Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (486K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (23)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Gravel, A.
Right arrow Articles by Roy, P. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gravel, A.
Right arrow Articles by Roy, P. H.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Nucleic Acids Research Pages 4347-4355  

DNA complexes obtained with the integron integrase IntI1 at the attI1 site
Introduction
Materials And Methods
   Bacterial strains and plasmids
   DNA fragments
   Protein purification
   Gel retardation assay
   DMS methylation interference
   Potassium permanganate interference
   Quantitation
Results
   IntI1 forms several distinct complexes with the attI1 sequence
   Methylation interference in formation of attI1-IntI1 complexes
   Potassium permanganate interference in formation of attI1-IntI1 complexes
Discussion
   IntI1 binds to four regions within the attI1 site
   attI1 binding sites and site-specific recombination
   IntI1 binds weakly to attCs
Acknowledgements
References

DNA complexes obtained with the integron integrase IntI1 at the attI1 site

DNA complexes obtained with the integron integrase IntI1 at the attI1 site

Annie Gravel, Bénédicte Fournier+ and Paul H. Roy*

Centre de Recherche en Infectiologie, Centre Hospitalier de l'Université Laval and Département de Biochimie,Faculté des Sciences et de Génie, Université Laval, Sainte-Foy, Québec G1V 4G2, Canada

Received July 8, 1998; Revised and Accepted August 7, 1998

ABSTRACT

Integrons are genetic elements that are able to capture genes by a site-specific recombination mechanism. Integrons contain a gene coding for a [lambda]-like integrase that carries out site-specific recombination by interacting with two different target sites; the attI site and the palindromic sequence attC (59 base element). Cassette integrations usually involve the attI site, while cassette excisions use attC. Therefore, the integrase should bind both sites to cleave DNA and perform site-specific recombination reactions. We have used purified maltose-binding protein fused with the integrase (MBP-IntI1) and native IntI1 protein and gel retardation assays with fragments containing the complete and partial attI1 site to show formation of four complexes in this region. Chemical modification of specific nucleotides within the attI1 site was used to investigate their interference with binding of the integrase protein. We attribute IntI1 specific binding to four regions in the attI1 site and a GTTA consensus sequence is found in three of the four regions. Interference by modified guanine and thymine residues in the DNA major groove and adenine residues in the minor groove were observed, indicating that the integrase interacts with both sides of the helix. Binding of IntI1 to attC is also discussed.

INTRODUCTION

Many antibiotic resistance genes in Enterobacteriaceae and Pseudomonas are on large, transferable extrachromosomal DNA elements, called plasmids, on which other mobile elements, called transposons, may reside. Some transposons code for single resistances [e.g. TEM [beta]-lactamase (Tn3), kanamycin (Tn5) and tetracycline resistance (Tn10)], but plasmids and transposons coding multiple drug resistance often possess another genetic element, the integron. The class 1 integrons consist of three DNA regions, two conserved segments and one of variable length, and sequence that includes the inserted antibiotic resistance gene cassettes (Fig. 1A; 1). The 5[prime]-conserved segment encodes a site-specific DNA recombinase (IntI1), a member of the tyrosine recombinase family (2-6), which is involved in gene cassette excision and integration via circular intermediates (7,8). This segment also contains a promoter region and a recombination site (attI1), which has been described as the last 40-70 nt of the 5[prime]-conserved segment (9). Three classes of integrons are defined, based on the nature of their integrase genes (10), but integrons of different classes were shown to carry identical cassettes (10-12). The 3[prime]-conserved segment includes an antiseptic resistance gene (qacE[Delta]1), a sulfonamide resistance gene (sulI) and an open reading frame (orf5) of unknown function (1,13,14). The variable segment is located between the two conserved segments and each junction contains the consensus sequence GTTRRRY, which is the recombination core site (Fig. 1A). One or more genes can be found in the variable region, but each gene is usually part of an independent gene cassette, although a few cases of two genes in the same cassette have been found (15,16). An attC is present at the 3[prime]-end of a cassette and it consists of a long inverted repeat sequence of variable length and sequence (17,18). The attC can potentially form an imperfect cruciform structure, which begins with an inverted core site (RYYYAAC) and ends with a core site (GTTRRRY) (17,18). Each gene cassette has its own version of attC, although they have similar DNA sequences at their extremities (17,18). Cassettes are mobile elements, but they do not code for genes involved in their own mobility. Their movements depend upon the integrase, which interacts with the recombination sites, the attI site located in the 5[prime]-conserved segment and the attC located at the 3[prime]-end of each cassette to excise and integrate cassettes (Fig. 1B; 7,8,19). Cassette movement occurs by a conservative site-specific recombination mechanism, in which DNA-binding proteins play the primary role in bringing the partners together. Up to now, only one protein has been shown to be necessary for cassette excision and integration, the integron integrase. IntI1 is a member of a very large family that contains proteins of diverse functions (2,4-6,20).


Figure 1. (A) General structure of class 1 integrons represented by the In2 integron. The 5[prime]-conserved and 3[prime]-conserved segments are shown with their respective genes and the variable region. In In2, the variable region contains only the aadA1 gene cassette. The attI1 site is shown by a black circle, the aadA1 attC by the black rectangle and core sites by ovals. The 3[prime]-conserved segment is longer than represented in this figure, but sequences beyond orf5 differ from one integron to another (40). (B) Excision and integration of a circular gene cassette into class 1 integrons by the action of the intI1 gene product.

In this study, we report the use of gel retardation assays for the detection of IntI1-DNA complexes. We show that the interaction of IntI1 with an attI1 site results in the formation of four distinct complexes. Analysis of these complexes by interference experiments using dimethyl sulfate and potassium permanganate as DNA modifying reagents provides evidence for protein interactions with guanine and thymine residues in the major groove of B-DNA and adenines in the minor groove. We also discuss the interaction of IntI1 with an attC site.

MATERIALS AND METHODS

Bacterial strains and plasmids

The maltose-binding protein (MBP) fusion derivative of IntI1 was produced from plasmid pLQ369 in Escherichia coli TB1 [ara[Delta](lacproAB) rpsL([phis]80lacZ[Delta]M15) hsdR].

For construction of pLQ369, the intI1 gene was amplified by PCR using pLQ860 as a template (13). The primers were designed such that a NdeI restriction site was introduced at the 5[prime]-end of the coding region of intI1 (with the ATG start codon within the NdeI site) and a BamHI restriction site was introduced at the end of the gene. The amplified DNA fragment was directly cloned into pTZ/PC digested with XcmI. This new clone (pLQ368) was then digested with NdeI and the extremities were filled-in with T using the E.coli DNA polymerase I Klenow fragment. These new extremities were then digested with the mung bean enzyme to make blunt ends, digested with BamHI and finally cloned into pMAL-c2 (New England Biolabs) digested with XmnI and BamHI. The intI1 gene and its extremities were then sequenced to verify than no mutations occurred during PCR or cloning.

DNA fragments

Five DNA fragments covering the attI1 site were used (Fig. 2A). They were made by PCR using pLQ362 as the template. The primers used were: Bam840 [d(CGACGGCCGAGGTCTTC)] and aadA-Tn21 [d(TCGATGACGCCAACTAC)] for the 5[prime] long site (5[prime]L) DNA fragment; 5[prime] site [d(GGCATCCAAGCAGCAAG)] and aadA-Tn21 for the 5[prime] site (5[prime]) DNA fragment; sulpro3 [d(GCCTGACGATGCGTGGA)] and -59attIrev [d(ATCGACCCACGGCGTAA)] for the -59 DNA fragment; 5[prime] site and -18attIrev [d(GCCCTGCTGCGTAACAT)] for the -18 DNA fragment; aadA-Tn21 and -19attIfor [d(GCAGTCGCCCTAAAACA)] for the -19 DNA fragment. One oligonucleotide of each pair was 5[prime]-end-labeled with [[gamma]-32P]ATP (NEN DuPont) using T4 polynucleotide kinase (Pharmacia Biotech Inc.) and used directly in a PCR reaction with its unlabeled primer pair partner. PCR reactions were purified on a G-50 Sephadex column and counted in an automatic counter. Unlabeled PCR control reactions were loaded on an agarose gel for quantification.


Figure 2. (A) DNA fragments, covering the attI1 site, used in gel retardation experiments. PCR primers and the extent of DNA fragments are indicated. Numbers below the DNA show the position of the 5[prime]-end of PCR primers (represented by arrowheads), relative to the G residue of the attI1 recombination site. intI1 translation starts at nt -136 on the complementary strand; the attI1 site is defined as nt -63 to +6; the recombination reaction takes place between G 0 and the T +1 of the core site. (B) SDS-PAGE analysis showing the IntI1 purification steps. Lane 1, molecular weight markers; lane 2, soluble crude extract from E.coli TB1 overexpressing pLQ369; lane 3, amylose column-purified MBP-IntI1; lane 4, MBP-IntI1 cleaved with factor Xa at 4°C for 18 h; lane 5, SP-Sepharose-purified IntI1.

Protein purification

MBP-IntI1 fusion protein was essentially purified as suggested by New England Biolabs. Protein concentration of the purified fusion protein was determined using the Bradford protein assay (BioRad). The protein solution was then made 20% with respect to glycerol and stored at -80°C. The purity of the MBP-IntI1 was evaluated at >90% on a SDS-PAGE gel (Fig. 2B).

The MBP-IntI1 fusion protein was cleaved by factor Xa at 4°C for 18 h and dialyzed twice against buffer A (50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1 mM EDTA, pH 8.0). The protein mix was loaded onto a SP-Sepharose High Performance column equilibrated with buffer A and run on a FPLC system (Pharmacia Biotech Inc.). IntI1 was eluted from the column with a NaCl gradient (0.2-1 M), concentrated, dialyzed against buffer A, quantified and stored at -80°C in 50% glycerol. The purity of IntI1 was evaluated at >95% on a SDS-PAGE gel (Fig. 2B).

Gel retardation assay

Binding reactions containing 20 000 c.p.m. (0.1-0.8 pmol) labeled DNA fragment were incubated with different concentrations of MBP-IntI1 in a 10 µl volume containing 10 mM HEPES (K+, pH 8.0), 60 mM KCl, 4 mM MgCl2, 100 µM EDTA (pH 8.0), 100 µg/ml BSA, 250 µM DTT, 100 ng poly(dI·dC) and 10% glycerol. Reactions were incubated at room temperature for 15 min, prior to electrophoresis through 4 or 5% pre-run, non-denaturing polyacrylamide gels buffered with 0.5× TBE. Dried gels were subjected to autoradiography. The poly(dI·dC) was only used with fusion protein; water was used with purified integrase protein. Unlabeled DNA was also added in competition experiments at concentrations 20-30 times that of labeled DNA, while 1 µl diluted MBP antibody (1/5) (New England Biolabs) was added in super-shift experiments.

DMS methylation interference

Labeled DNA (3 pmol) and 1 µl poly(dI·dC) (1 mg/ml) were dissolved in 200 µl 50 mM sodium cacodylate (pH 8.0) and 1 mM EDTA. DMS (1 µl) was added and the DNA modification reaction proceeded for 5 min at room temperature before termination by centrifugation of the reaction through Sephadex G-50 and precipitation. Under these conditions most molecules had undergone no more than one methylation. DNA-protein binding and a gel retardation assay were then done with the mix of methylated DNA fragments as described above. Electrophoresis through 4 or 5% polyacrylamide gels separated bound and unbound DNA, with protein-DNA complexes detected by autoradiography. The labeled DNA was eluted from gel slices, cleaved at modified positions using piperidine (1 M) at 90°C for 30 min, lyophilized, washed twice in water, counted, resuspended in loading dye and analyzed by electrophoresis through 6 or 15% sequencing gels.

Potassium permanganate interference

Labeled DNA (3 pmol) was dissolved in 55 µl 30 mM Tris-HCl (pH 8.0) followed by denaturation at 100°C for 3 min. KMnO4 (60 µl, 250 µM) was added and the mixture incubated for 12.5 min at room temperature, modifying and renaturing the DNA before purification on a Sephadex G-50 column and precipitation. Isolation and analysis of protein-DNA complexes and free DNA were as described above. Strand cleavage was performed in 1 M piperidine at 90°C for 30 min and DNA was analyzed as above.

Quantitation

Autoradiograms were analyzed by BioImage Visage 110s from Millipore (Ann Arbor, MI) with Whole Band Analysis software.

RESULTS

IntI1 forms several distinct complexes with the attI1 sequence

We examined binding of MBP-IntI1 fusion protein by gel retardation assays using five different fragments covering the attI1 site (Fig. 2A). A long fragment (5[prime]L) of 537 bp was used to detect any binding of the fusion protein to the complete attI1 site. We observed four DNA-protein complexes with the MBP-IntI1 fusion protein, while no complexes were obtained with the MBP protein alone (Fig. 3, lanes 1-5), indicating that the DNA-protein complexes obtained with the fusion protein were IntI1-specific. At lower concentrations of MBP-IntI1 protein (30-125 nM), the amount of the fastest migrating complex was increased and no complex IV was observed; at a high concentration (625 nM), the majority of bound DNA was in the form of complex IV andno unbound DNA was observed (data not shown). These observations indicate the dependence of complex formation on IntI1 concentration. In some experiments, a fifth complex was observed, however, we were not able to see this complex in all experiments performed (Fig. 4, lanes 2 and 3). All complexes were sensitive to competition by unlabeled 5[prime]L DNA fragment and could be converted to lower mobility complexes when an MBP antibody was used, indicating that they result from specific DNA-protein interactions (Fig. 4, lanes 4 and 5).


Figure 3. Gel retardation assays with the 5[prime]L DNA fragment and MBP-IntI1 (lanes 1-5) or native IntI1 (lanes 6-8). The aadA-Tn21 PCR primer was 5[prime]-end-labeled, a PCR amplification was done with its primer partner (Bam840) and the purified labeled fragments were incubated with different protein concentrations. Free DNA (F) and protein-DNA complexes (I, II, III and IV) are indicated by arrows. Proteins and concentrations used are indicated in tables above the corresponding figures. Lanes 1-5 and 6-8 were on separate gels.


Figure 4. Binding of MBP-IntI1 fusion protein to four DNA fragments covering the attI1 site. Gel retardation assays were done as described in Materials and Methods. A competition experiment was done with the unlabeled fragment used in the assay and a super-shift assay was done using a polyclonal MBP antibody. Protein concentrations, ratios of unlabeled DNA fragments to labeled fragments and MBP antibody used in this assay are indicated in the table above the figure, while DNA fragments used are indicated under the figure. Greater quantities of competitor DNA were required to completely eliminate the fastest migrating complexes (data not shown).

We then made a smaller fragment that still contains the complete attI1 recombination site; the 5[prime] site (5[prime]) DNA fragment (Fig. 2A). We used this fragment to do gel retardation assays with MBP-IntI1 fusion protein. Figure 4 shows the formation of four IntI1-DNA complexes with this fragment; all complexes could be competed by the unlabeled DNA fragment and super-shifted using the MBP antibody (Fig. 4, lanes 6-10). These results, along with those obtained using the 5[prime]L fragment, indicate that all binding regions of IntI1 are located between nt -96 and +71.

We then subdivided the DNA 5[prime]-fragment into two fragments to determine more precisely which regions were involved in binding of MBP-IntI1. We used DNA fragments -18 and -19 (Fig. 2A) to do gel retardation assays, the results of which are shown in Figure 4 (lanes 11-20). We observed two DNA-protein complexes with the -18 DNA fragment (complexes III and IV) and two complexes with the -19 fragment (complexes I and II). These results agree with those obtained with the 5[prime]-fragment and indicate that two complexes are formed upstream of nt -20 and two complexes are formed downstream of this region. These complexes can also be competed with unlabeled DNA fragment and super-shifted with the MBP antibody (Fig. 4).

We also performed gel retardation assays with native IntI1 using the 5[prime]L, -18 and -19 DNA fragments. Results were similar to those obtained with the MBP-IntI1 fusion protein but a greater quantity of purified integrase was required to obtain the slowest mobility complexes (Fig. 3, lanes 6-8).

We used another DNA fragment, -59 (Fig. 2A), to investigate binding of the integrase further upstream from the attI1 site. We obtained one specific complex with MBP-IntI1 that could be competed with unlabeled DNA fragment and super-shifted, however, we were unable to obtain significant interference at residues in this region (data not shown). We were also unable to obtain any complexes with the -59 DNA fragment using native IntI1 (data not shown).

DNase I footprinting analysis with the bottom strand of the DNA 5[prime]-fragment previously found three regions protected by IntI1 from DNase I digestion (data not shown). One region was located around nt -50, another around nt -30 and a larger one around the core site. The large region around the core site covered ~30 nt, from T -17 to A +11, and we at first believed that only one molecule of IntI1 was binding to the entire region. However, residues A -3, A -2 and A -1 of the top strand (and their complementary thymine residues) did not show any interference in formation of either complex I or II (see below), indicating that these residues are not involved in binding of IntI1. The DNase I footprinting experiment was not sensitive enough to distinguish the two binding sites for the integrase in the core site region and that is why interference experiments were performed.

Methylation interference in formation of attI1-IntI1 complexes

Of the several methods that are available to probe sequence-specific recognition of DNA by proteins, methylation interference has proved to be a powerful technique and has become one of the most widely used (21,22). DMS methylates double-stranded DNA at the N7 position of guanine in the major groove and the N3 position of adenine in the minor groove (23). A methylation interference experiment determines whether methylation of specific bases interferes with protein function (24). To determine which methylated residues within the attI1 site interfere with formation of complexes by inhibiting binding of IntI1, 5[prime]-end-labeled -19 and -18 DNA fragments were modified with DMS prior to incubation with MBP-IntI1 or native IntI1 and a G>A specific cleavage of methylated bases was performed to analyze DNA from isolated complexes. To identify residues which interfere with formation of complexes, we quantified interference of each nucleotide, as described in the legend to Figure 5. Where interference was found, the value for the intensity of a band from the substrate divided by the value for the same band from complexes was >1. We considered the interference to be significant when this ratio exceeded 1.5, while the ratios for unrelated residues were <1.3; a ratio between 1.3 and 1.5 is considered weak interference. The interference must, however, be significant as compared with free DNA (F).

   A

   B, C

Figure 5. Methylation interference of both strands of the -19 DNA fragment, identifying guanine major groove contacts and adenine minor groove contacts. (A) The modified 5[prime]-end-labeled DNA (M) was incubated with purified IntI1 and separated into bound (B) and unbound (F) DNA. Isolated DNA from each part was then cleaved and analyzed on denaturing polyacrylamide gels. BI and BII represent the fastest and slowest migrating complexes obtained with this fragment (see Fig. 4). The binding pattern, next to a Maxam-Gilbert A+G sequencing reaction, shows the sequence positions that interfere with protein binding. Numbers next to the autoradiograms represent the nucleotides as numbered in Figure 2, and strong ([solid triangle]) and weak ([open triangle]) interferences are shown. (B) Quantitation of interference data. Background values were initially subtracted from each band quantified. Subsequently, all bands to be compared in the modified (M), free (F) and complexed (B) lanes were normalized to correct for differences in the amount of sample loaded in each lane. This was done by quantifying the same unrelated band in the four lanes, dividing the larger of the four values by the other ones to give a correction factor and then multiplying quantitated intensities of each band in the lane by the correction factor. The fold interference (× Interference) was then calculated for each nucleotide, as the quantitated value for the modified band divided by the value for the other ones. Solid bars indicate the fold interference in the free DNA; dark hatched bars, in complex I; light bars, in complex II. (C) Sequence of the fragment, showing the regions that interfere with binding of IntI1.

The results of methylation interference experiments with the -19 DNA fragment are shown in Figure 5. In the top strand, we observed a reduction in the band intensities corresponding to the G 0, A +3 and A +4 residues in complexes I and II (Fig. 5A, lanes BI and BII, and 5B) and an enhancement of the same bands in DNA from the unbound fraction (F), as compared with untreated substrate DNA (M). These observations indicate interference of these three core site residues in formation of both complexes and thus that one molecule of IntI1 binds in the core site region of attI1 in both complexes. Residues A -8, A -7, A -6 and A -5 showed interference only in formation of complex II, indicating that one molecule of IntI1 binds in region I in both complexes and another molecule binds in region II in complex II (Fig. 5C). On the bottom strand, residues G +6 and G +9, as well as the two adenines on the complementary strand of the core site (A +1 and A +2) (Fig. 5A, lanes BI and BII, and 5B), showed significant interference in formation of both complexes. The latter two residues were quantified together because they were too close and too weak to be quantified separately. Interference of the A +8 residue was not determined due to non-significant quantitative data. The G -12, G -11, G -10, A -9 (weak interference) and G -4 residues showed significant interference only in formation of complex II, again indicating binding of one IntI1 molecule in region I and one molecule in region II (Fig. 5C). Results obtained for the top and bottom strands of the -19 DNA fragment indicate methylation interference of the same regions of the attI1 site (I and II) for each strand.

We then performed methylation interference experiments with the -18 DNA fragment (Fig. 6). Analysis of the top strand indicated that the G -32, A -29, G -27, A -25 and G -24 residues interfered with formation of complex III and the first two of these in complex IV (Fig. 6A, lanes BIII and BIV, and 6B). We do not know why the G -27, A -25 and G -24 residues do not show significant interference data in the formation of complex IV, but interference data for these residues in formation of complex III are highly significant and, thus, these residues should interfere with formation of both complexes. These observations indicate that one molecule of IntI1 binds in region III in both complexes. In complex IV, we observed interference of residues G -51 and A -58 in addition to the G -32 and A -29 residues, thus, this complex is formed by binding of one molecule of IntI1 in region III (as in complex III) and another molecule of IntI1 in region IV (Fig. 6C). Analysis of the bottom strand revealed a significant interference of residues A -33, A -31, A -30, G -28 (weak interference) and G -26 in formation of complexes III and IV (Fig. 6A, lanes BIII and BIV, and 6B); these residues are located in region III, the same region found to interfere with formation of complex III on the top strand. Residues A -50, A -49 and A -47 interfere only in formation of complex IV, showing again that two molecules bind to this fragment, one in region III and another one in region IV (Fig. 6C). Quantitative analysis of the residue A -52 indicated a ratio not significant enough to prove interference. We, however, think that this residue may indeed interfere with IntI1 binding, since the corresponding T on the top strand does show interference (see below). Results obtained for the top and the bottom strands of the -18 DNA fragment indicate methylation interference of the same regions of the attI1 site (III and IV) for each strand.

   A

   B, C

Figure 6. Methylation interference of both strands of the -18 DNA fragment, identifying guanine major groove contacts and adenine minor groove contacts. (A) The modified 5[prime]-end-labeled DNA (M) was incubated with purified IntI1 and separated into bound (B) and unbound (F) DNA. Isolated DNA from each part was then cleaved and analyzed as described in Figure 5. BIII and BIV represent the fastest and slowest migrating complexes obtained with this fragment (see Fig. 4). (B) Quantitation of interference data. Data were normalized as described in Figure 5. Solid bars indicate the fold interference in the free DNA; dark hatched bars, in complex III; light bars, in complex IV. (C) Sequence of the fragment, showing the regions that interfere with binding of IntI1.

We performed all experiments with both MBP-IntI1 fusion protein and native IntI1. The results were the same for both proteins, confirming that the complexes obtained with the fusion protein are related to the integrase protein and not to MBP dimerization.

Potassium permanganate interference in formation of attI1-IntI1 complexes

In general, methylation of G residues interfered with recombinase binding more strongly than A methylation because piperidine-mediated cleavage of methylated guanine is more efficient than cleavage of methylated adenine (23). The attI1 site contains several adenines whose contacts with IntI1 might be important. We determined the interference of several adenine residues with the binding of IntI1 in all regions previously found by methylation interference. However, a few adenines show no interference at all; for others, quantitative data were too weak to determine interference. We thus looked for interference of thymine residues using potassium permanganate. The non-polar methyl group of thymine at position C -5 is important for discrimination between thymine and cytosine and provides a surface for van der Waal's contacts with amino acids (25,26). Potassium permanganate primarily modifies thymine in single-stranded DNA by oxidizing the 5,6-double bonds to produce the glycol form (27). Thymine glycols have been shown to prevent protein binding to specific DNA sequences (27,28).

Analysis of the top strand of the -19 fragment (Fig. 7) showed strong interference of three modified thymine residues, located at T +1, T +2 and T +8, in formation of both complexes (Fig. 7A, lanes BI and BII, and 7B), indicating their participation in binding of one molecule of IntI1. The methylation interference of residue A +8 was not determined, but from these observations it probably interferes in the binding of IntI1. Another residue, T -9, showed interference only with formation of complex II, indicating its involvement in binding of a second molecule of IntI1. Analysis of the bottom strand revealed the same organization. Three thymine residues, located at positions T +4, T +7 and T +10, showed interference in formation of both complexes (BI and BII). Four residues interfere only with formation of complex II; their positions were T -8, T -7, T -6 and T -5 (Fig. 7A, lanes BI and BII, and 7B). Analysis of these data indicates that one molecule of IntI1 binds in region I and another molecule binds in region II, the same regions found with the methylation interference experiments (Fig. 7C). We found a base pair that does not show the same interference on both strands for this fragment; the T -9 residue located on the top strand, which interferes strongly in the formation of complex II, and its complementary A, which showed weak interference. We think that this difference results from G>A specific cleavage. One residue on the top strand (A +3) showed methylation interference but no potassium permanganate interference was observed. Two residues on the bottom strand (T +7 and T +10) showed potassium permanganate interference but no methylation interference of their complementary adenines was observed. These observations may reflect the flexibility of integrase binding in region I (see Discussion).

   A

   B, C

Figure 7. Potassium permanganate interference on both strands of the -19 DNA fragment, identifying thymine major groove contact points important for IntI1 binding. (A) The modified 5[prime]-end-labeled DNA (M) was incubated with purified IntI1 and separated into bound (B) and unbound (F) DNA. Isolated DNA from each part was then cleaved and analyzed as described in Figure 5. BI and BII represent the fastest and slowest migrating complexes obtained with this fragment (see Fig. 4). (B) Quantitation of interference data. Data were normalized as described in Figure 5. Solid bars indicate the fold interference in the free DNA; dark hatched bars, in complex I; light bars, in complex II. (C) Sequence of the fragment, showing the regions that interfere with binding of IntI1.

The top strand of fragment -18 showed seven modified thymine residues interfering strongly with binding of IntI1. The thymine residues located at T -33, T -31 and T -30 interfere with formation of both complexes (Fig. 8, lanes BIII and BIV, and 8B), while residues T -52, T -50, T -49 and T -47 interfere only in formation of complex IV. Three modified thymine residues, located at positions T -48, T -29 and T -25 on the bottom strand, showed interference in binding of IntI1; T -29 and T -25 interfere strongly in formation of both complexes while T -48 interferes weakly in formation of complex IV (Fig. 8A, lanes BIII and BIV, and 8B). These data indicate the involvement of region III in binding of one molecule of IntI1 and the involvement of region IV in binding of another molecule; these observations support the methylation interference data for these regions (Fig. 8C). In addition to the interferences previously found, we found a strong interference of the T -52 residue of the top strand in formation of complex IV, while interference was not demonstrated for the complementary A residue on the bottom strand, due to non-significant quantitative data. We believe that the A -52 residue does interfere in formation of complex IV, but G>A specific cleavage of methylated bases does not permit us to see this interference. Residue T -48 on the bottom strand and its complementary adenine showed weak and strong interference respectively, which could be explained by the same phenomenon.

DISCUSSION

IntI1 binds to four regions within the attI1 site

In gel retardation experiments, we obtained four DNA-protein complexes within the attI1 site (Figs 3 and 4). Changes in the mobility of protein-DNA complexes in non-denaturing gels can be attributed to alterations in the conformation of the DNA as well as to the change in mass of the DNA that results from binding of increasing amounts of proteins (29,30). Although our studies do not rule out IntI1-induced conformational changes in the attI1 site, it is likely that the four complexes are the result of binding of one to four molecules of IntI1 protein to the attI1 site, because each of the four complexes exhibits a unique pattern of interference (Figs 5-8). The interference data for the -19 DNA fragment are particularly important because they indicate the presence of two closely spaced binding sites for the integrase in the core site region that were not determined with the DNase I footprinting experiment.

We observed that three of the four binding sites showing strong interference in formation of complexes (Figs 5-8) contain a GTTA consensus sequence. This sequence shows similarities with the core site (GTTRRRY), but the complete core site is not repeated in all integrase binding sites, as nucleotides following the A residue are not conserved. This suggests that the GTTA residues are the most important in sequence recognition by the integrase protein. A few integrons carry the sequence GTTG instead of the GTTA usually found at the crossover site, however, they have been shown to be just as active as other integron attI1 sites in site-specific recombination (31). We can then assume that IntI1 binding would be observed at this site and, consequently, a consensus sequence of GTTR would be a better definition. Our interference results have shown three residues that clearly interfere in only one strand of the DNA (T +3, T +7 and T +10 on the bottom strand). All these residues are located in region I, in the variable region found in integrons, where different cassettes are inserted in the different integrons. These observations could reflect the flexibility of IntI1, with its weak binding in this region, to be able to recognize all integron sequences. The other IntI1 binding site (region II) is located upstream of the core site and together, regions II and I have an arrangement similar to the XerC/XerD recognition pattern (tAA-N6-8-TTA). The motif of IntI1 binding sites for regions II and I can be defined as TAA-N7-TTR. In XerC/XerD site-specific recombination, XerC binds to the upstream sequence of the N6-8 and XerD to the downstream sequence and DNA cleavage occurs at each extremity of this spacer region (N6-8), immediately 3[prime] of an AA dinucleotide on each strand (28,32). It is known that in the XerC/XerD system binding of XerC is weaker than that of XerD and that cooperative binding occurs (33,34). In the integron site-specific recombination system, we do not know whether cooperative binding occurs and, consequently, whether binding to the second site is weaker than to the first. However, our results indicate that binding of IntI1 to regions I and II is not necessary for binding to regions III and IV, and vice versa. Because the integron has a potential XerC/XerD recognition pattern and we found that one molecule of IntI1 binds on each side of the potential spacer region, we can speculate that IntI1 would cleave DNA at the extremities of the 7 bp spacer region and proceed like other members of the family (35-37). However, in vivo evidence accumulated to date shows that the crossover occurs at a unique position in the core site (18,31). These findings lead us to believe that either strand exchange occurs at the same site on both strands, or DNA synthesis occurs on one strand, or again only one strand exchange occurs in integrons, with the Holliday junctions being resolved by RuvC (38); other explanations are also possible. Therefore, in vitro experiments need to be done to determine the mechanism of strand cleavage and exchange in integrons.

Another group has recently studied the interaction between IntI1 and the attI1 site (39). They observed formation of two complexes using a gel retardation experiment with a DNA fragment covering the complete attI1 site and identified only one binding site by DNase I footprinting and methylation interference. This binding site corresponds to region III identified in this study, with a few differences in the modified residues that interfere with binding of IntI1. Based on an imperfect duplication of binding site III found at -50 from the core site, they concluded that the second complex obtained in the gel retardation experiment could represent binding of IntI1 to this site, although they did not obtain any interference data for this complex. This site corresponds to region IV identified in this study, in which we have shown modified residues that interfere with binding of IntI1. They did not observe the formation of any complexes that could represent binding of IntI1 around the core site region and thus did not identify any binding sites corresponding to regions I and II of this study.

   A

   B, C

Figure 8. Potassium permanganate interference on both strands of the -18 DNA fragment, identifying thymine major groove contact points important for IntI1 binding. (A) The modified 5[prime]-end-labeled DNA (M) was incubated with purified IntI1 and separated into bound (B) and unbound (F) DNA. Isolated DNA from each part was then cleaved and analyzed as described in Figure 5. BIII and BIV represent the fastest and slowest migrating complexes obtained with this fragment (see Fig. 4). (B) Quantitation of interference data. Data were normalized as described in Figure 5. Solid bars indicate the fold interference in the free DNA; dark hatched bars, in complex III; light bars, in complex IV. (C) Sequence of the fragment, showing the regions that interfere with binding of IntI1.


This study identified interactions between IntI1 protein and the attI1 site. We examined major groove contacts with guanine and thymine residues using DMS and potassium permanganate, respectively. Minor groove contacts were investigated with methylated adenines. In the presence of IntI1 protein, interference by the methylated guanines and adenines was noted in each of the four binding sites. When these residues are positioned on the helical DNA, major and minor groove contacts of IntI1 with both faces of the helix are observed and the integrase protein seems to wrap around the DNA (Fig. 9). For example, the minor groove contact at position A +3 is accessible from the front of the helix, while the major groove contact at position G 0 is from the back (Fig. 9). However, across all binding sites, IntI1 seems to contact DNA more often on the same side of the helix. The binding sites are also located two helical turns from each other in regions II, III and IV.


Figure 9. B-form of the DNA helix showing the attI1 sequence from nt -55 to +13. The top strand sequence is shown. Strong and weak interferences of modified nucleotides with IntI1 binding are denoted as [solid square] and [open square] for guanines, as [solid circle] and [open circle] for adenines and as [solid triangle] and [open triangle] for thymines. The IntI1 binding sites are indicated as regions IV, III, II and I.

We performed gel retardation assays with the -59 DNA fragment shown in Figure 2A. We were able to obtain one specific complex with the fusion protein, although not with native IntI1 (data not shown). Methylation interference experiments on the complex obtained with MBP-IntI1 showed a non-significant reduction in band intensity at the G -76 position. This may explain the fifth complex, however, we believe that this site is not a real binding site for IntI1 and is not involved in the recombination reaction, as it has been shown recently that sequences beyond nt -73 do not negatively affect recombination by IntI1 (31).

attI1 binding sites and site-specific recombination

We found four binding sites, but a question arises as to whether binding of IntI1 to all four sites is necessary for recombination to occur. We have not yet examined the effect of these binding sites on the recombination reaction. However, Hansson et al. have studied recombination frequency with deleted and mutated attI1 sites (31). Their data show that deletions between nt -36 and -27 have a strong negative effect on recombination between an attI and an attC but a marginal effect on an attI-attI recombination reaction. Further deletion into the -27 to -14 region has no additional effects. Mutations of the core site and deletions to within 7-14 nt upstream of the core site have an additional strong negative effect on recombination. Their deletion data for nt -20 to +10 correlate well with our data on protein binding sites located in this region. Our data also indicate the presence of a binding site in region III that was suggested by their deletion data in attI-attC recombination experiments. We can speculate that this binding site is involved in attI-attC site-specific recombination by some DNA bindings that bring the recombination partners into much closer contact. Our data suggest another binding site in region IV, but their data do not include deletion end points in this region.

IntI1 binds weakly to attCs

We also looked for possible binding sites present in the attC of In2. We were not able to determine the number of binding sites present in this element because of weaker binding of MBP-IntI1 to a DNA fragment covering the complete attC than to the attI1 fragments. We thought that the possible hairpin structure of attC could interfere with integrase binding. We then designed PCR primers to make two smaller DNA fragments each containing half of the attC site. Gel retardation assays have shown the formation of two weak complexes with the left half-site (the fragment containing the inverted core site); data for the right half-site were inconclusive (data not shown). It appears that two IntI1 molecules bind in the inverted core site region while several molecules bind the rest of attC and the 3[prime]-conserved segment. Recently, Stokes et al. postulated, by sequence comparison, four putative IntI1 binding domains in attC, two at each end (18). We can speculate that the two complexes obtained with the fragment covering the 5[prime]-end half of attC are a consequence of the binding of one and two molecules of IntI1 at these sites. We can also speculate that two other molecules of IntI1 bind to the 2R and 1R sites (18), but we believe that additional IntI1 molecules also bind in the 3[prime]-conserved segment.

ACKNOWLEDGEMENTS

We thank Steve Leclerc for his advice on DMS methylation interference, Ronald Maheux for the quantitation analysis with BioImage and Daniel C.Tessier for the kind gift of pTZ/PC. This work was supported by grant MT-13564 from the Medical Research Council (MRC) of Canada to P.H.R. A.G. held a fellowship from the MRC Canada and B.F. was supported in part by a fellowship from the Ministère de la Recherche et de l'Espace of France.

REFERENCES

1. Stokes,H.W. and Hall,R.M. (1989) Mol. Microbiol., 3, 1669-1683. MEDLINE Abstract

2. Argos,P., Landy,A., Abremski,K., Egan,J.B., Haggård-Ljungquist,E., Hoess,R.H., Kahn,M.L., Kalinoes,B., Narayana,S.V.L., Pierson III,L.S., Sternberg,N. and Leong,J.M. (1986) EMBO J., 5, 433-440. MEDLINE Abstract

3. Ouellette,M. and Roy,P.H. (1987) Nucleic Acids Res., 15, 10055. MEDLINE Abstract

4. Abremski,K.E. and Hoess,R.H. (1992) Protein Engng, 5, 87-91.

5. Esposito,D. and Scocca,J.J. (1997) Nucleic Acids Res., 25, 3605-3614. MEDLINE Abstract

6. Nunes-Düby,S.E., Kwon,H.J., Tirumalai,R.S., Ellenberger,T. and Landy,A. (1998) Nucleic Acids Res., 26, 391-406. MEDLINE Abstract

7. Collis,C.M. and Hall,R.M. (1992) Mol. Microbiol., 6, 2875-2885. MEDLINE Abstract

8. Collis,C.M. and Hall,R.M. (1992) J. Bacteriol., 174, 1574-1585. MEDLINE Abstract

9. Recchia,G.D., Stokes,H.W. and Hall,R.M. (1994) Nucleic Acids Res., 22, 2071-2078. MEDLINE Abstract

10. Recchia,G.D. and Hall,R.M. (1995) Microbiology, 141, 3015-3027. MEDLINE Abstract

11. Sundström,L., Roy,P.H. and Sköld,O. (1991) J. Bacteriol., 173, 3025-3028. MEDLINE Abstract

12. Arakawa,Y., Murakami,M., Suzuki,K., Ito,H., Wacharotayankun,R., Ohsuka,S., Kato,N. and Ohta,M. (1995) Antimicrobial Agents Chemother., 39, 1612-1615.

13. Bissonnette,L. and Roy,P.H. (1992) J. Bacteriol., 174, 1248-1257. MEDLINE Abstract

14. Paulsen,I.T., Littlejohn,T.G., Rådström,P., Sundström,L., Sköld,O., Swedberg,G. and Skurray,R.A. (1993) Antimicrobial Agents Chemother., 37, 761-768.

15. Tolmasky,M.E. (1990) Plasmid, 24, 218-226. MEDLINE Abstract

16. Bunny,K.L., Hall,R.M. and Stokes,H.W. (1995) Antimicrobial Agents Chemother., 39, 686-693.

17. Hall,R.M., Brookes,D.E. and Stokes,H.W. (1991) Mol. Microbiol., 5, 1941-1959. MEDLINE Abstract

18. Stokes,H.W., O'Gorman,D.B., Recchia,G.D., Parsekhian,M. and Hall,R.M. (1997) Mol. Microbiol., 26, 731-745. MEDLINE Abstract

19. Collis,C.M., Grammaticopoulos,G., Briton,J., Stokes,H.W. and Hall,R.M. (1993) Mol. Microbiol., 9, 41-52. MEDLINE Abstract

20. Blakely,G.W. and Sherratt,D.J. (1996) Mol. Microbiol., 20, 234-237. MEDLINE Abstract

21. Siebenlist,U., Simpson,R.B. and Gilbert,W. (1980) Cell, 20, 269-281. MEDLINE Abstract

22. Siebenlist,U. and Gilbert,W. (1980) Proc. Natl Acad. Sci. USA, 77, 122-126. MEDLINE Abstract

23. Wissmann,A. and Hillen,W. (1991) Methods Enzymol., 208, 365-379. MEDLINE Abstract

24. Luetke,K.H., Zhao,B.P. and Sadowski,P.D. (1997) Nucleic Acids Res., 25, 4240-4249. MEDLINE Abstract

25. Seeman,N.C., Rosenberg,J.M. and Rich,A. (1976) Proc. Natl Acad. Sci. USA, 73, 804-808. MEDLINE Abstract

26. Pabo,C.O. and Sauer,R.T. (1992) Annu. Rev. Biochem., 61, 1053-1095. MEDLINE Abstract

27. Panigrahi,G.B., Beatty,L.G. and Sadowski,P.D. (1992) Nucleic Acids Res., 20, 5927-5935. MEDLINE Abstract

28. Blakely,G.W. and Sherratt,D.J. (1994) Nucleic Acids Res., 22, 5613-5620. MEDLINE Abstract

29. Wu,H.-M. and Crothers,D.M. (1984) Nature, 308, 509-513. MEDLINE Abstract

30. Andrews,B.J., Beatty,L.G. and Sadowski,P.D. (1987) J. Mol. Biol., 193, 345-358. MEDLINE Abstract

31. Hansson,K., Sköld,O. and Sundström,L. (1997) Mol. Microbiol., 26, 441-453. MEDLINE Abstract

32. Blakely,G., Colloms,S., May,G., Burke,M. and Sherratt,D. (1991) New Biologist, 3, 789-798. MEDLINE Abstract

33. Blakely,G., May,G., McCulloch,R., Arciszewska,L.K., Burke,M., Lovett,S.T. and Sherratt,D.J. (1993) Cell, 75, 351-361. MEDLINE Abstract

34. Blakely,G. and Sherratt,D. (1996) Genes Dev., 10, 762-773. MEDLINE Abstract

35. Stark,W.M., Boocock,M.R. and Sherratt,D.J. (1992) Trends Genet., 8, 432-439. MEDLINE Abstract

36. Landy,A. (1993) Curr. Opin. Genet. Dev., 3, 699-707. MEDLINE Abstract

37. Jayaram,M. and Lee,J. (1995) Trends Genet., 11, 432-433. MEDLINE Abstract

38. Connolly,B., Parsons,C.A., Benson,F.E., Dunderdale,H.J., Sharples,G.J., Lloyd,R.G. and West,S.C. (1991) Proc. Natl Acad. Sci. USA, 88, 6063-6067. MEDLINE Abstract

39. Collis,C.M., Kim,M.-J., Stokes,H.W. and Hall,R.M. (1998) Mol. Microbiol., 29, 477-490. MEDLINE Abstract

40. Brown,H.J., Stokes,H.W. and Hall,R.M. (1996) J. Bacteriol., 178, 4429-4437. MEDLINE Abstract

*To whom correspondence should be addressed at: Centre de Recherche en Infectiologie, RC-709, CHUL, 2705 Boulevard Laurier, Sainte-Foy,Québec G1V 4G2, Canada. Tel: +1 418 654 2705; Fax: +1 418 654 2715; Email: paul.h.roy@crchul.ulaval.ca
+Present address: Infectious Disease Unit, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114-2696, USA

This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 19 Sep 1998
Copyright©Oxford University Press, 1998.
Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J Antimicrob ChemotherHome page
M. S. Ramirez, T. R. Parenteau, D. Centron, and M. E. Tolmasky
Functional characterization of Tn1331 gene cassettes
J. Antimicrob. Chemother., October 1, 2008; 62(4): 669 - 673.
[Abstract] [Full Text] [PDF]

Home page
 Nucleic Acids ResHome page
G. Demarre, C. Frumerie, D. N. Gopaul, and D. Mazel
Identification of key structural determinants of the IntI1 integron integrase that influence attC x attI1 recombination efficiency
Nucleic Acids Res., October 8, 2007; 35(19): 6475 - 6489.
[Abstract] [Full Text] [PDF]

Home page
 Nucleic Acids ResHome page
C. Johansson, M. Kamali-Moghaddam, and L. Sundstrom
Integron integrase binds to bulged hairpin DNA
Nucleic Acids Res., August 2, 2004; 32(13): 4033 - 4043.
[Abstract] [Full Text] [PDF]

Home page
 MicrobiologyHome page
C. M. Collis and R. M. Hall
Comparison of the structure-activity relationships of the integron-associated recombination sites attI3 and attI1 reveals common features
Microbiology, May 1, 2004; 150(5): 1591 - 1601.
[Abstract] [Full Text] [PDF]

Home page
 J Antimicrob ChemotherHome page
A. Reyes, H. Bello, M. Dominguez, S. Mella, R. Zemelman, and G. Gonzalez
Prevalence and types of class 1 integrons in aminoglycoside-resistant Enterobacteriaceae from several Chilean hospitals
J. Antimicrob. Chemother., February 1, 2003; 51(2): 317 - 321.
[Abstract] [Full Text] [PDF]

Home page
 Antimicrob. Agents Chemother.Home page
S. R. Partridge, C. M. Collis, and R. M. Hall
Class 1 Integron Containing a New Gene Cassette, aadA10, Associated with Tn1404 from R151
Antimicrob. Agents Chemother., August 1, 2002; 46(8): 2400 - 2408.
[Abstract] [Full Text] [PDF]

Home page
 J. Bacteriol.Home page
C. M. Collis, M.-J. Kim, S. R. Partridge, H. W. Stokes, and R. M. Hall
Characterization of the Class 3 Integron and the Site-Specific Recombination System It Determines
J. Bacteriol., June 1, 2002; 184(11): 3017 - 3026.
[Abstract] [Full Text] [PDF]

Home page
 J. Bacteriol.Home page
N. Messier and P. H. Roy
Integron Integrases Possess a Unique Additional Domain Necessary for Activity
J. Bacteriol., November 15, 2001; 183(22): 6699 - 6706.
[Abstract] [Full Text] [PDF]

Home page
 J. Bacteriol.Home page
C. M. Collis, G. D. Recchia, M.-J. Kim, H. W. Stokes, and R. M. Hall
Efficiency of Recombination Reactions Catalyzed by Class 1 Integron Integrase IntI1
J. Bacteriol., April 15, 2001; 183(8): 2535 - 2542.
[Abstract] [Full Text]

Home page
 Antimicrob. Agents Chemother.Home page
C. Goldstein, M. D. Lee, S. Sanchez, C. Hudson, B. Phillips, B. Register, M. Grady, C. Liebert, A. O. Summers, D. G. White, et al.
Incidence of Class 1 and 2 Integrases in Clinical and Commensal Bacteria from Livestock, Companion Animals, and Exotics
Antimicrob. Agents Chemother., March 1, 2001; 45(3): 723 - 726.
[Abstract] [Full Text]

Home page
 MicrobiologyHome page
S. R. Partridge, G. D. Recchia, C. Scaramuzzi, C. M. Collis, H. W. Stokes, and R. M. Hall
Definition of the attI1 site of class 1 integrons
Microbiology, November 1, 2000; 146(11): 2855 - 2864.
[Abstract] [Full Text]

Home page
 J. Bacteriol.Home page
B. Fournier, R. Aras, and D. C. Hooper
Expression of the Multidrug Resistance Transporter NorA from Staphylococcus aureus Is Modified by a Two-Component Regulatory System
J. Bacteriol., February 1, 2000; 182(3): 664 - 671.
[Abstract] [Full Text]

Home page
 J. Bacteriol.Home page
M. V. Francia, J. C. Zabala, F. de la Cruz, and J. M. García Lobo
The IntI1 Integron Integrase Preferentially Binds Single-Stranded DNA of the attC Site
J. Bacteriol., November 1, 1999; 181(21): 6844 - 6849.
[Abstract] [Full Text]

Home page
 Microbiol. Mol. Biol. Rev.Home page
C. A. Liebert, R. M. Hall, and A. O. Summers
Transposon Tn21, Flagship of the Floating Genome
Microbiol. Mol. Biol. Rev., September 1, 1999; 63(3): 507 - 522.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (486K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (23)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Gravel, A.
Right arrow Articles by Roy, P. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gravel, A.
Right arrow Articles by Roy, P. H.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?