., 1990a; Pan & Sadowski, 1993), (vertical arrows, Figure 1(a)) The minimal recombi- the Fre and Clp proteins were most likely interact- nation sequence for Flp is called FRT (Figure 1(b)) ing with the core-distal lox and FRT elements in and has a similar organization to lox. The FRT the substrate. To be sure that the ArkZ substrate symmetry elements are shown in red with the was capable of interacting with the COOH-term- core-proximal 4 bp underlined. The 8 bp core inal domains of Fre and Clp, we incubated the sub- region is in blue and the cleavage sites for Flp are strate with isolated Cre25 or His-P32 peptide. Not 8 bp apart (vertical arrows, Figure 1(b)). Previous only did Cre25 and His-P32 generate monomer footprinting and cross-linking studies had complexes with the ArkZ substrate, but each pro- suggested that the NH2-terminal domains of Flp tein demonstrated apparently greater af®nity for and Cre contacted the core-proximal nucleotides the binding sites in this substrate than the compar- and the COOH-terminal domains contacted the able chimera (Figure 4, lanes 20-23). As the native core-distal regions of the symmetry elements Cre and Flp proteins have a ®ve to ten times great- (Hoess et al., 1990a; Panigrahi & Sadowski, 1994; er af®nity for their cognate sites than their COOH- Shaikh, 1997). We therefore reasoned that hybrid terminal domains (Hoess et al., 1990a; Pan & symmetry elements of the FRT and lox sites might Sadowski, 1993), the reduced binding af®nity of provide unique binding targets for the Fre and Clp the Fre and Clp proteins for the lox and FRT chimeras (see Figure 3). elements may represent an inhibitory effect of the

The test substrates ArkZ, ArkP, ArkF, ArkL and heterologous NH2-terminal domain on the COOH- fox2b (Figure 1(c)-(g)), contained portions of both terminal binding domains of Cre and Flp in the lox and FRT as indicated. A hybrid symmetry chimeric proteins (see Discussion). element consisting of the core-proximal 4 bp of the As noted above, incubation with Cre and Flp right-hand FRT symmetry element and the core- revealed the complex Is of each protein as well as 30 Chimeras of the Flp and Cre Recombinases

Figure 1 (legend opposite)

a weak slower migrating complex, cII (Figure 4, tag in the Fre-Flp mix. In contrast, Cre and His-Fre lanes 12 and 13). Binding of a molecule of Cre or generated no mixed complex IIs (Figure 4, lanes 10 Flp seems to exclude binding of a second molecule. and 11) and a barely detectable amount of cI with Alternately, the absence of cII may re¯ect the His-Fre. Similar results were observed with a mix- absence of cooperativity between the two proteins ture of Flp and His-Clp (Figure 4, lanes 18 and 19). (see Discussion). A small amount of mixed com- We had expected the mixture of Cre and His-Clp plex II was observed when His-Fre and Flp were to generate a heterodimeric complex II, but only incubated together with ArkZ (Figure 4, lanes 16 complex I of each protein was observed (Figure 4, and 17). The slightly slower migration of this com- lanes 14 and 15), perhaps due to the very weak plex II is due to the presence of the additional His- binding of Clp to the substrate. Chimeras of the Flp and Cre Recombinases 31

Binding of proteins to the lox substrate ing (Figure 6, lanes 2 and 3). The degree of cooperativity previously observed with Cre binding to We next used a full-lox site (Figure 1(a)) to test the full-lox substrate (Figure 5(a), lanes 2 and 3) the abilities of Fre and Clp to bind to the lox site. was diminished as the Cre-bound complexes As expected, Cre readily forms monomer (cI), favored the monomeric cI over the dimeric cII. Fur- dimer (cII) and higher order (HO) complexes on its thermore, no Cre-dependent higher order com- cognate site (Figure 5(a), lanes 2 and 3). The Flp plexes were observed. In contrast, the His-Fre protein forms the analogous complex I and II on protein exhibited cooperativity (Figure 6, lanes 4 its cognate FRT site, though little higher order and 5) and bound with two to four times higher complex is seen (Figure 5(b)). As with the single af®nity to the ArkP substrate than to the full-lox lox element-containing site ArkZ, the His-Fre pro- substrate (Figure 5(a), lanes 4 and 5). Incubation tein bound weakly to the lox site compared to Cre with both Cre and His-Fre generated a complex II and the Cre25 peptide (Figure 5(a), lanes 4 and 5, band (cII Mix, Figure 6, lanes 6 and 7) that was dis- lane 7). Despite the decreased binding af®nity of tinct from both the Cre and Fre homodimer com- His-Fre for the lox site, the binding of the chimera plexes. The intermediate migration of the mixed was cooperative, as formation of complex II (cII complex between the homodimer complexes of Cre Fre) was favored over complex I (cI Fre). In con- and His-Fre was consistent with the binding of one trast to the binding of Cre25, protein-protein inter- monomer each of Cre and His-Fre to the lox and actions between Fre monomers also generated fox elements of the ArkP substrate. Indeed, the het- some higher order complexes. erodimeric cII of Cre and His-Fre was favored over The Clp protein was unable to bind the lox site, each of the homodimer species (Figure 6, cf. lanes consistent with both the absence of a Cre COOH- 2, 4 and 6). terminal binding domain in the fusion protein and the inability of the NH2-terminal domain to confer af®nity to the heterologous COOH-terminal Binding of proteins to the ArkF substrate domain (Figure 5(a), lane 6). We designed a second hybrid substrate to test Binding of the proteins to the ArkP substrate the binding of the native and chimeric proteins. The ArkF substrate was similar to the ArkP sub- We then examined the ability of the native and strate described above, except that the fox site was chimeric recombinases to bind to the ArkP sub- replaced with another composite symmetry strate (Figure 6). The substrate contains a lox core element, lrt (Figure 1(e)). This lrt element contained ¯anked by the left lox symmetry element and a fox the core-proximal region of a lox symmetry element (Figure 1(d)). The fox element contains the element and the core-distal region of an FRT 4 bp FRT core-proximal sequence and the 9 bp element and therefore contained the recognition core-distal sequence of the lox site. This fox sym- elements for both the NH2 and COOH-terminal metry element should provide the binding regions domains of the Clp protein. We performed gel for both the P13 and Cre25 portions of the Fre chi- mobility shift assays with this substrate using the mera. Cre, Flp, Cre25, His-P32, His-Fre and His-Clp pro- Neither Flp nor Clp was able to bind to the teins and the results are shown in Figure 7. ArkP substrate because of the absence of the core- Cre generated copious amounts of the mono- distal sequence of the FRT symmetry element in meric cI and, interestingly, formed some dimeric the substrate (Figure 6, lanes 8-10). cII as well (Figure 7, lanes 2 and 3) when higher The modi®ed lox symmetry element contained in levels of Cre protein were used. As with the single the ArkP substrate dramatically affected Cre bind- lox symmetry element-containing ArkZ substrate,

Figure 1. The lox and FRT-based substrates used in this study. The sequences of the oligonucleotides are shown. Each was 35 nucleotides in length. Where required, the 50-end was labeled as described in Materials and Methods. Symmetry elements (horizontal arrows) are derived from the lox site (green) or the FRT site (red). Core-proximal regions are underlined. lox and FRT core regions (open boxes) are magenta and light blue. Vertical arrows indicate the putative cleavage sites by Cre or Flp on the native lox and FRT sites. (a) Full-lox DNA substrate. The cognate site for the Cre protein contains two identical, inverted 13 bp lox symmetry elements that ¯ank an 8 bp lox core region. The cleavage sites are 6 bp apart. (b) Full-FRT DNA substrate. The cognate site for the Flp protein contains two inverted 13 bp FRT symmetry elements (a and b) that ¯ank an 8 bp FRT core region. The cleavage sites are 8 bp apart. The two symmetry elements differ by a single base-pair in the core-proximal region. (c) ArkZ DNA substrate. The hybrid substrate consists of a lox core region ¯anked by the left symmetry element of the lox site and the a symmetry element from the FRT site. (d) ArkP DNA substrate. The hybrid substrate consists of a lox core region ¯anked by the left symmetry element of the lox site and a composite binding element, fox, which is derived from the core- proximal 4 bp of the FRT a symmetry element and the core-distal 9 bp of the lox symmetry element. (e) ArkF DNA substrate. The hybrid substrate consists of a lox core region ¯anked by the left symmetry element of the lox site and a composite binding element, lrt, which is derived from the core-proximal 4 bp of the lox symmetry element and the core-distal 9 bp of the FRT symmetry element. (f) ArkL DNA substrate. The hybrid substrate consists of an FRT core region ¯anked by the left symmetry element of the FRT site and the composite binding element, fox. (g) fox2b DNA substrate. The substrate contains two inverted 13 bp fox symmetry elements ¯anking an FRT core region. 32 Chimeras of the Flp and Cre Recombinases

Figure 2. Peptide maps of proteins used in this study and the puri®cation of Fre and Clp. (a) Peptide maps of Flp and Cre. Flp and Cre are divided into NH2-terminal domains called P13 and Cre13 and COOH-terminal regions called P32 and Cre25. The amino acid residues included in each region are shown. The four conserved catalytic residues are shown for Flp and Cre. (b) Peptide maps of Fre and Clp. The chimeras were constructed by swapping the heterologous NH2-terminal domains between Cre and Flp. The Fre protein results from the fusion of the NH2-terminal domain of Flp and the COOH-terminal domain of Cre while the Clp protein results from the fusion of the NH2- terminal domain of Cre to the COOH-terminal domain of Flp. The amino acid residues from each parent protein are shown. The four conserved catalytic residues in the two chimeric proteins are shown with Cre or Flp numbering. (c) Puri®cation of His-tagged Clp and Fre. Puri®cation of His-Clp and His-Fre from an induced culture by Ni-NTA- af®nity chromatography. Marker lanes contain standards of the indicated molecular mass (kDa). Load lanes represent the soluble fraction of the total cell protein that was loaded on the Ni-NTA-agarose column. Pass lanes represent the ¯ow-through. His-Clp was eluted in the 150 to 200 mM imidazole-containing fractions. His-Fre protein was eluted by 175 and 200 mM imidazole. Both proteins were greater than 90 % pure at this point. Clp and Fre-containing fractions were concentrated by spin dialysis as described in Materials and Methods.

Cre25 generated a single complex with the ArkF The incubation of Cre and His-Clp with the substrate and His-Fre bound weakly to form a ArkF substrate generated abundant amounts of a monomeric cI (Figure 7, lanes 10-13). heterodimeric complex (cII Mix) in addition to the No complex was formed when the ArkF sub- individual complex I bands attributable to the strate containing the lrt symmetry element was binding of Cre and His-Clp (Figure 7, lanes 6 and incubated with Flp (Figure 7, lanes 8 and 9). In 7). The mixed complex II consists of one Cre mono- contrast, the His-Clp protein formed copious mer bound to the lox symmetry element and one amounts of monomer complexes with the ArkF His-Clp protein bound to the lrt symmetry element substrate (Figure 7, lanes 4 and 5): by phosphori- in the ArkF substrate. mager quanti®cation, the amount of complex I generated by Clp on the lrt site was about 20 times Binding of proteins to the ArkL substrate that observed in the Clp reaction with the FRT symmetry element in the ArkZ substrate (cf. While both the ArkP and ArkF substrates were Figure 4, lanes 8 and 9). Clp formed no dimeric essentially modi®ed lox substrates, it was also of complex II, showing that the protein does not bind interest to examine the binding of the chimeric pro- to the lox symmetry element of the ArkF substrate. teins to an altered FRT substrate. The ArkL sub- Chimeras of the Flp and Cre Recombinases 33

Figure 3. Scheme of binding of native and chimeric recombinases to symmetry elements. The two domains are shown in black (Cre) or gray (Flp). The sequences of the symmetry elements (horizontal arrows) are shown for the native lox site (left top) or FRT site (left bottom) with the core-proximal 4 bp underlined. Binding of the chimeric recombinases to the hybrid symmetry elements is shown on the right. Vertical arrows indicate cleavage sites. strate contains the FRT b symmetry element, the The combined results of the binding assays illus- FRT core and a fox symmetry element (Figure 1(f)). trate the altered binding af®nity of the Fre and Clp We conducted gel mobility shift assays to measure chimeras compared to the Cre and Flp proteins. the binding of Cre, Flp, Cre25, His-P32 and the Cre and Flp bound with high af®nity to the lox two chimeric proteins to the ArkL substrate. and FRT sites. In contrast, the Fre and Clp chi- As expected, both Flp and His-Fre formed only meras bound weakly to these symmetry elements. complex Is on the ArkL substrate (Figure 8, lanes The composite symmetry elements fox and lrt, 2-5). The Cre25 and His-P32 proteins also gener- were high-af®nity binding sites for the Fre and Clp ated a single complex band by binding to the fox proteins. Native Cre and Flp recombinases bound and the FRT symmetry elements contained in the weakly to these novel composite target symmetry ArkL substrate (Figure 8, lanes 8-10). The His-P32 elements. The Cre25 and P32 proteins bound their peptide exhibited a decreased binding af®nity to respective native and composite symmetry the ArkL substrate compared to the Flp protein, as elements with equal af®nity; hence the sequence of observed previously with the FRT symmetry the core-proximal nucleotides in a target symmetry element contained in the ArkZ substrate. In con- element did not affect the binding of the isolated COOH-terminal peptides of Cre or Flp. Finally, trast, the His-Fre chimera bound the ArkL DNA incubation of the native and chimeric recombinases with a higher af®nity than the Cre25 peptide. with the hybrid substrate containing both the cor- Both Cre and His-Clp exhibited weak inter- responding native and composite target symmetry actions with the substrate (Figure 8, lanes 11-14). elements generated distinct mixed dimer com- Modi®cations of the core-proximal nucleotides of plexes. The combinations of Cre and His-Fre with the lox symmetry element severely impaired the the ArkP substrate, Cre and His-Clp with the ArkF binding of Cre, while the absence of the core-proxi- substrate and the incubation of Flp and His-Fre mal nucleotides of the lox symmetry in the FRT with the ArkL substrate, are all examples of such symmetry element severely impaired the binding mixed reactions combining a native protein with a of the Clp. chimeric protein that generate heterodimeric com- A mixed complex II band (cII Mix) was observed plexes. when the Flp and His-Fre proteins were incubated with the ArkL substrate (Figure 8, lanes 6 and 7). Mode of cleavage: experimental design This heterodimeric complex II re¯ects the binding of one monomer each of Flp and His-Fre to the Previous experiments designed to illustrate trans FRT and fox symmetry elements of ArkL. cleavage by Cre and Flp made use of a comple- 34 Chimeras of the Flp and Cre Recombinases

Figure 4. Binding of Cre and Flp-derived proteins to the ArkZ substrate. Gel mobility shift assay. Binding reactions were done as described in Materials and Methods. Lane 1, no protein added. Lanes 2 through 23, the numbers above each lane refer to the pmol of indicated protein added. S, the 32P-labeled ArkZ DNA substrate, illustrated to the left with the 50 end labeled (asterisk). cI Cre, cI HFre, cI Flp, cI HClp, cI Cre25 and cI HP32, complexes generated by binding of one monomer of Cre, His-Fre, Flp, His-Clp, Cre25 and His-P32 proteins to the ArkZ DNA substrate. cII Mix, the heterodimeric complex generated by concurrent binding of two different proteins to the same DNA substrate. Lanes 13, and 20-23 originated from a different gel from lanes 1-12 and lanes 14-19.

mentation strategy whereby a protein was bound Cleavage of the ArkP substrate by Cre and His- to a target site and trans complementation for clea- Fre vage was measured (Shaikh & Sadowski, 1997). In a previous paper we described experiments where We incubated the labeled ArkP substrates (containing a lox and a fox symmetry element, a half-site that was non-cleavable was occupied Figure 1(d)) with a mixture of Cre and His-Fre, with a Cre molecule that acted as the tyrosine- and isolated the band containing the mixed com- donor (site B, see Figure 9). The site that was des- plex II. Fortunately, the preparative reactions gave tined to be cleaved (X25) was occupied by a Cre no Cre and His-Fre homodimeric complex IIs. As protein that was cleavage-incompetent due to a controls, we also isolated the complex I band mutation in the catalytic tyrosine. formed by Cre as well as the homodimeric cII A more rigorous protocol for examining the bands formed from reactions containing only Cre mode of cleavage is to use whole recombination or Fre. After electroelution, the labeled complexes sequences to which proteins can be targeted by were analyzed by SDS-PAGE and the results are site-speci®c DNA binding. The hybrid recombina- shown in Figure 10((a) top strand labeled; (b) bot- tion sites ArkP, ArkF and ArkL allowed us to tom strand labeled). place native and chimeric proteins at known sym- The Cre complex I bands showed no covalent metry elements within the substrates and thereby attachment (Figure 10(a) and (b), lanes 2). In con- permitted us to address directly whether the var- trast, the homodimeric complex II bands isolated ious proteins cleave in cis or trans (see Figure 1(d)- after treatments with Cre or His-Fre alone showed (f)). We labeled the top or bottom strand of each robust or very weak cleavage, respectively substrate, incubated it with the appropriate pair of (Figure 10(a) and (b), lanes 3 and 4). While the proteins and isolated the mixed heterodimeric more slowly migrating His-Fre covalent complex complexes (cIIs). After analysis of the complex by was barely detectable, the results illustrate that SDS-PAGE, we were able to determine the mode both strands of the ArkP substrate were cleavable of cleavage for the Cre, Fre, Flp and Clp proteins. by both proteins. Coupled with the absence of Chimeras of the Flp and Cre Recombinases 35

Figure 5. Binding of Cre and Flp-derived proteins to the lox and FRT substrates. Gel mobility shift assay. Binding reactions were done as described in Materials and Methods. (a) Binding of Cre and Flp-derived proteins to the lox substrate. Lane 1, no protein added. cI Cre and cII Cre, complexes generated by binding of one and two monomers of Cre protein, to the lox DNA substrate. cI HFre and cII HFre, the corresponding complexes generated by the His- tagged Fre protein. cI Cre25 and cII Cre25, the complexes generated by the Cre25 peptide. HO, higher order complexes. Lane 7 originated from a different gel from that shown in lanes 1-6. (b) Binding of Flp to the FRT substrate. Lane 1, no protein added. cI Flp and cII Flp, complexes generated by binding of one and two monomers of Flp protein to the FRT DNA substrate.

covalent attachment in the Cre monomeric com- ment of Cre solely to the top strand indicates a cis plexes, these results demonstrate that the cleavage mode of cleavage. Similarly, as the His-Fre chimera activity of Cre is dependent upon establishing pro- was bound adjacent to the cleavage site on the tein-protein interactions between monomers (Guo labeled bottom strand, and only a His-Fre covalent et al., 1997). These results also show the importance complex was detected, Fre was also cleaving in cis. of preparative isolation of the heterodimeric com- Since covalent attachment of His-Fre to the top plex II to avoid possible contamination by homodi- stand or Cre to the bottom strand was not meric complex IIs. detected, the results exclude a trans mode of clea- When the top strand was labeled, the isolated vage for Cre and His-Fre. heterodimeric complex showed only a Cre covalent complex (Figure 10(a), lane 5). Similarly, when the Cleavage of the Ark F substrate by Cre and His- bottom strand was labeled and the mixed dimeric Clp complex was isolated, only His-Fre covalent attachment was observed (Figure 10(b), lane 5). Since the We then used the ArkF substrate to monitor top covalent complex results from attachment of the strand cleavage adjacent to the Cre binding site protein after cleavage of the labeled DNA strand, and bottom strand cleavage adjacent to the Clp the location of the cleaving protein in the complex binding site (see Figure 1(e)). The preparative reac- allows us to determine the mode of cleavage. Since tion combining Cre, His-Clp and the ArkF sub- Cre is bound adjacent to the cleavage site on the strate generated the same pattern of complexes as labeled top strand (Figure 6), the covalent attach- the analytical scale reaction (Figure 7, lane 6). The 36 Chimeras of the Flp and Cre Recombinases

Figure 6. Binding of Cre and Flp- derived proteins to the ArkP substrate. Gel mobility shift. Com- plexes named as in Figures 4 and 5. Note that the heterodimeric cII is resolved from each of the of the homodimeric cIIs.

SDS-PAGE analysis of the isolated mixed com- showed a slightly greater amount of cleavage pro- plexes and the complex I bands formed by Cre and duct than that observed from the isolated dimer Clp is shown in Figure 11. (Figure 11(a), lane 5). As the complex I bands from No covalent complexes were observed in the Cre the unmixed protein reactions showed no covalent and His-Clp complex I bands isolated after treat- attachment, the presence of cleavage in these com- ment with single proteins. This again shows the plex Is may be attributable to the disassembly of dependence of the catalytic activity on the for- the heterodimeric complex after cleavage. mation of multimeric protein-DNA complexes. The results for the isolated mixed dimeric com- Most surprisingly, the isolated mixed dimer plex II in which the bottom strand was labeled are complex in which the top strand was labeled shown in Figure 11(b). A Cre covalent complex showed, in addition to Cre-dependent covalent was detectable (Figure 11(b), lane 6). But by phos- attachment, a large excess of the His-Clp covalent phorimage quanti®cation, the level of Cre-depen- complex (Figure 11(a), lane 6). These results, dent cleavage was of the order of 50 times lower coupled with the binding assignments of Cre and than the amount of Cre covalent complex observed Clp on the ArkF substrate, demonstrate that Cre on the top strand. While the amount of Clp-depen- cleaves the top strand in cis, while the Clp chimera dent covalent complex far exceeded the amount of is cleaving the same site in trans. A number of Cre cleavage product as before, the level of Clp additional bands appear that are also seen in cleavage was also reduced by 100-fold compared assays of covalent attachment by Flp (Figure 12). to the top strand cleavage results. This reduction in These most likely result from proteolytic fragments cleavage product from the Cre-His-Clp dimer was of Clp covalently attached to the cleaved labeled also evident in the complex I bands isolated from strand. By phosphorimager analysis, we deter- the mixed reaction (Figure 11(b), lanes 4 and 5). mined that Clp-dependent covalent attachment Here, the amount of Clp-dependent covalent com- exceeded that of Cre by a factor of 10-15 times plex was decreased by 100 times, while no clea- (data not shown). The Cre complex I band isolated vage product was detectable from the isolated Cre from a mixed reaction showed barely detectable complex I band. However, despite the reduced levels of covalent attachment (Figure 11(a), lane 4). levels of cleavage by both Cre and Clp of the bot- However, the isolated His-Clp complex I band tom strand, these results are consistent with trans Chimeras of the Flp and Cre Recombinases 37

Figure 7. Binding of Cre and Flp-derived proteins to the ArkF substrate. Gel mobility shift assay. Complexes named as in Figures 4 and 5.

cleavage by Cre and a cis mode of cleavage by Clp. The labeled top strand of the ArkL substrate The small amount of trans cleavage by Cre is con- monitored cleavage adjacent to the Flp binding sistent with our previous experiments (Shaikh & site. The heterodimeric complexes showed a barely Sadowski, 1997). detectable protein-dependent complex whose ori- gin could not be determined. We could not be certain whether it was indeed a Flp covalent complex Cleavage of the ArkL substrate by Flp and His- when comparing it to the position of the Flp-FRT Fre covalent complex (Figure 12(a), lanes 4 and 5). However, no His-Fre-dependent covalent complex Incubation of the ArkL substrate with Flp and was observed. Hence, Fre was unable to cleave the Fre allowed us to examine the mode of cleavage top strand in trans and little, if any, cis-cleavage by by the Flp and Fre proteins (Figure 1(f)). ArkL con- Flp occurred. tains one FRT and one fox symmetry element. The The mixed dimeric complex generated both Flp complex I bands isolated after incubation with the and His-Fre covalent complexes with the bottom Flp or His-Fre alone generated no covalent com- strand (Figure 12(b), lane 4). By phosphorimager plexes with either the top or bottom strand of the quanti®cation, we found that Flp and Fre produced ArkL substrate (Figure 12(a) and (b), lanes 2 and equal amounts of covalent complex. The level of 3). We were unable to isolate Flp and His-Fre com- Flp-dependent cleavage was greatly reduced complex I bands from a mixed reaction without con- pared to the activity of Flp on the FRT substrate. tamination by the other complex I bands (data not Interestingly, the Fre protein was more active in shown). the Flp-His-Fre heterodimer than in the Fre homo- 38 Chimeras of the Flp and Cre Recombinases

Figure 8. Binding of Cre and Flp-derived proteins to the ArkL substrate. Gel mobility shift assay. Complexes named as in Figures 4 and 5.

dimer (Figure 12(b), lane 6). Since the labeled bot- tic residues and exhibit the binding speci®city of the tom strand detects cleavage adjacent to the Fre full-length proteins for their cognate DNA target binding site, the results illustrate a trans mode of sites. Panigrahi & Sadowski (1994) identi®ed the cleavage for Flp and cis cleavage for the Fre chi- core-proximal and core-distal regions of the FRT mera. site as the cognate sites for the NH2-terminal P13 In summary, these results show that Cre cleaves and COOH-terminal P32 domains of Flp (see predominantly, if not exclusively, in cis while Fre Figure 3). Similarly, the lox symmetry element can cleaves only in cis. We also found that Flp cleaves be subdivided into a core-proximal site that is in trans and that Clp cleaves predominantly in bound by the NH2-terminal Cre13 and a core- trans. distal site that is bound by the COOH-terminal Cre25 (Shaikh, 1997). The common domainal struc- Discussion ture of Cre and Flp and the parallel organization of the symmetry elements of the lox and FRT sites The rationale for the design of the Fre and Clp allowed us to construct novel proteins, Fre and Clp, chimeras originated from the studies of the proteo- through swaps of the NH2-terminal regions of Cre lysis of the Cre and Flp proteins. Both recombinases and Flp. can be cleaved into analogous NH2 and COOH- The Fre and Clp chimeras qualify as novel pro- terminal domains (Hoess et al., 1990a; Pan & teins distinct from their Cre and Flp progenitors.

Sadowski, 1993). The COOH-terminal domains of The addition of the heterologous NH2-terminal both Cre and Flp contain the four conserved cataly- domain in both chimeras greatly reduced their bind- Chimeras of the Flp and Cre Recombinases 39

Figure 9. Rationale of the previous complementation test used (Shaikh & Sadowski, 1997, Reproduced with permission from Journal of Biological Chemistry (1997) 272, 5695-5702). The cleavable X25 site is loaded with the cleavage-incompetent CreHis Y324C protein and the non-cleavable B site with a cleavage-competent Cre protein (top). After the two reactions are mixed, the Cre protein bound to the B site donates its tyrosine 324 which cleaves the X25 site and covalently attaches the protein to the 32P-labeled top strand (middle). The radioactive label covalent complex is detected by SDS-PAGE (bottom). Squares, CreHis Y324C; triangles, CreHis; asterisk, 32P radioactive label.

ing af®nity for the cognate DNA target sites of Cre COOH-terminal peptides with relatively large NH2- and Flp. In addition, the composite recognition tar- terminal tags. Fre and Clp both exhibited greater get sites developed for Fre and Clp were speci®c for binding af®nity for their cognate sites than the Cre25 each chimera and did not permit binding of the par- and P32 peptides. Furthermore, Clp showed a level ental Cre and Flp proteins. Secondly, the Fre and of cleavage activity similar to intact Flp and much Clp proteins were not simply new versions of the greater than that generated by P32 (Shaikh, 1997). 40 Chimeras of the Flp and Cre Recombinases

Figure 10. Covalent attachment of Cre and Fre to the ArkP substrate. Equal counts of isolated complexes and substrate bands were analyzed by SDS-15 % PAGE, as described in Materials and Methods. The labeled (asterisk) ArkP substrate is shown at the top. The cartoons above the substrate show the locations of the proteins in the mixed complex II. The labels above each lane refer to the isolated substrate of complex band run as the sample. (a) Covalent attachment by Cre in cis. Lane 1 (S), isolated ArkP substrate. Lanes 2 and 3, isolated Cre complex I (cICre) and complex II bands (cIICre). Lane 4, isolated His-Fre complex II (cIIHFre) band. Lane 5, isolated Cre-Fre heterodimeric complex (cIIMix). covCre, covalent complex of Cre and top strand, indicative of cis-cleavage. Note that the absence of a His-Fre covalent complex (covHFre) in lane 5 means that no trans cleavage by Fre was detected. (b) Covalent attachment by Fre in cis. Lane 1 (S), isolated ArkP substrate. Lanes 2 and 3, isolated Cre complex I (cICre) and complex II bands (cIICre). Lane 4, isolated His-Fre complex II (cIIHFre) band. Lane 5, isolated Cre-Fre heterodimeric complex (cIIMix). covHFre, His-Fre bottom strand covalent complex, indicative of cis-cleavage. Note that the absence of a Cre covalent complex (covCre) in lane 5 means that no trans cleavage by Cre was detected.

Similarly, the Fre protein demonstrated the ability to COOH-terminal domain. Similar regulation of generate synaptic complexes and cleavage activity, DNA binding of the COOH-terminal domain of 70 in contrast to Cre25, which is defective for both the s protein by its NH2-terminal region has activities (Hoess et al., 1990a; Shaikh, 1997). been observed (Dombroski et al., 1993). This putative masking of the COOH-terminal binding domain by the NH terminus may also occur in the The mechanism of binding by Fre and Clp 2 native Flp and Cre proteins. Flp does not bind the In gel mobility shift assays, the Fre and Clp chi- lrt symmetry element (Figure 7, lanes 8 and 9), meric proteins exhibited very weak binding to whereas P32 binds well (lanes 14 and 15). Like- single lox and FRT symmetry elements. Fre showed wise, native Cre binds poorly to the fox symmetry much less complex formation than the COOH- element (Figure 8, lanes 11 and 12), whereas Cre25 terminal domain of Cre with a single lox symmetry does bind well (lanes 8 and 9). element-containing site. Similarly, the P32 peptide The nature of the interactions between the NH2 of Flp bound a single FRT symmetry element and COOH-terminal domains in Flp and Cre is much better than Clp. The decreased af®nity of Fre unknown. Such an interaction is not apparent in and Clp for the lox and FRT symmetry elements the cocrystal structure of Cre because the DNA of might suggest that the heterologous NH2-terminal the lox site is interposed between the two domains domain masks the intrinsic binding capacity of the (Guo et al., 1997). However, Subramanya et al. Chimeras of the Flp and Cre Recombinases 41

Figure 11. Covalent attachment of Cre and Clp to the ArkF substrate. Equal counts of isolated complexes and substrate bands were analyzed by SDS-15 % PAGE, as described in Materials and Methods. The labeled (asterisk) ArkF substrate is shown at the top. (a) Covalent attachment by Cre in cis, covalent attachment by Clp in trans. Lane 1 (S), isolated ArkF substrate. Lanes 2 and 3, isolated Cre complex I (cICre) and His-Clp complex I (cIHClp) from Cre and

His-Clp-only reactions. Lane 4 and 5, isolated Cre complex I (cICreMix) and His-Clp complex I (cIHClpMix) from the mixed Cre and His-Clp reaction. Lane 6, isolated Cre-Clp heterodimeric complex (cIIMix) from the mixed Cre and His-Clp reaction. covCre, Cre-top strand covalent complex, indicating cis cleavage by Cre. covHClp, His-Clp top strand covalent complex, indicating trans cleavage by Clp. The covalent attachment of Clp derived from a complex I is presumed to result from disassociation of the Cre subunit from the mixed dimeric complex. (b) Covalent attachment by Clp in cis, covalent attachment by Cre in trans. Lane 1 (S), isolated ArkF substrate. Lanes 2 and 3, isolated Cre complex I (cICre) and His-Clp complex I (cIHClp) from Cre and His-Clp-only reactions. Lanes 4 and 5, isolated

Cre complex I (cICreMix) and His-Clp complex I (cIHClpMix) from the mixed Cre and His-Clp reaction. Lane 6, isolated Cre-Clp heterodimeric complex (cIIMix) from the mixed Cre and His-Clp reaction. covCre, Cre bottom strand covalent complex, indicating trans cleavage by Cre. covHClp, His-Clp-bottom strand covalent complex, indicating cis cleavage by Clp. The covalent attachment of Clp derived from a complex I is presumed to result from disassociation of the Cre subunit from the mixed dimeric complex.

(1997) postulated that a large conformational protein interactions that arise between the NH2- change in the XerD protein might be needed to terminal domains of two opposing Cre monomers enable the DNA to contact the active site. (Guo et al., 1997). These cross-core interactions are Alternatively, the native and chimeric recombi- formed by the interface between the A and E nases may simply bind weakly to their non-cog- helices of the two monomers. Our experiments nate sites because the NH2-terminal domain is show the importance of these NH2-terminal repelled by the non-cognate 4 bp core-proximal domain interactions in establishing a stable dimer sequence. The NH -terminal domain would con- 2 complex. When Cre and Flp were incubated with tribute a positive free energy of interaction with the ArkZ substrate (containing inverted lox and the core-proximal sequence, thereby lowering the FRT symmetry elements) we detected only Cre and af®nity of the full-length protein for the non-cognate site. Flp complex I bands and barely detectable levels of a Cre-Flp heterodimeric complex. Therefore, the failure of Cre and Flp to form signi®cant amounts Cross-core interactions and the formation of of complex II on ArkZ may re¯ect a steric clash complex II between the Cre13 and P13 regions of Cre and Flp The crystal structure of the cleaved synaptic that deters mixed complex formation. This obser- complex of Cre highlights the cooperative protein- vation is further supported by the observation that 42 Chimeras of the Flp and Cre Recombinases

Figure 12. Covalent attachment of Flp and Fre to the ArkL substrate. Equal counts of isolated complexes and substrate bands were analyzed by SDS-17 % PAGE. The labeled (asterisk) ArkL substrate is shown at the top. (a) No covalent attachment by Flp or Fre. Lane 1 (S), isolated ArkL substrate. Lanes 2 and 3, isolated Flp complex I (cIFlp) and His-Fre complex I (cIHFre) from Flp- and His-Fre-only reactions. Lane 4, isolated Flp-Fre heterodimeric complex

(cIIMix) from the mixed Flp and His-Fre reaction. Lane 5, isolated complex II band of Flp (cIIFlpFRT) generated with a top strand labeled full-FRT substrate (see Figure 1(b)). Lane 6, isolated complex II band of His-Fre (cIIHFreFox2) generated with a top strand labeled fox2b substrate. This substrate contains two inverted fox symmetry elements surrounding an FRT site core. (Figure 1(g)). covFlp, covalent complex generated by cleavage of the labeled DNA strand by Flp. covHFre, covalent complex generated by cleavage of the labeled DNA strand by His-Fre. Note that the absence of covFlp and covHFre in lane 4 means no cis cleavage by Flp or trans cleavage by Fre occurred. (b) Covalent attachment by Fre in cis, covalent attachment by Flp in trans. Lane 1 (S), isolated ArkL substrate. Lanes 2 and 3, isolated Flp complex I (cIFlp) and His-Fre complex I (cIHFre) from Flp- and His-Fre-only reactions. Lane 4, isolated Flp- Fre heterodimeric complex (cIIMix) from the mixed Flp and His-Fre reaction. Lane 5, isolated complex II band of Flp

(cIIFlpFRT) generated with a top strand labeled full-FRT substrate (Figure 1(b)). Lane 6, isolated complex II band of His-Fre (cIIHFreFox2) generated with a top strand labeled fox2b substrate. covFlp, covalent complex generated by cleavage of the labeled DNA strand by Flp. covHFre, covalent complex generated by cleavage of the labeled DNA strand by His-Fre. Note that the presence of both covFlp and covHFre in lane 4 indicates cleavage by Flp in trans and cleavage by Fre in cis. incubation of the ArkZ substrate with Cre and P32 easily seen when Flp and Fre were incubated with from Flp or with Flp and Cre25 generated much the ArkL substrate (containing inverted FRT and greater levels of a mixed dimer complex (data not fox symmetry elements). Although all three sub- shown). On the other hand the failure to form strates would position the heterologous COOH- mixed complexes on this substrate may simply terminal domains of Cre and Flp opposite each re¯ect the absence of cooperative interactions other, only the CreClp and the FlpFre paired between the heterologous NH2-terminal and mixtures would position the same NH2-terminal COOH-terminal domains of Flp and Cre. domains together across the core. In the case of In contrast, Cre and Clp readily formed mixed ArkF, a Cre13-Cre13 interaction is created between dimeric complexes when incubated with the ArkF the two protein monomers bound in an inverted substrate (containing inverted lox and lrt symmetry con®guration on the substrate. In the case of the elements). Likewise, heterodimeric complexes were ArkL substrate, the two P13 domains are opposed. Chimeras of the Flp and Cre Recombinases 43

Hence, the formation of the mixed dimer complex actions between the COOH-terminal domains of was dependent upon presenting two identical Flp and Clp. homologous NH2-terminal domains cross-core at Footprinting studies suggest that protein-protein the dimer interface. interactions between the NH2-terminal domains of The Cre cocrystal structure showed that the Cre opposing Cre monomers trigger a conformational dimer was also stabilized by the insertion of the change in the COOH-terminal domain of Cre that COOH-terminal N helix of the cleaving subunit repositions the entire protein on the site (Guo et al., into a COOH-terminal hydrophobic pocket of the 1997; Hoess et al., 1990a; Shaikh, 1997). Hence, the opposed, non-cleaving monomer. Although pairing core-proximal region of the lox site may function as a nucleation site for the NH -terminal domain of of the heterologous NH2-terminal domains of Cre 2 and Fre seems unlikely, the binding of Cre and Fre Cre that allows communication between protein to the ArkP substrate (containing inverted lox and monomers leading to the formation of catalytically fox symmetry elements) resulted in the formation active complexes. Similarly, the P13 domain of Flp of a heterodimeric complex. It is possible that this is believed to induce a conformational change in complex was stabilized through protein-protein the COOH-terminal domain of Flp upon binding interactions involving the capture of the N helix of the FRT site (Panigrahi & Sadowski, 1994). The Cre by the Fre monomer bound across the core. core-proximal region in the FRT site may also func- Similarly, both Cre and Fre generated homodi- tion as an initiation site for complex formation by meric complex IIs on the ArkP substrate likely due the Flp protein. to protein-protein interactions between the COOH- terminal regions of the two proteins in the dimer. Cre formed a small amount of homodimeric The mode of cleavage complex with the ArkF substrate (containing inverted lox and lrt symmetry elements) despite Here, we have used the chimeric Fre and Clp generating only a monomeric complex with ArkZ proteins to test the mode of cleavage by the native (containing inverted lox and FRT symmetry Cre and Flp recombinases. We have also used the elements). It is likely that this homodimeric com- distinct site-speci®c binding activities of all four plex is mediated by cross-core interactions between proteins to target them to speci®c symmetry two opposing Cre13 domains. elements in hybrid recombination sites. Further- As observed from the crystal data and con®rmed more, by directly isolating the mixed dimeric com- by these results, Cre relies on major protein-protein plex, we were certain of the location of the two proteins on the hybrid substrate and that only pro- interactions between the NH and COOH-terminal 2 teins in the mixed complex II could contribute to domains of Cre monomers to stabilize multi-pro- the covalent complexes observed on subsequent tein-DNA complexes. The Flp protein, however, SDS-PAGE gels. This strategy allowed us to deter- does not show the same degree of binding coop- mine the mode of cleavage of both proteins in the erativity that is shown by Cre (Ringrose et al., heterodimeric complex. However, since the mode 1998). Formation of a stable heterodimeric complex of cleavage was determined on dimeric complexes when Flp and Fre were bound to the ArkL sub- containing single target DNA sites, we cannot strate (containing inverted FRT and fox symmetry assess how the formation of a synaptic complex elements) supports the idea that the P13-P13 inter- consisting of two sites might modify the mode of actions stabilize the mixed dimeric complex despite cleavage. the heterology between the COOH-terminal Our results showed that the native Cre protein domains bound cross-core on the same substrate. in this protocol cleaved predominantly in cis The diminished level of cooperative cross-core whereas the Flp protein cleaved almost entirely in interactions between the Flp COOH-terminal trans. The chimeric Fre protein also cleaved in cis domains may relate to the trans mode of cleavage while the Clp protein cleaved mostly in trans. Our by Flp. The Cre cocrystal shows that the N helix of results illustrate that the COOH-terminal domain the cleaving subunit is buried in a hydrophobic determined the mode of cleavage of the protein pocket of the non-cleaving Cre monomer situated and that the addition of the heterologous NH2- across the core (Guo et al., 1997). The catalytic tyro- terminal domain in the chimeric proteins did not sine present in the M helix of the COOH-terminal alter the mode of cleavage de®ned by the catalytic domain of the cleaving molecule of Cre has COOH-terminal region of the protein. attached to the scissile phosphate in cis. The Van In earlier experiments we showed that Cre Duyne group has modeled a trans cleaving mech- cleaved in trans (Shaikh & Sadowski, 1997). What anism in which the N helix of the trans-cleaving accounts for the discrepancy between the two sets molecule (e.g. Flp) would have to be buried in cis of experiments? Previously we made use of a half- to allow for donation of the active tyrosine in trans site complementation strategy whereby a Cre pro- (Gopaul & Van Duyne, 1999; Guo et al., 1999). This tein was pre-bound to a non-cleavable half-site and arrangement might effectively remove the major cleavage in trans was measured on another cleava- protein-protein interaction that is evident between ble half-site (Figure 9). The protocol also made use the COOH-terminal domains of Cre and may of cleavage-incompetent Cre variants. The use of account for the apparent lack of cooperative inter- tyrosine-de®cient and His-tagged versions of Cre 44 Chimeras of the Flp and Cre Recombinases and the non-cleavable half-lox sites may have con- cleaves predominantly in cis it can also cleave in tributed to the trans cleavage by Cre. trans even when the experimental protocol allows One of the half-lox sites, X25, was cleavable only cis cleavage. by the non-His-tagged version of Cre in the half- site reactions. Therefore, the His-tagged version of Action of Cre and Fre on ArkP Cre may have been unable to act in cis in reactions with the X25 site alone. Furthermore, the other When Cre and Fre were both bound to the ArkP half-lox site, B, was not cleavable by either version substrate, both proteins cleaved in cis although Cre of Cre. Hence, this site prevented cleavage in cis by cleaved more ef®ciently than Fre. This difference the protein bound to it and acted solely as a carrier may be related to the ef®ciency with which each of catalytically active Cre proteins to an assembly protein activates the scissile phosphate bond. Alter- with the Cre-X25 complex. natively the substrate may be bent differently by Most complementation experiments used to test the respective protein. In the Cre and Fre heterodi- the cleavage mode of Cre or Flp also relied on the meric complex the homologous catalytically active use of protein variants that replace the catalytic COOH-terminal domains were positioned together. tyrosine with an amino acid that may not form a In the Cre cocrystal structure, the catalytically hydrogen bond in the active site (Chen et al., 1992; active mixed dimeric complex is comprised of Gopaul et al., 1998; Guo et al., 1997, 1999). While cleaving and non-cleaving protein monomers (Guo the Cre Y324F protein assumes the conformations et al., 1997). The cis-cleaving monomer inserts its in crystal structures similar to that of wild-type N-helix into a hydrophobic pocket in the non- Cre (Guo et al., 1999), the effect of the Y324C cleaving monomer. Cre cleaves the top strand of mutation that we used may have allowed for trans the ArkP substrate in cis and covalently attaches its complementation to occur in vitro. The crystal catalytic tyrosine to the scissile phosphate adjacent structure of a Cre synaptic complex shows that the to the Cre binding site. The N helix of the cleaving active site of the cleaving Cre monomer is com- Cre monomer could have been donated in trans to posed of the conserved catalytic amino acid resi- the non-cleaving Fre protein. When Fre cleaves the dues Arg173, Arg292, His289 and Trp315 bottom strand of ArkP in cis, its N helix might be surrounding the scissile bond, with the catalytic donated in trans to the COOH-terminal docking tyrosine covalently attached at the cleavage site site in the non-cleaving Cre protein. (Guo et al., 1997). The non-cleaving monomer of Cre shows a similar arrangement except that both Assembly of Cre and Clp heterodimers on ArkF the histidine and tyrosine residues are shifted away from the scissile bond. If cis attack were pre- The Cre and Clp heterodimeric complexes were vented by the cleavage-incompetent nature of the comprised of the cis-cleaving COOH-terminal scissile bond, then the catalytic tyrosine of the Cre domain derived from Cre and a trans-cleaving monomer bound to one half-site may have been COOH-terminal domain of Flp. These complexes directed towards a trans attack of an opposing clea- veri®ed a cis cleavage mode for Cre (Figure 11(a)). vable half-site. Likewise the Cre monomer bound This may also mean that the N helix of Cre is to the X25 site may have also adopted the confor- donated in trans to the non-cleaving Clp protein mation required for the acceptance of an incoming bound cross-core on the substrate. This would catalytic tyrosine residue. require that the COOH-terminal domain of Clp It is possible that the original ®nding of trans (derived from Flp) must accept the heterologous cleavage by the phage l integrase was attributable COOH-terminal N-helix from Cre. to the same explanation (Han et al., 1993). It should The heterodimeric complex of Cre and Clp on be noted that the crystal structure of the COOH- the ArkF substrate showed that Clp cleaved the terminal domain of the l integrase showed con- top strand in trans (Figure 11(b)). Since Clp is siderable mobility of the catalytic tyrosine that derived from the COOH-terminal domain of Flp, might lead to trans cleavage (Kwon et al., 1997). this result is consistent with the known cleavage Construction of a Fre Y324C variant would allow a mode of Flp. Surprisingly, Cre was able to activate test of the prediction that such a mutation would the scissile phosphate and to accept the incoming change the mode of cleavage by Cre on the ArkP catalytic tyrosine from a Flp-derived COOH termi- substrate to trans. nus. The donation of the catalytic tyrosine of Clp It is important to note that the experiments in to the acceptor Cre monomer might require a cis our previous paper did not rule out cis cleavage by collapse of the N helix in both Cre and its equival- Cre (Shaikh & Sadowski, 1997). When both Cre ent in Clp (Gopaul & Van Duyne, 1999). and His-Cre were bound to their respective half- As both Cre cleavage in cis and Clp cleavage in sites, His-Cre cleaved in trans, but cleavage by Cre trans occur at the top strand of the ArkF substrate, was also observed. The design of the experiment the scissile bond activated by the adjacently bound did not allow us to distinguish cis from trans clea- Cre should be the normal Cre cleavage site. In con- vage by Cre. The present results from the mixed trast, the putative cleavage site of the bottom Cre and Clp heterodimeric complex demonstrated strand of the substrate is in fact a Cre cleavage site, that Cre could cleave in cis or support trans clea- which may be inef®ciently activated by the adja- vage by Clp. Here, we show that although Cre cently bound Clp protein. This may have contribu- Chimeras of the Flp and Cre Recombinases 45 ted to the inef®cient trans cleavage by Cre and cis Shaikh & Sadowski, 1997). All plasmids were prepared cleavage by Clp at this site. Although we consider using the Qiagen plasmid isolation kit. it unlikely, it is possible that cleavages by Cre and Clp at this site occur after redistribution of the pro- Oligonucleotides tein(s) that occurs after isolation of the mixed complex II. Oligonucleotides were synthesized at the Hospital for Sick Children/Pharmacia Biotechnology Service Center at the Banting Institute, University of Toronto or at Dal- Action of Fre and Flp on ArkL ton Chemical Laboratories, Toronto. They were puri®ed using the OPC cartridge method. Where needed the oli- In contrast to the robust cleavage activity gonucleotide was 50-labeled with [g-32P]ATP with T4 demonstrated by both Cre and Clp in the heterodi- polynucleotide kinase. After phenol/chloroform-extrac- meric complex with the ArkF substrate, the Flp tion, the oligonucleotide was separated from unincorpo- and Fre heterodimeric complex on the ArkL sub- rated nucleotides using a BioRad P6 spin column strate showed weak catalytic activity of both pro- (BioRad). It was then annealed to the appropriate comp- tein partners. The amount of covalent attachment lementary oligonucleotide by heating and slow cooling in 0.1 M NaCl, 5 mM MgCl of Flp to the bottom strand of the ArkL substrate 2. was much lower than that generated by Flp on the full-FRT substrate. This may have been due to a Proteins decreased ability of Fre to activate the scissile bond for cleavage or to a defect in accepting the catalytic Native Cre protein was puri®ed from an induced cul- tyrosine in the active pocket from either a cis or ture containing pShe11 and Cre25 from an induced culture containing pShe5. His-tagged P32 and native Flp trans-cleavage mode. However, it should be noted were puri®ed from induced cultures containing pShe2 that Cre had no dif®culty in accepting the catalytic and pLD3, respectively. Puri®cation conditions for all tyrosine from Clp in trans. Therefore, we favor the proteins have been described previously (Shaikh, 1997; former explanation. Shaikh & Sadowski, 1997). In none of these experiments have we determined the actual location of the scissile bond. Construction of His-Fre expression vector Although it is reasonable to assume that the protein bound next to the scissile bond would deter- The Fre protein is a chimera of the NH2-terminal mine the position of cleavage, it is possible that the domain of Flp, P13, residues 1-123, and the COOH-term- source of the catalytic tyrosine could also play a inal domain of Cre, Cre25, residues 119-343 (Figure 2(a) role. For example, in the Cre and Clp heterodimer, and (b)). The P13-coding region of Flp was isolated by it is likely that the scissile bond is in the ``normal'' PCR using as template the pShe2 plasmid that contains the Flp NH -terminal coding region. The 50 NH -terminal position on the top strand for cis cleavage by Cre. 2 2 However, trans cleavage by Clp might shift the site primer, FLPN, was 44 nucleotides long and hybridized to the start of the P13 coding region. This primer contained of cleavage by the incoming tyrosine residue. Simi- an NdeI restriction site (underlined) for later cloning steps larly, Flp cleavage at the bottom strand of the and had the sequence: 50 TAGGGCAGCCATATGCCA- ArkL substrate might be inef®cient if Fre had acti- CAATTTGATATATTATGTAAAACACC 30. vated the phosphodiester bond one nucleotide The COOH-terminal primer was P13B and hybridized core-proximal to the expected site of trans cleavage to the sequence encoding the amino acid residues 117- by Flp. The cleavage sites for Flp in the FRT site 123 of the P13 coding region. This primer contained a are 8 bp apart whereas the scissile bonds for Cre in BamHI restriction site for later cloning steps (underlined) the lox site are 6 bp apart. These differences in spa- and had the sequence: 50 TCTAGGGGATCCGCAAAC- cing may have in¯uenced cross-core interactions TACTTACAATATCAGT 30. between the native recombinases and the chimeras PCR reactions contained 10 mM KCl, 20 mM Tris-HCl (pH 8.8), 10 mM (NH ) SO , 2 mM MgSO , 0.1 % (v/v) and hence the ef®ciency and sites of cleavage. 4 2 4 4 Unfortunately, the SDS-PAGE analysis we used to Triton X-100, (ThermoPol buffer, New England BioLabs), 1 mM each primer FLPN and P13B, 0.8 pM pShe3 tem- distinguish the covalent complexes of the different plate, and 400 mM each dNTP (Pharmacia) in a total proteins was not sensitive enough to determine the reaction volume of 100 ml. Reaction mixtures were pre- site of cleavage to one nucleotide resolution. heated to 95 C for four minutes, two units of Vent DNA polymerase were added and the 95 C incubation was continued for 1.5 minutes. Cycling occurred in three Materials and Methods stages. The ®rst stage consisted of two cycles of 50 C, one minute, 72 C, 42 seconds, 95 C, 30 seconds. The Enzymes second stage consisted of 30 cycles of 65 C, one minute, 72 C, 42 seconds, 95 C, 30 seconds. The third stage con- All enzymes were obtained from New England Bio- sisted of a 65 C, one minute annealing step and a 72 C, Labs and used according to the manufacturer's instruc- ®ve minutes extension. Enzyme, dNTPs and oligonucleo- tions. tide primers were removed by the QIAquick column. The puri®ed 365 bp PCR product was double-digested Plasmids with NdeI and BamHI enzymes, and was puri®ed by the QIAquick column as before. The pShe6 vector, contain- The pShe2, pShe3, pShe5, pShe6, pShe11 and pLD3 ing the full Cre coding sequence fused to a 10Â His plasmids have been described previously (Shaikh, 1997; NH2-terminal tag, was digested with BamHI. The 710 bp 46 Chimeras of the Flp and Cre Recombinases

DNA fragment that contains the sequences from amino An aliquot of this mixture was transformed into the XL1- acid 119 of Cre to beyond the stop codon for Cre in Blue strain and the desired plasmid, encoding the Cre13- pShe6 (Cre25-coding fragment) was isolated from a 1.0 % P32 chimera, Clp, in frame with an NH2-terminal 10Â (w/v) agarose gel and puri®ed using the QIAquick col- His-tag, was denoted pShe20 and maintained in both the umn. Similarly, the pShe6 plasmid was digested with XL1-Blue strain and the BL21 expression strain. The pre- NdeI and BamHI enzymes, and the 5.7 kbp vector frag- dicted molecular mass of the chimera is approximately ment was isolated and puri®ed as before. The 360 bp 48 kDa (45 kDa Clp 3 kDa His leader sequence). NdeI-BamHI digested PCR product, the 710 bp BamHI- BamHI Cre25-coding fragment and the NdeI-BamHI digested She6 vector fragment were ligated together Expression and purification of His-Fre and His-Clp using T4 DNA ligase. An aliquot of the ligation mixture was used to transform competent XL1-Blue cells (Strata- The BL21 strains transformed with the Fre and Clp 0 q expression vectors were grown at 37 C and protein gene: F LacI /recA1 hsdR17 (rK mK)) as described by Sambrook et al. (1989). Resulting colonies from overnight expression induced with 1 mM isopropyl-b-D-thiogalac- growth on LB-ampicillin agar plates were picked and toside for 30 minutes at 37 C; the culture was shifted to grown in liquid medium. DNA from individual isolates room temperature for four hours. When produced in was puri®ed as described by Sambrook et al. (1989). 5 ml cultures the proteins were at least 85 % soluble as A unique Bsu36 restriction site in the P13-coding region assayed by SDS-PAGE and Coomassie Blue staining and a unique XhoI restriction site in the Cre25-coding after sonication and low-speed centrifugation. The cell region were used in restriction analysis of the DNA and pellet from a 500 ml culture was resuspended in three con®rmed that all clones contained both the P13 PCR volumes of sonication buffer (50 mM sodium phosphate product and the Cre25-coding fragment as inserts in the (pH 8.0), 300 mM NaCl). Sonication was done with six correct orientation. One such isolate was sequenced to 20 second bursts (40 % gain, Vibra Cell sonicator) on ice verify the accuracy of the PCR and the cloning. The with two minute intervals between bursts. The sonicate desired plasmid, encoding the P13-Cre25 chimera, Fre, in was centrifuged at 100,000 g for one hour at 4 C. All subsequent manipulations were done at 4 C. The super- frame with an NH2-terminal 10Â His-tag, was denoted pShe21 and maintained in both the XL1-Blue strain and natant was applied to a 2 ml Ni-NTA agarose column the expression strain BL21 (DE3, pLysS), (Studier et al., (Qiagen) previously equilibrated in wash buffer (50 mM 1990). The predicted molecular mass of His-Fre is sodium phosphate (pH 8.0), 300 mM NaCl, 10 % glycer- approximately 41 kDa (38 kDa Fre 3 kDa His leader ol). The column was washed with ®ve column volumes sequence). of wash buffer and then successively with three column volumes of wash buffer containing 50 mM, 75 mM, 100 mM and 125 mM imidazole to remove proteins binding non-speci®cally to the column. The His-Fre pro- Construction of His-Clp expression vector tein was eluted from the column with 175 and 200 mM imidazole in wash buffer. Similarly, the His-Clp protein The Clp protein is a chimera of the NH -terminal 2 was eluted between 150 and 200 mM imidazole-contain- domain of Cre, Cre13, residues 1-122, and the COOH- ing buffer (Figure 2(c)). Both proteins were greater than terminal domain of Flp, P32, residues 124-423 (Figure 2). 95 % pure as assayed by SDS-PAGE and migrated as The Cre13-coding region of Cre was isolated by PCR predicted by their respective molecular mass. The yields using the pShe6 plasmid as template. The 50 NH -term- 2 of His-Fre and His-Clp were 10 mg/l of induced culture. inal primer, NCC, was 32 nucleotides long and hybri- Fractions were pooled and the imidazole-containing buf- dized to the 10Â His-leader sequence upstream of the fer was exchanged for a buffer containing 20 mM Hepes start codon of the Cre gene in pShe6. The primer (pH 7.8), 0.1 mM EDTA and 0.1 mM (NH ) SO using contained an NcoI restriction site (underlined) for 4 2 4 10DG desalting columns (BioRad). Initially, both chi- later cloning steps and had the sequence: meric proteins caused a non-speci®c stimulation of Cre 50 TAGGGCTACCATGGGCCATCATCATCATCATC 30. 0 binding as assayed in gel mobility shift assays (data not The 3 COOH-terminal primer was Cre13B and hybri- shown). This activity was removed by passing the His- dized to the sequence encoding amino acid residues 116- Fre and His-Clp samples through a Centricon 50 fol- 122 of the Cre protein. Cre13B was 43 nucleotides long lowed by a Centricon 10 spin-concentrator (Millipore). and contained sequences that encode the amino acid Even though the SDS-PAGE pro®le did not differ from residues 124-128 of the Flp protein as well as a BstBI 0 the original eluted fractions, apparently the molecular restriction site (underlined) immediately 5 to the start of mass sieves used in the concentrators removed factors the Cre portion (italics) of the primer. The sequence of 0 that stimulated Cre binding. These His-Fre and His-Clp this primer was: 5 TCTAGGTTCGAACTG- samples were resuspended in buffer containing 20 mM TAATTGTTTTCGGATCCGCCGCATAACC 30. Hepes (pH 7.8), 0.1 mM EDTA and 0.1 mM (NH4)2SO4 A two-stage PCR was done as described before using and their concentrations were determined by the Brad- Vent DNA polymerase, these two primers and pShe6 as ford assay (Bradford, 1976) with reagents and IgG stan- the template. The ®rst stage consisted of 30 cycles of dard from BioRad. Both proteins were stored at 70 C. 68 C, one minute, 72 C, 42 seconds, 95 C, 30 seconds. The second stage consisted of a 68 C, one minute hybridization and a 72 C, ®ve minutes extension. The Gel mobility shift analysis of proteins PCR product was puri®ed and the 455 bp Cre13-PCR product was digested with NcoI and BstBI. The pShe2 Gel mobility shift assays were used to test the ability plasmid encodes the COOH-terminal P32 fragment of of His-Fre, His-Clp, Cre and Flp, and the COOH-

Flp in frame with an NH2-terminal His-tag. This plasmid terminal peptides of Flp and Cre to bind to a number of was digested with NcoI and then BstBI and the 5.8 kbp substrates shown in Figure 1 (Andrews et al., 1987; DNA vector fragment isolated from an agarose gel. The Wierzbicki et al., 1987). The design of the substrates is 450 bp NcoI-BstBI digested PCR product and the 5.8 kbp described in detail in Results. Substrates were singly NcoI-BstBI digested pShe2 vector were ligated together. 50-end labeled and the binding reactions were performed Chimeras of the Flp and Cre Recombinases 47 in 20 ml containing 50 mM Tris-HCl (pH 7.4), 30 mM Argos, P., Landy, A., Abremski, K., Egan, J. B.,

NaCl, 2 mM MgCl2, 3 % glycerol and 0.33 mg/ml of Haggard-Ljungquist, E., Hoess, R. H., Kahn, M. L., sonicated and denatured calf thymus DNA as competi- Kalionis, B., Narayana, S. V. & Pierson, L. S. D., tor. Binding reactions were done with 0.077 pmol of sub- et al. (1986). The integrase family of site-speci®c strate and incubated at 30 C for 45 minutes with Cre, recombinases: regional similarities and global diver- Cre25, His-Fre, Flp, His-P32 peptide and His-Clp. Actual sity. EMBO J. 5, 433-440. levels used for each protein are described in Results and Bradford, M. M. (1976). A rapid and sensitive method the Figures. Reactions were ended by the addition of for the quantitation of microgram quantities of pro- 2 ml of binding dye solution (1 mM Tris-HCl (pH 7.4), tein using the principle of protein-dye binding. 0.1 mM EDTA, 10 mg/ml BSA, 2 % glycerol, and 0.01 % Anal. Biochem. 72, 248-254. (w/v) xylene cyanol and bromophenol blue dyes) and Chen, J. W., Lee, J. & Jayaram, M. (1992). DNA cleavage then run on a non-denaturing 10 % (w/v) polyacryl- in trans by the active site tyrosine during Flp amide gel at 4 C (Andrews et al., 1987). recombination: switching protein partners before exchanging strands. Cell, 69, 647-658. Craig, N. L. (1988). The mechanism of conservative site- Isolation and analysis of complexes speci®c recombination. Annu. Rev. Genet. 22, 77-105. Dombroski, A. J., Walter, W. A. & Gross, C. A. (1993). Analytical binding reactions were scaled up for pre- Amino-terminal amino acids modulate sigma-factor parative reactions from 20 to 50 ml containing 100 mM DNA-binding activity. Genes Dev. 7, 2446-2455. Tris-HCl (pH 7.4) 60 mM NaCl, 4 mM MgCl , 6 % gly- 2 Esposito, D. & Scocca, J. J. (1997). The integrase family cerol and 0.66 mg/ml of sonicated and denatured calf of tyrosine recombinases: evolution of a conserved thymus DNA as competitor. Binding reactions were active site domain. Nucl. Acids Res. 25, 3605-3614. done with 0.385 pmol of substrate, and were incubated Evans, B. R., Chen, J. W., Parsons, R. L., Bauer, T. K., at 30 C for 45 minutes and run on a polyacrylamide gel Teplow, D. B. & Jayaram, M. (1990). Identi®cation as described. For all scaled up reactions, the lower levels of the active site tyrosine of Flp recombinase. Poss- of protein used in the analytical reactions were increased ible relevance of its location to the mechanism of by a factor of 5 for reactions containing both one and recombination [published erratum appears in J. Biol. two proteins (see Results). After electrophoresis, sub- Chem. (1991) Apr 15; 266(11):7312]. J. Biol. Chem. strate or complexes were located by a brief exposure of 265, 18504-18510. the wet gel to X-ray ®lm. Complexes were electroeluted Gopaul, D. N. & Van Duyne, G. D. (1999). Structure from the gel slices into Centricon 10 spin units (Milli- and mechanism in site-speci®c recombination. Curr. pore) in SDS reservoir buffer (25 mM Tris-HCl (pH 6.8), 250 mM glycine, 0.1 % (w/v) SDS); the sample volume Opin. Struct. Biol. 9, 14-20. was reduced to about 40 ml by centrifugation. SDS Gopaul, D. N., Guo, F. & Van Duyne, G. D. (1998). sample buffer was added to the samples. The samples Structure of the Holliday junction intermediate in were heated to 95 for ®ve minutes and equal counts of Cre-loxP site-speci®c recombination. EMBO J. 17, each sample were analyzed by either 15 % or 17 % poly- 4175-4187. acrylamide SDS-PAGE. The gel was soaked in 50 % Gronostajski, R. M. & Sadowski, P. D. (1985). The FLP methanol, 20 % glycerol, dried and exposed to X-ray recombinase of the Saccharomyces cerevisiae 2 micron ®lm. plasmid attaches covalently to DNA via a phospho- tyrosyl linkage. Mol. Cell. Biol. 5, 3274-3279. Guo, F., Gopaul, D. N. & Van Duyne, G. D. (1997). Structure of Cre recombinase complexed with DNA in a site-speci®c recombination synapse. Nature, Acknowledgements 389, 40-46. Guo, F., Gopaul, D. N. & Van Duyne, G. D. (1999). This work was supported by a grant from the Medical Asymmetric DNA bending in the Cre-loxP site- Research Council of Canada. A.C.S. was supported by a speci®c recombination synapse. Proc. Natl Acad. Sci. University of Toronto Open Scholarship. We thank USA, 96, 7143-7148. Linda Beatty, Linda Lee, Barbara Funnell, Andy Becker Hamilton, D. L. & Abremski, K. (1984). Site-speci®c and Tom Yager for helpful discussions. recombination by the bacteriophage P1 lox-Cre system. Cre-mediated synapsis of two lox sites. J. Mol. Biol. 178, 481-486. References Han, Y. W., Gumport, R. I. & Gardner, J. F. (1993). Complementation of bacteriophage lambda inte- Abremski, K. E. & Hoess, R. H. (1992). Evidence for a grase mutants: evidence for an intersubunit active second conserved arginine residue in the integrase site. EMBO J. 12, 4577-4584. family of recombination proteins. Protein Eng. 5, 87- Hoess, R. H. & Abremski, K. (1990). The Cre-lox recom- 91. bination system. In Nucleic Acids and Molecular Amin, A., Roca, H., Luetke, K. & Sadowski, P. D. (1991). Biology (Eckstein, F. & Lilley, D. M. J., eds), vol. 4, Synapsis, strand scission, and strand exchange pp. 99-109, Springer-Verlag, Berlin. induced by the FLP recombinase: analysis with Hoess, R. H., Abremski, K. & Sternberg, N. (1984). The half-FRT sites. Mol. Cell. Biol. 11, 4497-4508. nature of the interaction of the P1 recombinase Cre Andrews, B. J., Beatty, L. G. & Sadowski, P. D. (1987). with the recombining site loxP. Cold Spring Harbor Isolation of intermediates in the binding of the FLP Symp. Quant. Biol. 49, 761-768. recombinase of the yeast plasmid 2-micron circle to Hoess, R. H., Wierzbicki, A. & Abremski, K. (1986). The its target sequence. J. Mol. Biol. 193, 345-358. role of the loxP spacer region in P1 site-speci®c Arciszewska, L. K. & Sherratt, D. J. (1995). Xer site- recombination. Nucl. Acids Res. 14, 2287-2300. speci®c recombination in vitro. EMBO J. 14, 2112- Hoess, R. H., Abremski, K., Irwin, S., Kendall, M. & 2120. Mack, A. (1990a). DNA speci®city of the Cre recom- 48 Chimeras of the Flp and Cre Recombinases

binase resides in the 25 kDa carboxyl domain of the complementation analysis. Mol. Cell. Biol. 13, 3167- protein. J. Mol. Biol. 216, 873-882. 3175. Hoess, R. H., Wierzbicki, A. & Abremski, K. (1990b). Pan, H., Clary, D. & Sadowski, P. D. (1991). Identi®- Synapsis in the Cre-lox Site Speci®c Recombination cation of the DNA-binding domain of the FLP System. In Structure & Methods (Sarma, R. H. & recombinase. J. Biol. Chem. 266, 11347-11354. Sarma, M. H., eds), vol. 1, pp. 203-213, Human Panigrahi, G. B. & Sadowski, P. D. (1994). Interaction of Genome Initiative & DNA Recombination, Adenine the NH2- and COOH-terminal domains of the FLP Press. recombinase with the FLP recognition target Jayaram, M. (1997). The cis-trans paradox of integrase sequence. J. Biol. Chem. 269, 10940-10945. [see comments]. Science, 276, 49-51. Qian, X. H., Inman, R. B. & Cox, M. M. (1990). Protein- Jayaram, M. & Lee, J. (1995). Return to sobriety after the based asymmetry and protein-protein interactions catalytic party. Trends Genet. 11, 432-433. in FLP recombinase-mediated site-speci®c recombi- Kwon, H. J., Tirumalai, R., Landy, A. & Ellenberger, T. nation. J. Biol. Chem. 265, 21779-21788. (1997). Flexibility in DNA recombination: structure Ringrose, L., Lounnas, V., Ehrlich, L., Buchholz, F., of the lambda integrase catalytic core. Science, 276, Wade, R. & Stewart, A. F. (1998). Comparative kin- 126-131. etic analysis of FLP and cre recombinases: math- Landy, A. (1989). Dynamic, structural, and regulatory ematical models for DNA binding and aspects of lambda site-speci®c recombination. Annu. recombination. J. Mol. Biol. 284, 363-384. Rev. Biochem. 58, 913-949. Sadowski, P. D. (1993). Site-speci®c genetic recombina- Lee, G. & Saito, I. (1998). Role of nucleotide sequences tion: hops, ¯ips and ¯ops. FASEB J. 7, 760-767. of loxP spacer region in Cre-mediated recombina- Sadowski, P. D. (1995). The Flp recombinase of the 2- tion. Gene, 216, 55-65. micron plasmid of Saccharomyces cerevisiae. Prog. Lee, J., Whang, I., Lee, J. & Jayaram, M. (1994). Directed Nucl. Acid Res. Mol. Biol. 51, 53-91. protein replacement in recombination full sites Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecu- reveals trans-horizontal DNA cleavage by Flp lar Cloning: A Laboratory Manual, 2nd edit., Cold recombinase. EMBO J. 13, 5346-5354. Spring Harbor Laboratory Press, Cold Spring Har- Lee, J., Jayaram, M. & Grainge, I. (1999). Wild-type Flp bor, NY. recombinase cleaves DNA in trans. EMBO J. 18, Shaikh, A. C. (1997). Assays of the functional domains 784-791. of the Cre and Flp recombinases. MSc thesis, Mack, A., Sauer, B., Abremski, K. & Hoess, R. (1992). University of Toronto. Stoichiometry of the Cre recombinase bound to the Shaikh, A. C. & Sadowski, P. D. (1997). The Cre recom- lox recombining site. Nucl. Acids Res. 20, 4451-4455. binase cleaves the lox site in trans. J. Biol. Chem. Nunes-Duby, S. E., Tirumalai, R. S., Dorgai, L., Yagil, E., 272, 5695-5702. Weisberg, R. A. & Landy, A. (1994). Lambda inte- Studier, F. W., Rosenberg, A. H., Dunn, J. J. & grase cleaves DNA in cis. EMBO J. 13, 4421-4430. Dubendorff, J. W. (1990). Use of T7 RNA polymer- Nunes-Duby, S. E., Kwon, H. J., Tirumalai, R. S., ase to direct expression of cloned genes. Methods Ellenberger, T. & Landy, A. (1998). Similarities and Enzymol. 185, 60-89. differences among 105 members of the Int family of Subramanya, H. S., Arciszewska, L. K., Baker, R. A., site-speci®c recombinases. Nucl. Acids Res. 26, 391- Bird, L. E., Sherratt, D. J. & Wigley, D. B. (1997). 406. Crystal structure of the site-speci®c recombinase, Pan, G. & Sadowski, P. D. (1993). Identi®cation of the XerD. EMBO J. 16, 5178-5187. functional domains of the FLP recombinase. Separ- Vetter, D., Andrews, B. J., Roberts-Beatty, L. & ation of the nonspeci®c and speci®c DNA-binding, Sadowski, P. D. (1983). Site-speci®c recombination cleavage, and ligation domains. J. Biol. Chem. 268, of yeast 2-micron DNA in vitro. Proc. Natl Acad. Sci. 22546-22551. USA, 80, 7284-7288. Pan, G., Luetke, K. & Sadowski, P. D. (1993). Mechan- Wierzbicki, A., Kendall, M., Abremski, K. & Hoess, R. ism of cleavage and ligation by FLP recombinase: (1987). A mutational analysis of the bacteriophage classi®cation of mutations in FLP protein by in vitro P1 recombinase Cre. J. Mol. Biol. 195, 785-794.

Edited by M. Gottesman

(Received 13 April 2000; received in revised form 5 June 2000; accepted 13 June 2000)