General Strategies, Experimental Rationale, and Interpretations
This section can be divided into two parts. The first set of experiments ( and ) was aimed at mapping the peptide regions of Flp that are protected from protease digestion when it is bound to its DNA substrate or to the Holliday intermediate of recombination. The protection profile could be overlaid on the Cre-DNA crystal structure to assess the overall degree of fit between the modes of substrate recognition in the two systems. The second set of experiments (–) addresses how two alternative active site orientations, one for cis and the other for trans cleavage, can be derived from the same global architecture of DNA-protein interactions.
Chymotryptic footprint patterns of N-tagged Flp or Flp(Y343F) and C-tagged Flp
Footprinting profiles of N-tagged Flp with trypsin and proteinase K, distribution of protected regions on the Flp protein
The DNA substrate in the majority of protease footprinting experiments was the Flp full site (containing two Flp-binding elements, S2; see ). In one set of experiments, a half-site (containing one Flp-binding element, S1; ) and a Holliday junction substrate (containing four Flp-binding elements, HS4; ) were also employed. The nonspecific DNA used as a control in these reactions was the same size as S2 and was obtained by changing 12 positions within the left and right Flp-binding elements () to the “least favored” bp, as deduced from the analysis of Senecoff et al
The Flp protein derivatives used for the footprinting assays were tagged using the T7 epitope tag or the S peptide tag at either the amino terminus or at the carboxyl terminus. They are schematically represented in . T7-Flp (row 1, ) and His6-S-Flp(Y343F) (row 3, ) harbored the T7 and S tags, respectively, at their amino termini; T7-Flp-S contained both tags, T7 at the amino terminus and S at the carboxyl terminus (row 2, ).
Wild type Flp or Flp(Y343F) was subjected to partial proteolysis in the presence or absence of DNA under a set of standardized conditions ( and ). The protease-digested samples were gel-fractionated, transferred to PVDF membranes, and probed with the T7 antibody or the S protein probe. The ladder of bands observed, in the order of increasing molecular mass, provides a measure of the relative distances of the protease-sensitive sites from the position of the tag. In principle, for every band derived from the protein tagged at the amino terminus upon digestion with a given protease, there should be a corresponding band from the protein tagged at the carboxyl terminus such that the two together add up to the full-length protein. The footprinting profiles obtained with wild type Flp or Flp(Y343F) were virtually the same, except that the protein-DNA adduct resulting from Flp cleavage was absent in the Flp(Y343F) reactions.
For each of the proteases used, the digestion patterns of Flp in the absence and presence of the nonspecific DNA fragment (NS) were not identical. This is easily seen by comparing lanes 4 to lanes 2 in , for example. The difference could be either due to an effect of DNA on the activity of the proteases or due to some nonspecific Flp-DNA interaction. In deriving the footprint profiles, only those proteolytic bands that could be distinguished between the Flp/S2 and the Flp/NS samples were considered as significant, except in one case that is explained below. Prior to assigning the composite protected peptide regions (see ), the molecular masses of proteolytic products were corrected for the size differences in the respective tags they harbored (1.5 kDa for the amino-terminal T7 tag; 2 kDa for the carboxyl-terminal S tag; 5 kDa for the aminoterminal His6-S tag).
Chymotryptic Footprinting of Flp Bound to DNA Substrates Containing One, Two, and Four Flp-binding DNA Arms
A normal Flp recombination event is initiated by the binding, separately, of two Flp monomers from solution to each of the two binding elements of the minimal recombination target site. Dimerization of Flp occurs only between two DNA-bound Flp monomers. Furthermore, the strand exchange mechanism gives rise to a Holliday junction as an obligatory intermediate in the reaction. We wished to probe the footprinting patterns of Flp bound to DNA harboring a single Flp-binding element (S1, ), or to a functional substrate containing two binding elements (S2, ), or to a Holliday junction containing four Flp binding arms (HS4, ). The results of these assays using partial proteolysis by chymotrypsin are presented in . The two profiles shown in A and B correspond to identical sets of chymotryptic reactions fractionated differentially to highlight the region corresponding to 30–14 kDa in A and to expand that spanning 50–25 kDa in B.
The overall protection patterns were quite similar with the three types of substrates (lanes 3–5 of ). A conspicuous region of protection spanning the 22–30-kDa range was seen in all three cases (PII in ). Within PII, a doublet at approximately 24–25 kDa showed a higher degree of protection in Flp bound to S2 than in Flp bound to S1 or HS4. Although not readily apparent in , this difference can be seen by comparing lanes 3 and 5 to 4 in ’ (A’ shows the relevant region of A for a longer exposure of the Western blot luminogram). Similarly, the two bands at 14–16 kDa (PI) showed significant protection in the Flp-S2 complex but were essentially unprotected in the Flp-S1 and Flp-HS4 complexes (compare lane 4 to lanes 3 and 5 in ). Finally, a 34–35-kDa proteolytic fragment was absent from the S1, S2, and HS4 lanes, defining a third chymotryptic footprint, PIII (). The possible significance of the band marked PIV in is discussed in reference to the results shown in (see below).
An interesting feature of the footprints was the enhancement of a doublet band just above PIII (labeled E; approximately 38–39 kDa) in the presence of the Flp substrates (lanes 3–5;
) but not in the presence of nonspecific DNA (compare lane 6 to lane 2 in ). The cleavage enhancement was stronger for S1 and S2 compared with HS4. This hypersensitivity to chymotrypsin is suggestive of a potential conformational transition, commensurate with DNA binding, near the carboxyl-terminal portion of Flp adjacent to PIII. This region maps to the amino-terminal side of the active site tyrosine (Tyr-343) of Flp.
The band migrating slower than 50 kDa in lanes 3–5
), was the result of DNA cleavage by Flp and the formation of the covalent Flp-DNA adduct. It should be pointed out that, a priori
, the presence of CL in the Flp/S1 case (lanes 3
) was not expected, since it takes two Flp monomers to assemble a single active site for strand cleavage (7
). However, once a Flp monomer has bound its cognate sequence in S1 and oriented the scissile phosphodiester, fortuitous transient binding of a second Flp monomer on the same DNA or encounter with a second DNA-bound Flp monomer may result in strand cleavage.
Substrate-mediated Protection against Chymotrypsin in Flp Tagged at the Carboxyl Terminus or Flp(Y343F) Tagged at the Amino Terminus
To verify the authenticity of the footprints obtained with Flp tagged at the amino terminus, a similar chymotryptic digestion was carried out using Flp tagged at the carboxyl terminus in the presence of S2 or NS. The results are shown in . The sum of the molecular masses of a protected fragment from amino-terminally tagged Flp and its complement from carboxyl-terminally tagged Flp, corrected for the differences in tag sizes, should be approximately 46.5 kDa, the mass of native Flp. Since the amino-terminal T7 tag and the carboxyl-terminal S tag are approximately 1.5 and 2.0 kDa, respectively, the observed sum should be 46.5 + 3.5 = 50 kDa. The regions of protection marked P’I (32–36 kDa), P’II (20–28 kDa), and P’III (14–16 kDa) match the PI (14–16 kDa), PII (22–30 kDa), and PIII (34–35 kDa) protections, respectively, deduced from the data in .
The protected band at 7–8 kDa (p
, ) is likely the result of chymotryptic cleavage at one or more tyrosine residues at positions 361, 362, and 364. This assignment, although not confirmed by sequencing, is consistent with a previously mapped V8 cleavage site at Glu-370 (8
). These are the only aromatic residues within a 20 amino acid segment spanning Glu-370 (from 361 to 380). The approximately 10-kDa band P’IV (lane 2
, ) protected in the S2 and NS lanes (lanes 3
, ) signifies cleavage at Tyr-343 (or Phe-343 for Flp(Y343F)), as determined by amino-terminal sequencing (see ). The high protease sensitivity of the peptide segment housing the Flp catalytic tyrosine was noted in a previous study as well (8
). Protease susceptibility in the vicinity of the catalytic tyrosine has been also observed for the lambda Int protein (29
). We interpret the P’IV protection to be functionally relevant even though it was observed in both S2- and NS-containing reactions. Since Tyr-343 forms a covalent adduct with the DNA backbone upon strand cleavage, it has to be in close contact with DNA. The band corresponding to P’IV expected from Flp tagged at the amino terminus is approximately 40 kDa (the combined masses of Flp together with the carboxyl- and amino-terminal tags should be 50 kDa). Such a band was observed in the partial chymotryptic digest of N-tagged Flp (PIV, lane 2
, ) and was not detected in the presence of the Flp substrates (lanes 3–5
, ). As with P’IV, PIV was also absent in the reaction containing the nonspecific DNA fragment (lane 6
The pattern with the C-tagged Flp also included a 46-kDa protected polypeptide (p’0, ). The equivalent 4-kDa protection from N-tagged Flp would not have been detected under the electrophoretic conditions employed in .
Note that the Flp cleavage product (>50 kDa; CL
) was not seen with the C-tagged Flp (absence of CL in lane 3
of ). This result is consistent with previous observations that even minor alterations of the carboxyl terminus of Flp (in the present instance, the addition of the S tag) result in loss of catalytic activity (30
). The crystal structure of the Cre-DNA complex also indicates that the carboxyl-terminal region is responsible for the allosteric activation of one Cre monomer within a dimer to cleavage competence (10
The chymotryptic footprints obtained with N-tagged Flp(Y343F) in the presence of the S2 DNA () were qualitatively identical to those yielded by wild type Flp and S2 (compare lane 3 of with lanes 4 in ). The three protected areas (PI, PII, and PIII) were clearly discernible, and specific bands within a region of protection (for example, the doublets harbored by PI and PII) were slightly shifted up in agreement with the longer amino-terminal tag in Flp(Y343F). As expected from the inability of Flp(Y343F) to cleave DNA, no band migrating above 50 kDa was observed in its footprint.
Footprints of Flp Derived by Tryptic and Proteinase K Digestion
The general inferences drawn from the chymotrypsin assays were further tested by probing N-tagged Flp in the presence of S2 with two other proteases (). Individually, the information contained within each of the two footprint patterns was sparse relative to the chymotryptic data. Collectively, however, they support and augment the conclusions derived from the latter. For instance, the PII footprint (the protected doublet at approximately 24 kDa) was detected in the trypsin digestion (lane 3, ). Similarly, the approximately 33-kDa band protected in the proteinase K assays (p3, lane 3, ) maps adjacent to the amino-terminal border of the PIII patch. The proteinase K profile also contained a 47–48-kDa protected band (p5, lane 3, ), placing the corresponding proteolytic site near the extreme carboxyl-terminal region of Flp.
Summary of the Footprints of the Flp-DNA Complexes
The outcomes of the footprint analyses are schematically diagrammed in . The diagram below
relates the protected regions to the location of the principal protease-sensitive sites mapped within Flp when it was not bound to DNA. Because the cleanest internally cleaved and electrophoretically fractionated polypeptide fragments (suitable for amino-terminal microsequencing) were obtained with V8 protease, the primary landmarks in are provided by Glu residues (Glu-128, Glu-271, and Glu-370). In addition, the protected P’IV cleavage fragment from the chymotrypsin digest of C-tagged Flp was characterized as the product of cleavage at Tyr-343. Hyper-cleavage of unbound Flp at Asp-134 and Asp-332 that we observed during digestion with endoproteinase Asp-N would be consistent with V8 and chymotryptic cleavages at Glu-128 and Tyr-343, respectively (data not shown). Earlier experiments had revealed that one of the preferred cleavages by trypsin also occurs proximal to Tyr-343, at Arg-340 (8
The three areas of protection (PII, PIII, and PIV/p’5) spanning the major portion of the carboxyl-terminal domain of Flp would be consistent with the extensive contacts between the corresponding region of the Cre protein and its binding site revealed by the crystal structure (10; see also). The two protected segments within the Flp amino-terminal domain (p’0 and P1) would also agree with the crystal structure in which the amino- and carboxyl-terminal domains of Cre form a crescent over the DNA target. However, the predicted secondary structure of Flp, as well as the footprinting data, suggests that the amino-terminal domains of Flp and Cre may be less related to each other. The significance of the footprints in DNA recognition by Flp is further addressed under “Discussion.”
Comparison of Flp footprints to DNA contacts observed in the crystal structure of Cre bound to DNA; a model for DNA recognition by Flp
Overall, the footprinting analysis on Flp conforms to the reasonable expectation that the mode of substrate binding is conserved among members of the Int family. This provides the justification for the functional assays for Flp and Cre described below.
Cleavage Assays in Bulge-containing DNA Substrates
The logic of forcing Flp and Cre substrates into similar configurations for mapping strand cleavage was based on our previous work (22
) with Flp substrates containing nucleotide bulges in specific strands. NMR and FRET studies have shown that the bulge acts as a one-way hinge in DNA (25
). Depending on the strand that it resides in and its position within that strand, a bulge promotes the bending of the flanking DNA arms in one direction (away from the bulge) but not the other. Thus, by placing nucleotide bulges at equivalent positions within DNA sequences of similar size and organization (as is the case for the Flp and Cre target sites), they can be forced to assume similar geometries. The structural features of the different bulged substrates are illustrated in the appropriate figures and described in the corresponding text.
Preferential Cleavage of Bulge-containing Strands by Cre and Flp
The experiments displayed in test whether trans cleavage by Flp and cis cleavage by Cre can be related by I and II or by I and III in .
For the Flp system, strand-specific nucleotide bulges between the central two nucleotides of the spacer can constrain the substrate geometry to two nearly exclusive states that yield left cleavage in one case and right cleavage in the other (22
). It is always the strand harboring the bulge that becomes cleavage-susceptible. Furthermore, this selective cleavage occurs by the trans
mechanism, as would be expected from the known behavior of the Flp protein (7
). We have now constructed bulged synthetic Cre substrates (analogous to the Flp substrates, containing nucleotide bulges in the middle of the spacer), and we assayed their cleavage preferences in the presence of wild type Cre (). The corresponding results with bulged Flp substrates and Flp are shown for comparison (see Ref. 22
In the control substrate with no bulges, there was an intrinsic 3–4-fold bias that favors left end cleavage by Cre over the right end (, lane 2). When the bulge was located in the top strand (the strand containing the labile phosphate at the left end), cleavage was directed almost exclusively to the left end, magnifying the natural cleavage bias (, lane 4). When the bulge was present on the bottom strand, cleavage became biased in the opposite direction, toward the right end by a factor of 3 (, lane 6; CR:CL ~3:1). By taking into account the inherent cleavage preference of the control substrate, the bottom strand bulge enhances the odds of right cleavage by a factor of 9–12. This strong proclivity for cleavage of the bulge-containing strand was identical to that observed with Flp (, lanes 7–12). The Flp substrates in the reactions represented by lanes 10 and 12 contained the bulge on the top and bottom strands, respectively.
Thus, for a defined geometry of their substrates (imposed experimentally by the strand bulge), the cis
cleaving Cre and the trans
cleaving Flp orient the reactive tyrosine toward the same end of the spacer (as in I and III of ). It should be noted that the bulge geometry affects cleavage directly and not indirectly by slowing down the reverse reaction that reseals a broken strand. The cleavage bias observed in substrates containing spacers with centrally placed bulges was identical for Flp as well as the joining incompetent variant Flp(H305L) (22
). We assume that this condition holds for the Cre reactions as well.
There is one report in literature (32
) that, in contradiction to the crystal structure (10
), claims trans
cleavage by Cre. Our repeated efforts have failed to provide any evidence of Cre cleaving substrates with or without bulges in the trans
The criterion set by us for trans
cleavage was a cleavage followed by joining reaction when a recombinase mutant lacking the active site tyrosine was paired with an RHR triad mutant. This is because a cleavage produced adjacent to the tyrosine mutant (as is the case for trans
) can readily proceed through the joining step. The active site tyrosine is not required for strand joining. In every situation that Flp answered the test (7
), Cre failed. Thus, in our assays, Cre and Flp are truly cis
Scanning the Flp and Cre Spacer with Bulges for Position-dependent Cleavage Bias
If the bulge induces similar DNA geometry in the Flp and Cre substrates (as we have supposed), the position effects on strand cleavage displayed by the spacer bulges should be similar for both the Flp and Cre systems. The experiments that address this issue are summarized in –.
Strand cleavage by Flp(H305L) and Flp in substrates with spacer bulges and spacer mismatches
To simplify interpretations, the following points may be emphasized. Strand cleavage assays for the wild type recombinases normally measure the balance between the cutting (forward) and joining (reverse) reactions. In the case of Flp, strand joining can be suppressed by introducing mismatches in the spacer immediately adjacent to the cleavage site or by using the mutant protein Flp(H305L). The assays shown in utilize a combination of both these conditions to circumvent possible differential effects on the joining reaction of bulges neighboring a scissile phosphodiester versus bulges internal to the spacer. Currently, we do not have a mutant of Cre that catalytically mimics Flp(H305L), being normal in cleavage but defective in strand joining. In the Cre assays, done with the wild type protein (), bulges immediately adjacent to the cleavage positions were omitted to avoid “mismatch” effects on the joining reaction. Comparison of the cleavage results with wild type Cre and Flp(H305L) is valid qualitatively. Quantitatively, the Flp (H305L) reactions yielded higher levels of cleavage as expected (compare to ).
Bulges within the Flp Spacer
In , the outcomes from a cleavage reaction of Flp(H305L) with substrates containing spacer bulges at various positions are displayed. The lack of cleavage by Flp(H305L) at a given spacer end in the presence of a neighboring bulge (absence of CL in lane 6
and of CR in lane 15
of ) was expected. The mutant is known to cut poorly at mismatched spacer ends (a bulge would effectively mimic base noncomplementarity; see also ). By contrast, as expected from the blockage in strand sealing, wild type Flp yielded enhanced cleavage adjacent to the bulge (CL
CR in lane 5
CL in lane 14
of ). In plotting the cleavage bias in , the values of CL and CR for bulges located at the left and right spacer ends, respectively, were derived from the wild type Flp reaction (CL
from lane 5
from lane 14
of ). All other values were from the Flp(H305L) reactions. Thus, represents the cleavage profile, with minimal strand religation, as a function of the bulge location within the spacer. Even without these adjustments to the cleavage outputs by Flp(H305L), the patterns in (see below) would not be significantly altered qualitatively.
The cleavage yields by Flp(H305L) from , CL and CR, respectively, were calculated (as described under “Materials and Methods”) and graphed as a function of the bulge position (indicated by the vertical arrows on the abscissa)
The results displayed in document the differential cleavage competence of Flp(H305L) in response to mismatches at the ends of the spacer or internal to it. As shown in lanes
2 and 3
, end mismatches severely compromised the activity of Flp(H305L) but not of wild type Flp. By contrast, base non-complementarity at the spacer center did not affect cleavage by Flp(H305L) (lane 6
of ). Note that the differences in cleavage by wild type Flp in lanes 2
of reflect the role of “end homology” in cleavage reversal, namely the slow down in joining due to mismatched bases adjoining the nick (33
The cleavages at the left and right spacer ends for the different bulge positions () were quantified as a weighted fraction of the substrate strand converted to CL or CR (see “Materials and Methods”) and are graphically represented in . Large cleavage discriminations were observed when the bulge was located at or near the center of the spacer. A strong diminution in the bias, or even its reversal, was observed at the spacer extremities. This pattern holds true for bulges on the other strand as well. The dashed line in represents the cleavage profile obtained for substrates containing bulges on the bottom strand (raw data not shown). The two cleavage curves, drawn as the algebraic sum of the left and right cleavages, demonstrate the inverse relationship in the choice of the phosphodiester target imposed by a matched pair of bulges located at equivalent positions on opposite strands.
The above results suggest that the DNA arms for the centrally placed bulges assume a geometry that most closely resembles their relative disposition within the cleavage-competent complex formed between a Flp dimer and a normal DNA substrate. However, large cleavage preferences were observed even with bulges placed eccentrically within the spacer (between 2 and 3 or 6 and 7; lanes 8 and 12 of ). Presumably, the Flp dimer can manipulate these substrates sufficiently to establish a correctly oriented active site. The cleavage bias for wild type Flp for substrates with bulges at the internal spacer positions (between 2′ and 3′, 4′ and 5′, and 6′ and 7′) was qualitatively the same as that for Flp(H305L) (data not shown), further indicating that cleavage bias was a function of substrate geometry and did not result from indirect effects of bulges on strand ligation.
Bulges within the Cre Spacer
The cleavage patterns obtained with wild type Cre for substrates containing bottom strand spacer bulges are arranged in . In the plot showing the cleavage yields as a function of the bulge location (), the CL yields were divided by a correction factor of 4 to compensate for the cleavage bias in the native substrate (lane 2 of ; see also “Materials and Methods”). The dependence and the magnitude of the bias on the bulge position qualitatively paralleled those observed for the Flp substrate (shown in and ). The bias was strongest at an internal position within the Cre spacer (bulge between 2′ and 3′; ) and either faded or reversed itself as the bulge was moved to either side (between 1′ and 2′ to the right; and between 3′ and 4′, 4′ and 5′, and 5′ and 6′ to the left; ). As already noted, the intrinsic cleavage bias of the control Cre substrate significantly favors the top strand (CL > CR; lane 2, ). Because of the low cleavage susceptibility of the native bottom strand, spacer bulges introduced on this strand provide a strikingly clear cut demonstration of their role in determining the choice of strands as targets for cleavage.
The similarity in the correlation between cleavage bias and bulge location for the Flp and Cre substrates supports the functional similarity of the bulge-containing substrates in the cleavage reactions by the two recombinases. The cleavage curves
in and are suggestive of a sinusoidal trend, as would be expected for the phasing of the DNA kinks induced by bulges located along the helical path of a DNA strand (reviewed in Ref. 35
). Since the functional spacer length for recombination (the distance between the scissile phosphodiester bonds) is less than 10 bp (6 bp for Cre and 8 bp for Flp), it is not possible to test directly this prediction. In addition, if the primary determinant of the geometry of the protein-DNA complex is the interactions between the two bound recombinase monomers, the “cleavage periodicity” may differ significantly from the helical periodicity of DNA. The effects of the nucleotide bulges are most easily explained by their role in offering steric inducements or posing steric barriers, in a position dependent manner, to the establishment of a particular cleavage geometry as follows: one that targets the left spacer end, or one that targets the right spacer end.