The Integron Integrase Efficiently Prevents the Melting Effect of Escherichia coli Single-Stranded DNA-Binding Protein on Folded attC Sites

SSB binds folded attC sites.It was shown previously that single-stranded DNA-binding protein (SSB) destabilizes secondary structures in ssDNA during major cellular processes such as DNA replication, repair, and homologous recombination (19, 25, 26). We therefore hypothesized that SSB can bind folded attC sites and destabilize them. To determine if this was the case, we investigated the binding capacity of SSB on folded attC sites using an electrophoretic mobility shift assay (EMSA). In order to mimic the attC site's bottom strand, we used a VCR derivative bottom-strand (VCR_dbs) substrate, a single-stranded oligonucleotide of ∼90 bases containing a shorter version of the VCR_2/1bs site (23). In this substrate, the central part of the site (VTS) was removed and replaced with GAA (Fig. 2A), a modification previously described and shown to be fully recombinogenic (23).

• Open in new tab
• Download powerpoint

FIG 2

Comparison of SSB binding to several VCR_2/1 site derivatives. (A) Nucleotide sequences and secondary structures of the bottom strands (bs) of VCR_2/1 site derivatives. Structures were determined by the UNAFOLD online interface at the Pasteur Institute. VCR_dbs is a VCR derivative used in EMSA in which the variable terminal structure (VTS) was replaced by GAA. The black arrow shows the cleavage point in the R box, and the asterisk shows the extrahelical G nucleotide. We introduced mutations (boldface type) into this VCR_dbs site in order to generate, after folding, paired (pd) and unpaired (unpd) VCR_dbs site derivatives. The ΔG (free energy) values of VCR_dbs, paired VCR_dbs, and unpaired VCR_dbs folding are −21.7, −37.2, and −5.3 kcal mol⁻¹, respectively. (B) EMSA of the binding of SSB to VCR_2/1 site derivatives. Equal quantities of radiolabeled VCR site derivative DNA fragments (0.6 pmol) were incubated with increasing amounts (0, 13, 26, 52, and 104 nM) of SSB. Complexes I and II probably represent several degrees of SSB multimerization. (C) Densitometry analysis of EMSA. Binding of SSB was measured by monitoring the fraction (×100) of the labeled DNA substrate bound by SSB in complexes I and II to labeled (bound and free) DNA, detected by an automatic peak search of Image Gauge.

As controls, we constructed two other synthetic attC sites, namely, the unpaired and paired VCR_dbs site derivatives. To obtain a fully paired VCR_dbs site derivative, we introduced specific mutations in the unpaired-strand parts of the folded VCR_dbs site in order to introduce complementarity and pair them. Finally, to obtain the unpaired VCR_dbs derivative, we introduced specific mutations in the double-stranded parts of the folded VCR_dbs site in order to unpair them (Fig. 2A).

The results from EMSA are presented in Fig. 2B. In the presence of SSB, distinct shifted bands (bands I and II), corresponding to SSB-DNA complexes, can be seen for both VCR_dbs and unpaired VCR_dbs site derivatives. Bands I and II probably correspond to different SSB-binding modes and/or different types of intertetramer cooperativity (see Discussion). As expected, for the paired VCR_dbs derivative, we failed to observe distinct and significantly shifted bands. The band intensities on the gels were quantified, confirming the ability of SSB to bind single-stranded parts of attC sites very efficiently (Fig. 2C). We observed, among free DNA, higher electrophoretic migration bands in the VCR_dbs and paired VCR_dbs derivative lanes. Note that these supplementary bands probably correspond to DNA dimer formation favored by the presence of inverted repeats. The presence of these DNA dimers does not alter the interpretation of the results.

SSB destabilizes folded attC sites.We hypothesized that the consequence of SSB binding on the folded attC sites could be a destabilization effect. Indeed, Reddy and collaborators have previously shown the ability of SSB to efficiently and rapidly bind stem-loops, melting these secondary structures and facilitating the binding of RecA (26). To study this, we developed a replication slippage assay. Indeed, a direct correlation between the stability of secondary structures and the frequency of replication slippage was demonstrated previously (27). The model proposed is that the formation of a stem-loop structure between a pair of flanking direct sites during replication permits a slippage event. This results in precise or nearly precise loss of the structured region. As SSB is essential in E. coli, the slippage assays were performed in the wild-type SSB context or with an ssb-200 mutant. The ssb-200 mutation corresponds to a G→A substitution at nucleotide 41, resulting in an amino acid substitution at position 4 (G4D). This mutation induces a tex (transposon excision) phenotype, leading to an increase in the precise excision frequency of a Tn10 derivative (Tn10dKan) (28). Interestingly, Reddy and collaborators showed that tex of the ssb-200 mutant was not affected in any other functions mediated by SSB, such as replication, recombination, or repair. This suggests that the observed tex phenotype of this mutant is probably due only to an incapability of the SSB-200 protein to destabilize stem-loop structures formed by the inverted repeats of Tn10dKan. Thus, if the previously observed binding of SSB on the attC folded sites induces their melting, it is very likely that such an effect would not be observed with SSB-200.

We constructed a pBR325 derivative plasmid carrying an attC_aadA7 site inserted into the cat resistance gene in such way that the loss of this attC site by “replication slippage” would lead to the reconstitution of a functional cat gene (p5087) (see Materials and Methods) (Fig. 3A). This plasmid was transformed into an ssb-200 mutant strain (ω6738 [28]) and into its ssb wt parental strain, GJ1885 (ω6737 [28]). The Cm^r reversion frequency obtained for the parental strain was 5.5 × 10⁻¹¹. Interestingly, the frequency of reversion for the ssb-200 strain was approximately 14-fold higher (7.6 × 10⁻¹⁰) (Fig. 3B). Note that several of the Cm^r clones were sequenced in order to confirm deletion of the attC sites (Fig. 3B). These results suggest that SSB reduces the attC site loss mediated by replication slippage events, probably by binding and destabilizing the single-stranded folded attC site. To confirm that the obtained frequency of replication slippage is not influenced by the local context (transcription, replication, and nucleotide sequences), we performed a set of experimental controls. attC sites were cloned into further pBR derivatives in both orientations within the cat resistance gene. Moreover, the cat gene was cloned in both orientations on the plasmid (see Materials and Methods). In all cases, we obtained the same order of Cm^r reversion frequencies for the four tested constructions in the wt context and in the ssb-200 context (data not shown, and see Materials and Methods; see also Tables S1 and S2 in the supplemental material). As an additional control, we constructed a synthetic derivative of the attC_aadA7 site presenting a lower free energy of single-stranded attC site folding (ΔG = −5.1 kcal mol⁻¹) than the natural attC_aadA7 site (ΔG = −19.1 kcal mol⁻¹). For this, we introduced specific mutations into the double-stranded parts of the natural attC_aadA7 site in order to unpair them. We called this synthetic attC site an unpaired attC_aadA7 site derivative (Fig. 4A). As with the attC_aadA7 site, we also observed a decrease in attC deletion frequencies in the ssb-200 context (∼13-fold) (ω8609) compared to the wt context (ω8605) (Fig. 4B). This result shows that the probable destabilization effect of SSB is not only on the attC sites but also more generally on single-stranded folded secondary structures. We also cloned the unpaired attC_aadA7 site derivative in the opposite orientation within the cat resistance gene, and we obtained the same order of Cm^r reversion and attC deletion frequencies for this construct in both the wt context (ω8604) and the ssb-200 context (ω8608) (data not shown).

• Open in new tab
• Download powerpoint

FIG 3

In vivo effects of SSB and IntI1 on attC site folding. (A) Experimental setup of the replication slippage assay. The model proposed is that folding of the attC site between the pair of flanking direct repeats (EcoRI restriction sites) (black rectangles) permits a replication slippage event. This results in precise or nearly precise attC deletion, thus reconstituting a functional cat resistance gene (Cm^r reversion frequencies). The precise attC deletion was measured by sequencing (attC deletion frequencies). (B) Effects of SSB and IntI1 on Cm^r reversion and attC deletion frequencies (see Materials and Methods). The replication slippage assay was used to study the effects of IntI1Y312F and SSB on attC_aadA7 site folding. The Cm^r reversion frequencies and attC deletion frequencies are represented in the absence (−) and in the presence (+) of IntI1 for the GJ1885 wild-type strain and the GJ1885 ssb-200 strain (GJ1890). The results represent means of four independent experiments (see Materials and Methods). Error bars show the standard deviations.

• Open in new tab
• Download powerpoint

FIG 4

Specificity of the in vivo effects of SSB and IntI1 on attC site folding. (A) Secondary structures of the attC_aadA7 site bottom strand (bs) and the unpaired attC_aadA7 bottom-strand (bs) site derivative. Structures were determined by the UNAFOLD online interface at the Pasteur Institute. The black arrow shows the cleavage point. Primary sequences of the attC sites are shown. The ΔG (free energy) values of attC_aadA7 bs and unpaired attC_aadA7 bs folding are −19.1 and −5.1 kcal mol⁻¹, respectively. (B) The replication slippage assay was used to study the effect of SSB on folding of the attC_aadA7 site (black bar) and the unpaired attC_aadA7 site derivative (gray bar). The relative deletion rates obtained for the ssb-200 strain versus GJ1885 (ssb-WT) are indicated. The results represent means of four independent experiments. (C) The replication slippage assay was used to study the effect of IntI1Y312F on folding of the attC_aadA7 site (black bar) and the unpaired attC_aadA7 site derivative (gray bar). The relative deletion rates obtained in the presence of integrase versus in its absence (intI1+/intI1−) are indicated. The results represent means of four independent experiments.

Altogether, these assays confirmed that SSB can destabilize folded attC sites in a nonspecific manner.

The integrase stabilizes attC site folding.We showed a destabilizing effect of SSB on folded attC sites, in the absence of integrase. As the integrase likely stabilizes the folded attC structure, we could expect this effect to be counterbalanced in the presence of this protein. In order to test this hypothesis, we assessed the effect of IntI1 expression on the slippage frequency in the assay described above. In strain GJ1885 (ω6791), we introduced plasmid p8741 carrying the intI1 gene under the control of the inducible P_bad promoter (ω8749). We chose to use an intI1 gene mutated for the catalytic tyrosine (Y312F) in order to observe the effect of IntI1 binding without an influence of potential recombination events that could occur between copies of the plasmid. It was previously shown that when replacing the reactive nucleophilic tyrosine with the inert phenylalanine (IntI1Y312F), the ability of the integrase to attack the DNA was lost, but the attC site binding capability was preserved (22). The results showed that in the IntI1Y312F expression context (arabinose induction), the Cm^r reversion and attC deletion frequencies were 25-fold higher than those in the absence of the integrase (glucose repression) (Fig. 3B and 4C). This suggests that IntI1, contrary to SSB, increases the attC site loss mediated by replication slippage events, probably by stabilizing folded attC sites. To confirm that this observed stabilizing effect of IntI1 on attC folding is specific to attC sites, we performed the same experiment by using the previously used unpaired attC_aadA7 derivative (Fig. 4A), which theoretically lost its IntI1-binding boxes (ω8752). As expected in this case, we obtained a very low and probably nonsignificant stabilizing effect of IntI1 (6-fold higher) compared to that of the wild-type attC_aadA7 site (25-fold higher) (Fig. 4C). We demonstrated here that the integron integrase stabilizes the folding of attC sites in a specific manner.

In vitro study of effects of IntI1 and SSB on attC site folding.In order to specify the effect of the integrase on attC site folding, we performed additional in vitro experiments. Indeed, it has been shown that IntI1 possesses the intrinsic property to efficiently catalyze strand transfer in vitro between attC fragments in the absence of other bacterial proteins, suggesting the ability of the integrase to favor attC site folding or to stabilize it (29, 30). To confirm this observation, we set up an in vitro folding assay to allow us to easily monitor the single-stranded or double-stranded folding of short DNA fragments containing attC sites. The variations in optical density measured at 260 nm (OD₂₆₀) directly reflect the DNA hybridization modifications. Indeed, the OD₂₆₀ was comparatively lower when the DNA fragment was in the double-stranded form rather than in the single-stranded form. The preheated attC_aadA7 fragment led to an OD₂₆₀ close to 1.2, corresponding to the fully ssDNA form of the fragment (Fig. 5A, red), and the prechilled attC_aadA7 fragment led to an OD₂₆₀ close to 0.6, corresponding to the fully dsDNA form (Fig. 5A, blue). By stabilizing the temperature to 37°C, the OD₂₆₀ of the preheated fragment decreased to 0.9, corresponding to the partially dsDNA form of the fragment. We also obtained the same partially dsDNA structure from the prechilled fragment. These results were also reproduced by using VCR_2/1-containing fragments (Fig. 5B) but not for attI1 sites (sites not forming a hairpin structure) (Fig. 5C). This confirms that attC sites share a partially dsDNA form, in contrast to attI1, which remains fully double stranded. Addition of IntI1 to both prechilled and preheated attC_aadA7 (Fig. 5A) and VCR_2/1 (Fig. 5B) fragments led to an OD₂₆₀ close to 1.2, suggesting that the enzyme favors the formation of the fully single-stranded folded attC site and stabilizes this structure. In contrast, no such effect was found by using the att1I site, suggesting that this activity of IntI1 is highly specific to attC sites. As a supplementary control, we tested a mutated attC_aadA7 site (attC_aadA7m) (Fig. 5D) where the extrahelical G base has been deleted in order to prevent the specific binding of IntI1 on the L box (Fig. 1B) (22). As expected, by using this mutated site, we failed to observe any influence of IntI1 on attC site folding, highlighting the specificity of the IntI1 properties for the attC sites.

• Open in new tab
• Download powerpoint

FIG 5

In vitro effect of IntI1 on attC_aadA7 (A), VCR_2/1 (B), attI1 (C), and attC_aadA7m (D) site folding. The OD₂₆₀ of each fragment (50 μg/ml) was measured at 37°C either after prechilling (prechilled attC_aadA7, VCR_2/1, attI1, and attC_aadA7m) or after preheating for 3 min at 95°C (preheated attC_aadA7, VCR_2/1, attI1, and attC_aadA7m) in the absence of IntI1 as well as after the addition of IntI1 (2 pmol) 30 min after the beginning of the measurement with prechilling (prechilled attC_aadA7 + IntI1, VCR_2/1 + IntI1, attI1 + IntI1, and attC_aadA7m + IntI1) or after preheating at 95°C (preheated attC_aadA7 + IntI1, VCR_2/1 + IntI1, attI1 + IntI1, and attC_aadA7m + IntI1). The OD₂₆₀ measurement of the proteins alone was previously performed and subtracted from the measurements obtained with the mixture in order to determine accurate values for the fragment solution. Results represent means of three independent experiments. Error bars show the standard deviations.

Interestingly, when SSB was used instead of IntI1, perturbation of the fragments' folding was observed only for preheated DNA. This suggests that the protein binds the single-stranded fragments, keeps them in a single-stranded form (Fig. 6A and B), and does not intervene in the attC folding property, as observed with IntI1. This also highlights the specificity of the effect observed with IntI1, confirming that the folding effect was due to an intrinsic property of the enzyme to favor and stabilize the folding of DNA strands and not to an unspecific DNA avidity.

• Open in new tab
• Download powerpoint

FIG 6

In vitro effect of SSB on attC_aadA7 (A) and VCR_2/1 (B) site folding. The OD₂₆₀ of each fragment (50 μg/ml) was measured at 37°C either after prechilling (prechilled attC_aadA7 and VCR_2/1) or after preheating for 3 min at 95°C (preheated attC_aadA7 and VCR_2/1) in the absence of SSB as well as after addition of SSB (2 pmol) 30 min after the beginning of the measurement with prechilling (prechilled attC_aadA7 + SSB and VCR_2/1 + SSB) or after preheating at 95°C (preheated attC_aadA7 + SSB and VCR_2/1 + SSB). The OD₂₆₀ measurement of the proteins alone was previously performed and subtracted from the measurements obtained with the mixture in order to determine accurate values for the fragment solution. Results represent means of three independent experiments. Error bars show the standard deviations.

In conclusion, while both SSB and IntI1 bind the folded attC site, the consequences of their binding would be antagonistic, since SSB seems to destabilize the folded attC site, while IntI1 assists with attC site folding.

SSB does not modulate integron recombination.We tested the recombination of attC sites under conditions of SSB overexpression and in an ssb-200 mutant context. We used an excision assay described previously (31). This assay allowed us to directly measure attC recombination through restoration of the dapA reading frame. In an MG1655 ΔdapA strain, we inserted a plasmid containing a dapA gene interrupted by two specific recombination sites, attI1 and attC_aadA7 (ωB768) (Fig. 7A), attC_aadA7 and attC_ereA2 (ωB661), or attI1 and attI1 (ωB662) (see Table S1 in the supplemental material). Integrase expression causes site-specific recombination and excision of the synthetic cassette, restoring a functional dapA gene and allowing these strains to grow on DAP-free medium (Fig. 7A). We transformed these strains with the pBAD::intI1 multicopy plasmid carrying the intI1 gene under the control of the P_bad promoter (p7383) and a second plasmid, p7305 (pHYD620 from reference 28), containing the ssb gene, ensuring the overexpression of the SSB protein. pCL1920, the pHYD620 progenitor plasmid (p8008) (devoid of the ssb gene), was used as a control. Integrase-driven recombination frequencies were calculated as the ratio of the number of cells growing in the absence of DAP over the total number of cells. Results are presented in Fig. 7B.

• Open in new tab
• Download powerpoint

FIG 7

In vivo effect of SSB on cassette excision. (A) Experimental setup of the cassette excision assay. The dapA gene is interrupted by a synthetic cassette (black line) encountered by the attI1 (black triangle) and attC_aadA7 (white triangle) sites. Recombination leads to the excision of the integron cassette and restores a functional dapA gene. (B) Excision frequencies of the attI1 × attI1, attC_aadA7 × attC_ereA2, and attI1 × attC_aadA7 cassettes in the WT or ssb-200 strain (bottom) and in the presence of the pCL1920 control vector (vector) or the pHYD620 plasmid (ssb⁺), which overexpresses the SSB protein (bottom). Black bars represent the results obtained in the presence of IntI1, and gray bars represent those obtained in the absence of IntI1. “nd” indicates “nondetected” recombination events. Results represent means of three independent experiments. Error bars show the standard deviations.

The frequencies of excision in the absence (vector) and presence (ssb⁺) of SSB overexpression were significantly similar for each tested att combination. These results show that SSB overexpression does not affect integron recombination.

We also tested the recombination of these three att combinations in the above-described ssb-200 context (ωB769, ωB771, and ωB770). For this, we constructed an MG1655 ΔdapA ssb-200 strain (ωB704) (see Table S1 in the supplemental material). Results are presented in Fig. 7B. Once more, excision frequencies in the wt and ssb-200 strains were similar for each tested att combination. These results show that SSB did not affect integron recombination.

As a control, we also performed both assays described above in the absence of integrase expression (Fig. 7B). We observed a few recombination events only for the attI1 × attI1 reaction, probably due to slippage events and/or homologous recombination between both identical 58-bp-long attI1 sites.

Effect of RecA on integron recombination.We tested the recombination of att sites under conditions of RecA overexpression (recA⁺) and in a recA-deleted mutant. Tests were performed by using the above-described in vivo excision assay. We also tested the recombination of the three previously used att combinations (see Materials and Methods; see also Tables S1 and S2 in the supplemental material). The frequencies of excision in the RecA overexpression conditions and recA-deleted strains are significantly similar for each tested att combination (Fig. 8). These results show that RecA does not affect integron recombination.

• Open in new tab
• Download powerpoint

FIG 8

In vivo effect of RecA on cassette excision. Shown are excision frequencies of the attI1 × attI1, attC_aadA7 × attC_ereA2, and attI1 × attC_aadA7 cassettes in the WT or recA-deleted strain (bottom) and in the presence of the pGB2 control vector (vector) or the pGB2-recA plasmid (recA⁺), which overexpresses the RecA protein (top). Black bars represent the results obtained in the presence of IntI1, and gray bars represent the results obtained in the absence of IntI1. “nd” indicates “nondetected” recombination events. Results represent means of three independent experiments. Error bars show the standard deviations.

We also performed these assays in the absence of integrase expression (Fig. 8). As in the SSB studies, we observed recombination events only for the attI1 × attI1 combination, probably due to slippage events and/or homologous recombination (only in the recA⁺ strains) between both attI1 sites.