SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

The integration host factor (IHF) of Pseudomonas putida connects cell growth to transcriptional activity of distinct promoters. The IHF site of the σ54 promoter Pu of the TOL (m-xylene biodegradation) plasmid pWW0 of P. putida has been examined to define experimentally a relationship between occupation of the promoter by this factor, the biological activity of the protein and the tolerance of the target site to single-base changes through the bound DNA core sequence. The use of an in vivo high-intensity UV imprinting procedure to examine such an occupation of Pu by IHF allowed inspection of the interplay between the factor and cognate site variants under the physiologically relevant conditions of monocopy gene dosage. The resulting data were merged in a structural model for establishing key features of the IHF–DNA interaction. A functional consensus for first-order IHF binding was instrumental for a genome-wide survey of sequences with potential regulatory value. This search revealed that very few, if any, of the maximum 330 sites within intergenic regions were placed in locations controlling expression of central metabolic genes. It thus seems that the IHF regulon of P. putida has a degree of functional specialization that is not evenly distributed through all gene categories.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Integration host factor (IHF) is a small (20 kDa) basic and heterodimeric protein belonging to the family of prokaryotic nucleoid-associated proteins that binds and bends DNA specifically in a fashion complementary to other factors such as HU, FIS and H-NS (Schmid, 1990; Browning et al., 2010). IHF and its variants are widely present in prokaryotes (Muñoz et al., 2010) as a highly abundant component of the bacterial cell (Ditto et al., 1994; Valls et al., 2002). The canonical structure of the IHF of Escherichia coli consists of two homologous subunits, α and β, encoded by two unlinked genes (ihfA and ihfB; Haluzi et al., 1991; Weisberg et al., 1996). IHF was originally discovered as a factor required for the integration of bacteriophage λ into the E. coli chromosome (Dorman, 2009). However, it was later shown to be involved in a variety of cellular processes representing nearly all the major DNA functions, such as replication, transcription, site-specific recombination, transposition, partitioning, transfer and packaging into phage (Nash, 1996; Ilves et al., 2004; Dorman, 2009). IHF functionality relies on its unique ability to sharply bend DNA at distinct target sites, thereby allowing interactions at a distance between genomic segments otherwise separated by various helical turns. Such a DNA bending is brought about through contacts with the minor groove of specific sequence motifs (Wang et al., 1995; Yang and Nash, 1995), on which the factor produces a distinct protection of 30–35 bp (Friedman et al., 1988; Yang and Nash, 1989; Wang et al., 1995). In E. coli, the consensus IHF binding site approaches the asymmetric motif WATCARnnnnTTR (where W is A or T, R is purine; Goodrich et al., 1990; Lee et al., 1991), often accompanied by an upstream and less defined 4–6 nt A/T tract (Yang and Nash, 1989; Goodrich et al., 1990; Hales et al., 1996; Goodman et al., 1999). Apart of a role in structuring the chromosome, both wet and in silico experiments have established the role of IHF both as a global co-regulator of expression of many genes (Martinez-Antonio and Collado-Vides, 2003) as well as a specific transcriptional factor of particular promoters (Pérez-Martín et al., 1994a; Delic-Attree et al., 1995; Pérez-Martín and de Lorenzo, 1997; Ussery et al., 2001; Browning et al., 2010). These enable expression of genes belonging to virtually all functional profiles (Freundlich et al., 1992; Browning et al., 2010). The question at stake is whether the E. coli's IHF paradigm is maintained in all other bacteria where orthologues of the factor have been found (reviewed in Muñoz et al., 2010) or does it diverge in a host-specific fashion?

In this work, we have addressed this issue by revisiting the binding of the IHF of Pseudomonas putida to target sites in vivo in order to create reference for discerning the regulatory scope of the factor. Our approach entirely stems from the single IHF site located at the intervening region between the upstream activating sequences and the RNAP of the σ54 promoter Pu of the TOL (toluene and m-xylene biodegradation) plasmid pWW0 of P. putida (Ramos et al., 1997). The role of IHF in this promoter is to facilitate the interaction between the m-xylene responsive enhancer-binding activator protein XylR and the promoter-bound σ54-RNAP (de Lorenzo et al., 1991; Calb et al., 1996; Pérez-Martín and Lorenzo, 1996a,b). Specifically, IHF binding to Pu fixes an optimal promoter geometry that facilitates contacts between otherwise distant proteins and aids the recruitment of σ54-RNAP (Bertoni et al., 1998). As a consequence, activity of the normally strong Pu promoter is virtually undetectable in P. putida cells lacking IHF (de Lorenzo et al., 1991; Calb et al., 1996; Valls et al., 2002). We have exploited such a dependence of Pu activity on IHF to set up a reliable in vivo system that measures accurately the functional weight of each of the nucleotides of the bound sequence in IHF binding and activity. A combination of site-directed mutagenesis, in vivo UV imprinting in monocopy gene dosage and lacZ reporter technology has allowed us to generate a reliable standard for surveying potential first-order IHFP. putida binding to target sequences in the genome of this bacterium. Despite some remaining uncertainties regarding the predicted vs. actually bound IHF sites, their location in respect to adjacent genes in the chromosome suggests that IHF has become specialized to regulate a number of housekeeping functions in P. putida that conspicuously rules out central metabolic duties.

Results and discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Surveying specificity determinants of the IHFP. putida–DNA interplay

The starting point of the present study was the examination of both the DNA sequences and the matching amino acid residues that bring about strong interactions between the IHF protein of P. putida and its target sites. To this end, a dependable tridimensional model of such interactions was made based on the considerable similarity of the primary structure of the P. putida IHF α and β subunits with the counterpart factor from E. coli (Rice et al., 1996; Muñoz et al., 2010). The alignment of Fig. 1 reveals an identity of 87% between the α chains of either bacterium and 74% for the corresponding β. Threading the sequence of the IHFP. putida on the known crystal structure of the E. coli protein yields an rmsd value of 0.09 for all residues of both structures (model and template) as can be expected from the high degree of primary sequence homology. The overall distribution of amino acid residues involved in DNA binding is kept, although there are some noticeable differences (see below). Otherwise, acidic amino acids placed on highly conserved regions of either subunit are identical. This structure accounts for the effective mutual complementation between IHF variants of either bacterium (Calb et al., 1996). If one places the E. coli extended consensus binding sequence at-aatt--attaaWATCAR-aggTTR------a-a (Goodrich et al., 1990) on the IHF structure, three regions become immediately noticeable to be important, namely an upstream AT-rich segment which seems to be indirectly recognized, as well as WATCAR and TTR sequences that account for much of the specificity of the interaction (Rice et al., 1996; Rice, 1997). Comparison of such DNA sequences with the strong IHF-binding site present in the σ54-dependent promoter Pu (de Lorenzo et al., 1991; Bertoni et al., 1998; Fig. 2) verifies the gross correspondence between the P. putida protein sites and the nucleotides that are involved in the interaction. Yet, it also pinpoints manifest differences at the critical WATCAR and TTR motifs which seem to primarily determine first-order DNA–protein recognition (Muñoz et al., 2010). On this basis, we set out to identify the optimal DNA sequence primarily targeted by IHFP. putida by examining the effects of directed changes in each of the nucleotides of such a core motif.

image

Figure 1. IHF sequence of P. putida and DNA recognition. Top: Alignment of IHF subunits of P. putida and E. coli. The main interacting regions with DNA are highlighted. Bottom: Structural model of IHF structure of P. putida in complex with its target sequence in the Pu promoter of the TOL plasmid.

Download figure to PowerPoint

image

Figure 2. The IHF site of the Pu promoter of P. putida and its mutant variants. The organization of the relevant region of the TOL plasmid pWW0 is sketched with an indication of the functional elements at play. These include the upstream activating sequences occupied by the m-xylene responsive, the σ54-dependent transcriptional regulator XylR, the IHF site itself (note alignment with the E. coli consensus and the codes and changes of the mutant variants) and the binding site for σ54-RNAP.

Download figure to PowerPoint

In vivo activity of Pu variants containing mutant IHF sites

Integration host factor is strictly required for transcriptional activation of the m-xylene responsive Pu promoter. Figure 2 shows the location of the target site between −47 and −86 upstream of the transcription initiation site. IHF not only assists formation of the right promoter geometry (Pérez-Martín et al., 1994b), but it also helps recruitment of σ54-RNAP (Bertoni et al., 1998; Carmona et al., 1999; Macchi et al., 2003) and suppresses promiscuous activation by other enhancer-binding proteins (Pérez-Martín and de Lorenzo, 1995). This provides an reliable genetic system to examine and quantify the effects caused by replacing one at a time each of the bases directly involved in IHF action as exposed in the structural model discussed above. For this, we generated nine single-base mutant of the IHF site of Pu by replacing the positions with the same co-ordinates of the WATCAR and TTR motifs by the second most frequent nucleotide of the E. coli consensus (Goodrich et al., 1990), as shown in Fig. 2. For the sake of comprehension, we have numbered 1 to 13 each of the bases AATCAAtaatTTA of the Pu promoter that correspond to the E. coli consensus W1A2T3C4A5R6-aggT11T12R13. Note that the UP region of Pu maps further upstream of the IHF site (Bertoni et al., 1998) and therefore none of these mutants interfere with that component of the activation mechanism. This granted, mutant promoters generated as explained in Experimental procedures were passed to vector pBK16 vector for delivery into the same unique genomic site of a specialized reporter P. putida strain bearing a chromosomal copy of the xylR regulator (P. putida KT2442 hom.fg. xylR/S; Kessler et al., 1992). Conjugative transfer of pBK16 derivatives followed by plating on Sm/X-gal plates led to the generation of mono-copy transcriptional Pu-lacZ fusions that maintain the gene dosage and the stoichiometry of the native system (Fig. 3). With this procedure, we generated 10 entirely isogenic reporter strains, which differed only by the sequence of the IHF site at the Pu promoter. They were then separately cultured in LB medium and exposed to saturating vapours of the XylR effector (and thus Pu inducer) m-xylene (Abril et al., 1989). The accumulation of β-galactosidase by each of the strains after 4 h of induction is plotted in Fig. 3 as a percentage of the levels found in the wild type (4538 ± 630 Miller units). The basal activity of the strains without induction or lacking IHF (i.e. in an ihfA mutant; Valls et al., 2002) was < 200 Miller units. LacZ activities of the strains examined break down the collection of mutants into two groups. In one case, mutants in positions 2, 4, 5, 12 and 13 (numbered according to the co-ordinates of the core sequence: see Fig. 2) appear to lose, to different extents, much of the responsiveness to IHF, the most dramatic being 4, 5 and 12. Mutants 2 and 13 keep a residual activity which is in any case very low compared to the wild-type promoter. The second group of mutants (1, 3, 6, 11 in Fig. 2) do keep Pu activity at high levels but display also some differences. While mutants 1 and 3 came down by a mere 20% in respect to the reference promoter, mutants 6 and 13 displayed an activity indistinguishable from the promoter with an intact IHF site – if not even higher. Alignment of these results with the core consensus site of E. coli (Fig. 2) suggested dissimilarities in the way IHFP. putida and IHFE. coli interact with their respective sites. In fact, the trend is that of a relaxation of the consensus that can be recognized by IHF in P. putida. For instance, mutant Pu-1, which undergoes only a ∼ 30% reduction in promoter output, enters a C residue in a position where the consensus accepts only A or T. Similarly, Pu variant 3 affects a very conserved T residue (exchanged by an A in the mutant), but the promoter still works within the range of the wild type. This is in contrast with mutations in adjacent, equally conserved bases that are changed in Pu variants 2, 4 and 5 which, as expected, lost much of their activity. Expectedly also, replacement of the existing A by a G in mutant Pu-6 originated a 100% active promoter. The most surprising occurrence, though, happened at the downstream TTA motif (TTR in the consensus). While changes T[RIGHTWARDS ARROW]C and A[RIGHTWARDS ARROW]C of mutants 12 and 13 respectively did bring down Pu output, replacement of the very conserved T by an A in mutant 11 had no consequences.

image

Figure 3. Transcriptional activity of Pu-lacZ fusions bearing mutated variants of the IHF site. A. Integration of Pu variants into the chromosome of xylR+ strain P. putida hom. fg. The homology fragment pre-inserted in the chromosome bears a selectable aphA (KmR) gene flanked by an N-terminally truncated aadA gene (SmR) and a divergently oriented N-truncated lacZ gene. These DNA segments exchange cognate sequences engineered in the lacZ transcriptional fusion vector pBK16 as shown by means of homologous recombination. See text and Kessler et al. (1992) for details. B. β-galactosidase accumulation of xylR+ strains bearing chromosomal insertions of Pu variants to lacZ. Cultures of each strain with the Pu mutant indicated were grown to early exponential phase in LB medium in the presence of saturating vapours of the XylR inducer m-xylene. Aliquots of the bacterial cultures were assayed for β-galactosidase activity after lysis with chloroform and SDS (Miller, 1972). Transcriptional activity is shown in reference to that of the wild-type Pu sequence (4538 ± 630 Miller units).

Download figure to PowerPoint

A provisional conclusion of all of the above is that the IHF–DNA interplay paradigm of E. coli may not be automatically translated to P. putida (and perhaps to any other species either) and that a specific consensus sequence has to be raised for this bacterium. Still, the data above were generated through an indirect procedure for measuring the action of IHF on Pu and it says little on the actual occupation of the sites in vivo. To examine whether promoter output was faithfully correlated with IHF binding to the various mutant promoters we resorted to a direct method to assess protein–DNA interactions in vivo, as explained below.

Monitoring IHF binding in vivo to Pu promoter variants with high-intensity UV imprinting

The degree of occupancy of the IHF site of the Pu promoter by the factor under the same induction conditions employed to measure β-galactosidase above was inspected with a non-disruptive in vivo footprinting/imprinting procedure. The method employed (which is explained in detail in Valls and de Lorenzo, 2003) is based on the pattern of [32P] 5′-end-labelled primer extension products caused by the arrests of the proceeding DNA polymerase when it finds intra-strand pyrimidine dimers originated by UV irradiation of whole cells. Since formation of such T^T or T^C dimers is greatly influenced by the neighbouring structure of the DNA, in particular by its local curvature, the binding (or lack of it) of a DNA-bending protein can easily be assessed (Engelhorn et al., 1995; Valls et al., 2002). To have a reliable indication of the attachment of IHF to its target site in Pu with this method we first run the procedure in vitro with plasmid DNA encompassing the region of interest and the purified IHF protein of P. putida. The [32P]-labelled oligonucletide was designed to prime upstream of the transcription initiation site for having a maximum (i.e. single nucleotide level) resolution of the IHF binding sequence. Figure 4 shows the result of such an in vitro experiment, where the gel loaded with the extension products displayed clear-cut descriptor of IHF binding in the relative ratio of bands −58/57 vs. −66/65 as well as in the protection of five out of the nine bands corresponding to the long T-rich track at −75/86. These patterns are identical to those found with an alternative high-energy UV laser imprinting procedure (Valls et al., 2002). With this reference in hand, we set out to run the same protocol with live cells. To this end, the strains under examination (carrying either the wild-type Pu sequence or its mutant variants integrated in the chromosome) were grown in LB plus m-xylene up to early stationary phase, the condition where IHF concentration in the cells is maximal (Valls et al., 2002). The cultures were then irradiated with a non-coherent high-intensity UV light beam (Experimental procedures) and the genomic DNA extracted for primer extension. Figure 4 shows a gel loaded with each of the corresponding labelled products. The control sample with the wild-type Pu promoter exhibited intense bands at −66/65, very faint signals at −58/57 and a considerable protection of the upstream T-track, all indicating a complete occupation of the promoter by IHF. The same was true for Pu mutant 1, whose change in respect to the E. coli consensus was as irrelevant for IHF binding as it was for promoter activity (Fig. 3). In contrast, mutant 2 gave a pattern of bands virtually identical to that of the wild-type promoter in vitro in the absence of IHF, indicating that this sequence cannot bind (or binds poorly) IHF in vivo either. In the case of mutants 3 and 4, the bands corresponding to positions −66/65 disappeared because the native TC duplet is changed to AC and TA respectively, which cannot yield pyrimidine dimers with UV light. Yet, the other descriptors indicated that Pu mutant 3 was bound by IHF while variant 4 was not at all. This is also in good agreement with the Pu activity data of Fig. 3. Mutant Pu-5 diverged considerably from the previous ones, because both doublets −66/65 and −58/57 were produced with similar intensities. While highlighting of TC doublet at −66/65 does indicate IHF binding, the bands of the upstream T track match those of the unbound form of the sequence. This pattern is not trivial to explain structurally, but it suggests that IHF can interact with such a Pu variant yet it fails to bend DNA. This could explain the virtual lack of activity of the cognate Pu-lacZ fusion (Fig. 3). Pu mutants 6 and 11 yield interaction patters virtually indistinguishable from the wild type, even more pronounced in the bands that demonstrate binding. It comes as no surprise that promoter activity is also indistinguishable from the non-mutated Pu. Finally, the extension products of variants 12 and 13 are very similar to those of Pu mutant 5 discussed above (i.e. possible IHF interaction with DNA, but little or bending) and the transcriptional activities that they endow are low as well.

image

Figure 4. In vivo UV imprinting of the IHF binding region in the Pu promoter. A. Extension profile of the IHF region of the Pu promoter. The lanes to the left show extension products resulting from priming the IHF region of the promoter in vitro in the absence (−) or presence (+) of purified IHFP. putida following UV irradiation. Note the bands originated in T^T and T^C products at the upstream AT extension of the site at −76/−86 as well as the two doublets at −66/65 and −58/57 of the core binding sequence. For the other samples, P. putida strains bearing Pu variants with the changes in the IHF site indicated were grown in LB medium and irradiated with UV at stationary phase. Genomic DNA was then extracted subject to primer extension as explained in the text and loaded in the same gel. B. Mutations born by the Pu promoter variants. Each sequence is aligned and numbered in respect to the co-ordinates of each of the nucleotides that form the core binding site in the wild-type promoter (AATCAAtaatTTA; Fig. 2). The T^C and T^T doublets that produce the diagnostic bands in the gel are highlighted.

Download figure to PowerPoint

Structural understanding of the IHFP. putida–DNA interplay

Figure 5 summarizes the results of both the Pu-lacZ experiments and the UV in vivo imprinting assays. The gross effects of the mutations examined in this work match the role of the corresponding nucleotide in the structural model, although some differences with the E. coli case become apparent. First, exchange A1[RIGHTWARDS ARROW]C1 results in a still active promoter (Fig. 3), perfectly able to bind IHF (Fig. 4). That position thus tolerates a base other than a W (A,T) perhaps because – according to the structural model – there may not be direct interactions of the protein with that DNA position. In contrast, location 2 of the target sequence is not permissive, for the reason that a A2[RIGHTWARDS ARROW]G2 change dramatically decreased both Pu promoter occupation and transcriptional activity. The basis of this requirement could be that the phosphate group of A2 is predicted to directly interact with amino acid residues H54 and R56 of the IHF chain β (Fig. 1). In position 3 (Fig. 2), most known E. coli sequences have T, although an A is occasionally found (Muñoz et al., 2010). This fits the behaviour of the T3[RIGHTWARDS ARROW]A3 change, which causes only a small reduction of both IHF binding and Pu promoter activity. Threading the IHFP. putida–DNA sequences in the crystal structure of Rice et al. (1996) and Rice (1997) prognoses a DNA bending of 32° at this location as a result of protein-induced buckling, and either T or A in position 3 tolerates a narrower minor groove as predicted. In addition, the base at position 3 interacts with IHF indirectly through a water molecule, what may increase its tolerance to mutation. According to Rice et al. (1996) and Rice (1997) the DNA molecule buckles up upon insertion of the P64 of IHF subunit α in the positions 4 and 5 of the consensus. The C and A nucleotides in these positions are conserved in every known IHF site (Muñoz et al., 2010), probably because these two base pairs are the only ones of the E. coli consensus involved in direct interactions with the protein. This seems to be the case of the P. putida protein, because Pu variants C4[RIGHTWARDS ARROW]A4 and A5[RIGHTWARDS ARROW]C5 mutations lack both IHF binding and transcriptional activity. Position 6 is believed to have a strong preference for an A (and less for G, thereby the W in the consensus code), because mutation of its paired T reduces binding by fivefold (Muñoz et al., 2010). Our data indicate this not to be essential in P. putida, because Pu mutant 6 behaves in a fashion impossible to tell apart from the wild type. It thus seems that either A or G at position 6 further favour intercalation of IHFα P64, allowing extensive hydrophobic contacts among the proline and base rings. Next, according to the crystal structure of the E. coli's protein (Rice et al., 1996), the second portion of the consensus (T11T12R13) is recognized by the core body of IHF rather than its distinct upwards extensions (Fig. 1). In this case, amino acid R46 of IHF chain β reaches out the minor groove of DNA to make both direct and water-mediated hydrogen bonds to the three conserved bases and van der Waals interactions with the ribose moieties. Such contacts through a single protein residue require a narrower minor groove and explain the bias against bases with bulky amino groups (G, C) at positions 11–12 but tolerance to T[RIGHTWARDS ARROW]A transversions. Our results with variants T11[RIGHTWARDS ARROW]A11 mutation (full activity, full binding) and T12[RIGHTWARDS ARROW]C12 (unproductive binding, no activity) are consistent with this view. Finally, stable IHF–DNA interactions require a somewhat odd kink between bases 12 and 13 of the consensus that adds to the overall sharp bending. Therefore, an easily deformable duplet such as T12 R (A,G)13 often occurs in existing native IHF sites. Consistently with this, our Pu mutant 13 (A13[RIGHTWARDS ARROW]C13) had only a marginal transcriptional activity, which surely resulted from a suboptimal binding to the factor.

image

Figure 5. Distribution of intergenic IHF sites in the genome of P. putida KT2440. The upper part of the figure aligns the various IHF binding sequences studied in this work. The 330 sites located within intergenic sequences are indicated around the chromosome of P. putida KT2440 (ORFs encoded in each DNA strand indicated in colour). There seem to be islands by regions 0.5, 1.5, 3,75 and 5.7 Mb with very low frequency of sites. This is in contrast with, e.g. genomic site at c. 5.0 Mb, which is endowed with abundant IHF binding sequences.

Download figure to PowerPoint

Figure 5 shows the apparent core consensus WWCARnnnnWTR for IHFP. putida binding that recapitulates all of the above results and analyses. Note that the sequence is somewhat more permissive than the one proposed for E. coli (WATCARnnnnTTR). The P. putida's sequence was then employed to survey its chromosome for such sites and thus examine the potential regulatory landscape of the factor in this bacterium. To this end, it should be noted that recognition of IHF target sites is driven by indirect readout (Rice et al., 1996; Muñoz et al., 2010), which involves also interactions with less defined adjacent sequences (see above). Although such extensions (typically AT-rich) are important, their role in mediating binding vs. bending vs. functionality is not straightforward (Goodman et al., 1999). Also, since only single-base mutations were employed to produce the consensus, then we cannot rule out that given combinations of other changes could also accommodate IHF binding. This sets some limitations on the use of the consensus IHFP. putida first-order binding motif just proposed for identifying sites at a genome scale. Yet, the above core sequence is expected to generally rule out false negatives, although not false positives. With these considerations in mind we set out to employ the core sequence WWCARnnnnWTR as a reference to estimate the maximum extent of the IHF regulon in P. putida.

Distribution of possible IHF binding sites in the P. putida genome

Estimations of the number of IHF binding sites in the genome of E. coli have yielded disparate results ranging from 70 to over 600 locations (Boffini and Prentki, 1991; Freundlich et al., 1992; Ussery et al., 2001; Grainger et al., 2006; Grainger and Busby, 2008). In our case, we resorted to use the functional information regarding the IHF site of the Pu promoter presented above to estimate the upper limit of this figure in the chromosome of P. putida KT2440. Using the consensus sequence determined before, we found 2653 sites, which matched the motif WWCARnnnnWTR and thus represent the maximum edge of the functional extent of the factor. Evidently, these reflect core target DNA sequences that are necessary for specific recognition, but functional binding may not occur unless there are other less defined characteristics in the adjacent sequences, e.g. an upstream AT-rich track and overall bendability of the region (see above). However, given the increasing number of IHF molecules of P. putida along growth (approx. from 100 to > 4000; Valls et al., 2002) it could well happen that many of the sites are in fact occupied in cells grown to stationary phase. Out of them, 330 WWCARnnnnWTR sequences were found in intergenic regions (Table S1) and are thus likely to have a role in regulating expression of contiguous genes. The set of sites encompasses ORFs for functions belonging to different categories (Table S2), including amino acid transport/metabolism, cell envelope and outer membrane biogenesis, coenzyme metabolism, energy production and conversion, signal transduction mechanisms, inorganic ion transport and metabolism, transcription and translation, ribosomal structure and biogenesis (summary in Fig. 6). Also, a large share of IHF sites includes hypothetical proteins within or close to mobile elements, suggesting an important role of the factor in the occurrence of horizontal gene transfer (Fass and Groisman, 2009). Surprisingly, we did not find IHF sites in any regulatory position near genes encoding central carbon metabolism (i.e. core pathways for consumption/production of carbohydrates and amino acids), an issue that is discussed below. Finally, we examined the positioning of IHF binding sites in respect to repetitive extragenic palindromic (REP) sequences from P. putida. Enterobacterial REP sequences are associated to IHF (Boccard and Prentki, 1993; Oppenheim et al., 1993), but the counterparts found in P. putida may not follow the E. coli paradigm (Aranda-Olmedo et al., 2002; Ramos-Gonzalez et al., 2006). Inspection of the ∼ 800 REP sites of P. putida (Aranda-Olmedo et al., 2002) in respect to predicted IHF sites revealed that there was not one case where the two occurred on top of each other. Moreover, only in 49 instances the distance between IHF and REP sites was < 100 bp (not shown). This indicates that the arraying of REP sequences with IHF sites that seem to be important for nucleoid structure and evolution in some bacteria (van Belkum et al., 1998) has little importance in P. putida. Perhaps this is related to the high GC content found in REP sequences (Aranda-Olmedo et al., 2002), which stands in contrast with the elevated AT share of IHF sites.

image

Figure 6. The IHF regulon in P. putida. The figure summarizes the functional classes of the genes which are likely to be controlled by IHF because of the presence of consensus binding sites at their nearmost intergenic regions (see Table S2). The predicted IHF regulon encompasses different functional categories (a few relevant examples are indicated), which conspicuously lack central metabolic functions.

Download figure to PowerPoint

Conclusion

Each bacterial group seems to possess a limited number of global regulators. Some of them are associated to the nucleoid and regulate or co-regulate expression of large sets of genes regardless of their specific function (Martinez-Antonio and Collado-Vides, 2003). In E. coli, as little as 7–10 factors that respond to a suite of environmental and metabolic conditions suffice to submit expression of individual genes to the general physiology of the cells at any given time. IHF appears among the top factors that rule this over-imposed control (Ma et al., 2004) and it has bona fide orthologous in many bacteria, including P. putida. The question at stake is whether sharing the same molecular mechanism of action (i.e. binding and sharply bending DNA) goes necessarily together with sharing the same functional scope in different bacteria as well. While the key properties of IHF in its interplay with DNA may be preserved along the phylogenetic tree, small differences in interaction parameters might entirely change the functional window of activity in vivo. This has been observed, e.g. in Crp proteins borne by non-E. coli species, which are structurally very close to the archetypal enterobacterial regulator, but they run an entirely different biological duty depending on the specific host (Milanesio et al., 2011). We show above that the IHF of P. putida shares much of the overall protein structure and target DNA sites with the E. coli counterpart. Yet, it seems that minor disparities in their preferred bound sequences can shift their window of functionality towards dissimilar regulatory duties. Note that the consensus that we propose on the basis of mutagenizing the IHF site of Pu sets an upper limit to the number of genes likely to be influenced by the factor in P. putida, but does not predict accurately what sites may or may not be occupied at given times. Intuitively, this should depend on the specific affinity of IHF for each site, a parameter surely influenced by the sequences flanking the core recognition site (Goodman et al., 1999) and the concurrent binding of other proteins to overlapping or adjacent DNA segments. In this context, it is remarkable that even an overestimation IHF-controlled genes (Table S1) hardly contains any that participates in regulation of central metabolic functions. This conspicuous absence is intriguing and will be the subject of future studies.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Bacterial growth conditions and general methods

Escherichia coli strains were always grown at 37°C in LB or LB-agar supplemented, when required with 100 µg ml−1 of ampicillin, 50 µg ml−1 of streptomycin or 30 µg ml−1 of chloramphenicol, as needed. Unless indicated otherwise, P. putida was grown in LB at 30°C in the presence of 200 µg ml−1 of streptomycin, when required. Plasmids were transferred from donor E. coli strains into the P. putida receptor strain by mobilization using a filter mating technique (de Lorenzo and Timmis, 1994). Tripartite matings were set with equal amounts of donor, receptor and helper E. coli HB101 (pRK600) strains and incubated for 8 h at 30°C on the surface of LB plates. Cells were resuspended in 10 mM MgSO4 and appropriate dilutions plated on M9 minimal medium with 0.2% citrate supplemented with streptomycin, X-gal and 3 mM 3-methyl-benzyl-alcohol (3-MBA). This medium permitted the counter-selection of E. coli, and the identification of double recombinants as pale blue or blue colonies, depending on the activity of the Pu promoter mutant (see below). Double recombinants were re-streaked and verified by sensitivity to kanamycin. For quantification of promoter output, β-galactosidase activity assay was measured from 100 µl of bacterial cultures following the procedure of Miller (1972). The linearity of the assay within the range of cell densities and the time of reaction with o-nitrophenyl-β-D-galactoside was verified in all cases. β-galactosidase activity values given throughout this study represent the average of two independent experiments using three different clones in duplicate samples.

Construction of site-directed mutants in Pu and introduction into the P. putida chromosome

Plasmid pEZ9 was used as a template for mutagenesis of the IHF-binding site in Pu promoter. This is a pUC18 derivative that carries the entire region between co-ordinates −208 and +93 of the Pu sequence in respect to transcription initiation, inserted as an EcoRI-BamHI fragment (de Lorenzo et al., 1991). Single-base mutations were introduced in pEZ9 by using the QuikChange™ XL site-directed mutagenesis kit (Stratagene) with pairs of complementary 33-mer oligonucleotides bearing the desired mutation in the central position. This mutagenic procedure originated pEZ9 derivatives each bearing a different single-base change in Pu that was verified through DNA sequencing of the insert region. For recombination of the different Pu variants at the same location in the P. putida chromosome, the wild-type Pu sequence and its nine single-base variants were excised from the pEZ9 plasmids with EcoRI/BamHI and re-cloned in the corresponding sites of the delivery plasmid pBK16. This plasmid contains a small polylinker flanked by the aadA streptomycin/spectinomycin resistance cassette and the lacZ gene, both containing an amber codon in their sequences (Kessler et al., 1992). Cloning of the Pu promoter and its mutant derivatives in the EcoRI/BamHI polylinker sites originated a corresponding collection of Pu-lacZ transcriptional fusions. pBK16-Pu and its mutant variants were routinely maintained in an E. coli supF strain that enabled to select plasmid maintenance by resistance to streptomycin (Kessler et al., 1992). pBK16-derived plasmids were mobilized by conjugative transfer in P. putida KT2442 hom. fg. xylR/S– a specialized Pseudomonas strain that bears a Km resistance gene between truncated aadA and lacZ genes in the chromosome (Kessler et al., 1992). P. putida KT2442 hom. fg. xylR/S also bears a chrosomosomal insertion with the xylR gene under the control of its native promoter. Selection of transconjugants (see above) enabled the isolation of double recombinants in which the Pu variants had replaced the KmR cassette in the chromosome (Fig. 3).

In vivo UV footprinting/imprinting

The P. putida strains under examination were grown in liquid LB until an OD600 = 1.0 in the presence of inducer m-xylene. The cultures were then divided in 50 µl of aliquots into 96-well plates and each well was irradiated with UV during 25–30 s as described (Valls and de Lorenzo, 2003). Aliquots were immediately collected, mixed and chilled on ice. Genomic DNA was extracted out of 500 µl of light-exposed bacterial cultures and processed (see Valls et al., 2002, for details), in order to reveal the photo-reactivity of the DNA region under study. To this end, 5–10 µg of the treated genomic DNA were subject to 45 cycles of primer extension (1 min 94°C, 1 min 59°C, 1 min 72°C) using Taq polymerase (Perkin Elmer, MA) and the radioactively [32P]-5′-labelled oligonucleotide 5′CCCGCTTTGAGGATATACATGGCGAAAGC3′. This sequence is complementary to positions +63/+91 on the coding strand of the Pu promoter DNA. Amplifications were performed in a standard PCR buffer with 2.5 mM MgCl2, 5 pmol of labelled primer and five units of polymerase. After a final extension for 10 min at 72°C, the amplified products were ethanol-precipitated, washed with 70% ethanol and resuspended in 5 µl of TE and 10 µl of formamide loading buffer (Ausubel et al., 1994). Half volume of the samples was analysed by electrophoresis in 6% denaturing polyacrylamide gels (Ausubel et al., 1994), which were subsequently dried, exposed for 1 week to a Molecular Imager System ®-sensitive screen and visualized with the Quantity One software (Bio-Rad). For in vitro control experiments, 50 ng of supercoiled pEZ9 plasmid – which contains the Pu promoter – were incubated in the presence or absence of the purified P. putida IHF protein as described (Valls et al., 2002). Ten-microlitre samples were then irradiated for 15–20 s with the UV lamp in 0.5 ml Eppendorf tubes. Each sample was subjected to 10 amplification cycles in a total volume of 20 µl in the same conditions described for genomic DNA (see above). After addition of 6 µl of formamide dye, 2 µl of each sample were loaded for electrophoresis and analysis.

Homology modelling

To generate a structural model of the IHF protein of P. putida and its target DNA sequences (Muñoz et al., 2010), we generated multiple alignments Clustal-W on the Psi-Blast list for IHF-like proteins, using the sequence corresponding to both chains A and B from P. putida (according to SMART database). The two chains of IHF were treated independently (one at a time), since experimental information suggests that the implication of both chains is not the same in the binding of DNA: while IHF-α participates in the recognition of the first segment of the DNA-consensus sequence (WATCAA for E. coli), IHF-α interplays with the second portion of the consensus, TTG. Models for each of the chains were generated with the alignments obtained previously from clustal-W and using Swiss-Model, a protein structure homology modelling server (Arnold et al., 2006). The quality of model structures was analysed for packing quality, rotamer normality, bond angles and lengths, side chain planarity, dihedral angles, etc., using the tools available at the What-If package (Vriend, 1990). Four residues not involved in protein–DNA interactions in the crystal structure are observed to be facing the protein groove and therefore were not considered in this study. A complete account of the methods employed for this model is to be found in Muñoz et al. (2010).

Prediction of possible IHF sites in P. putida KT2440 genome

The consensus sequence driven from the structural and functional analysis in vivo (WWCARnnnnWTR, where n = A, T, G or C; W = A or T; R = A or G) was employed to probe the genome of P. putida (accession number NC_002947) for matching sites using the Virtual Footprint server http://prodoric.tu-bs.de/vfp (Munch et al., 2005). Genomic search was restricted, where indicated, to non-coding, putative intergenic regions and manually curated, e.g. sites located at the intergenic region between the 3′ ends of two convergently transcribed (and thus unlike to play a role in gene expression) were removed from the analysis. To examine relationships with REP sequences, the genomic co-ordinates of the IHF sites were compared pair-to-pair with DNA segments previously reported for P. putida (Aranda-Olmedo et al., 2002) using an ad hoc perl script.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

This work was defrayed by generous grants of the CONSOLIDER program of the Spanish Ministry of Science and Innovation, by the BACSIN and MICROME Contracts of the EU and by funds of the Autonomous Community of Madrid.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information
  • Abril, M.A., Michan, C., Timmis, K.N., and Ramos, J.L. (1989) Regulator and enzyme specificities of the TOL plasmid-encoded upper pathway for degradation of aromatic hydrocarbons and expansion of the substrate range of the pathway. J Bacteriol 171: 67826790.
  • Aranda-Olmedo, I., Tobes, R., Manzanera, M., Ramos, J.L., and Marques, S. (2002) Species-specific repetitive extragenic palindromic (REP) sequences in Pseudomonas putida. Nucleic Acids Res 30: 18261833.
  • Arnold, K., Bordoli, L., Kopp, J., and Schwede, T. (2006) The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics 22: 195201.
  • Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, K. (1994) Current Protocols in Molecular Biology. New York: John Wiley and Sons.
  • van Belkum, A., Scherer, S., van Alphen, L., and Verbrugh, H. (1998) Short-sequence DNA repeats in prokaryotic genomes. Microbiol Mol Biol Rev 62: 275293.
  • Bertoni, G., Fujita, N., Ishihama, A., and de Lorenzo, V. (1998) Active recruitment of σ54-RNA polymerase to the Pu promoter of Pseudomonas putida: role of IHF and αCTD. EMBO J 17: 51205128.
  • Boccard, F., and Prentki, P. (1993) Specific interaction of IHF with RIBs, a class of bacterial repetitive DNA elements located at the 3′ end of transcription units. EMBO J 12: 50195027.
  • Boffini, A., and Prentki, P. (1991) Identification of protein binding sites in genomic DNA by two-dimensional gel electrophoresis. Nucleic Acids Res 19: 13691374.
  • Browning, D.F., Grainger, D.C., and Busby, S.J. (2010) Effects of nucleoid-associated proteins on bacterial chromosome structure and gene expression. Curr Opin Microbiol 13: 773780.
  • Calb, R., Davidovitch, A., Koby, S., Giladi, H., Goldenberg, D., Margalit, H., et al. (1996) Structure and function of the Pseudomonas putida integration host factor. J Bacteriol 178: 63196326.
  • Carmona, M., de Lorenzo, V., and Bertoni, G. (1999) Recruitment of RNA polymerase is a rate-limiting step for the activation of the σ54-promoter Pu of Pseudomonas putida. J Biol Chem 274: 3379033794.
  • Delic-Attree, I., Toussaint, B., and Vignais, P.M. (1995) Cloning and sequence analyses of the genes coding for the integration host factor (IHF) and HU proteins of Pseudomonas aeruginosa. Gene 154: 6164.
  • Ditto, M.D., Roberts, D., and Weisberg, R.A. (1994) Growth phase variation of integration host factor level in Escherichia coli. J Bacteriol 176: 37383748.
  • Dorman, C.J. (2009) Nucleoid-associated proteins and bacterial physiology. Adv Appl Microbiol 67: 4764.
  • Engelhorn, M., Boccard, F., Murtin, C., Prentki, P., and Geiselmann, J. (1995) In vivo interaction of the Escherichia coli integration host factor with its specific binding sites. Nucleic Acids Res 23: 29592965.
  • Fass, E., and Groisman, E.A. (2009) Control of Salmonella pathogenicity island-2 gene expression. Curr Opin Microbiol 12: 199204.
  • Freundlich, M., Ramani, N., Mathew, E., Sirko, A., and Tsui, P. (1992) The role of integration host factor in gene expression in Escherichia coli. Mol Microbiol 6: 25572563.
  • Friedman, D.I., Weglenska, A., Jacob, B., Sirko, A., Pratt, T.S., Steiner, T., et al. (1988) Integration host factor: a protein for all reasons. Cell 55: 545554.
  • Goodman, S.D., Velten, N.J., Gao, Q., Robinson, S., and Segall, A.M. (1999) In vitro selection of integration host factor binding sites. J Bacteriol 181: 32463255.
  • Goodrich, J.A., Schwartz, M.L., and McClure, W.R. (1990) Searching for and predicting the activity of sites for DNA binding proteins: compilation and analysis of the binding sites for Escherichia coli integration host factor (IHF). Nucleic Acids Res 18: 49935000.
  • Grainger, D.C., and Busby, S.J. (2008) Global regulators of transcription in Escherichia coli: mechanisms of action and methods for study. Adv Appl Microbiol 65: 93113.
  • Grainger, D.C., Hurd, D., Goldberg, M.D., and Busby, S.J. (2006) Association of nucleoid proteins with coding and non-coding segments of the Escherichia coli genome. Nucleic Acids Res 34: 46424652.
  • Hales, L.M., Gumport, R.I., and Gardner, J.F. (1996) Examining the contribution of a dA+dT element to the conformation of Escherichia coli integration host factor-DNA complexes. Nucleic Acids Res 24: 17801786.
  • Haluzi, H., Goitein, D., Koby, S., Mendelson, I., Teff, D., Mengeritsky, G., et al. (1991) Genes coding for integration host factor are conserved in Gram-negative bacteria. J Bacteriol 173: 62976299.
  • Ilves, H., Horak, R., Teras, R., and Kivisaar, M. (2004) IHF is the limiting host factor in transposition of Pseudomonas putida transposon Tn4652 in stationary phase. Mol Microbiol 51: 17731785.
  • Kessler, B., de Lorenzo, V., and Timmis, K.N. (1992) A general system to integrate lacZ fusions into the chromosomes of gram-negative eubacteria: regulation of the Pm promoter of the TOL plasmid studied with all controlling elements in monocopy. Mol Gen Genet 233: 293301.
  • Lee, E.C., MacWilliams, M.P., Gumport, R.I., and Gardner, J.F. (1991) Genetic analysis of Escherichia coli integration host factor interactions with its bacteriophage lambda H' recognition site. J Bacteriol 173: 609617.
  • de Lorenzo, V., and Timmis, K.N. (1994) Analysis and construction of stable phenotypes in gram-negative bacteria with Tn5- and Tn10-derived minitransposons. Methods Enzymol 235: 386405.
  • de Lorenzo, V., Herrero, M., Metzke, M., and Timmis, K.N. (1991) An upstream XylR- and IHF-induced nucleoprotein complex regulates the sigma σ54-dependent Pu promoter of TOL plasmid. EMBO J 10: 11591167.
  • Ma, H.W., Buer, J., and Zeng, A.P. (2004) Hierarchical structure and modules in the Escherichia coli transcriptional regulatory network revealed by a new top-down approach. BMC Bioinformatics 5: 199.
  • Macchi, R., Montesissa, L., Murakami, K., Ishihama, A., de Lorenzo, V., and Bertoni, G. (2003) Recruitment of σ54-RNA polymerase to the Pu promoter of Pseudomonas putida through integration host factor-mediated positioning switch of alpha subunit carboxyl-terminal domain on an UP-like element. J Biol Chem 278: 2769527702.
  • Martinez-Antonio, A., and Collado-Vides, J. (2003) Identifying global regulators in transcriptional regulatory networks in bacteria. Curr Opin Microbiol 6: 482489.
  • Milanesio, P., Arce-Rodriguez, A., Munoz, A., Calles, B., and de Lorenzo, V. (2011) Regulatory exaptation of the catabolite repression protein (Crp)-cAMP system in Pseudomonas putida. Environ Microbiol 13: 324339.
  • Miller, J.H. (1972) Experiments in Molecular Genetics. New York: Cold Spring Harbor Laboratory.
  • Munch, R., Hiller, K., Grote, A., Scheer, M., Klein, J., Schobert, M., and Jahn, D. (2005) Virtual Footprint and PRODORIC: an integrative framework for regulon prediction in prokaryotes. Bioinformatics 21: 41874189.
  • Muñoz, A., Valls, M., and de Lorenzo, V. (2010) Extreme DNA bending: molecular basis of the regulatory breadth of IHF. In Bacterial Chromatin. Dame, R.T., and Dorman, C.J. (eds). Heidelberg: Springer, pp. 365393.
  • Nash, H.A. (1996) The HU and IHF proteins: accessory factors for complex protein-DNA assemblies. In Regulation of Gene Expression in Escherichia Coli. Lin, E.E.C., and Lynch, A.S. (eds). Austin, TX: R.G. Landes Company, pp. 149179.
  • Oppenheim, A.B., Rudd, K.E., Mendelson, I., and Teff, D. (1993) Integration host factor binds to a unique class of complex repetitive extragenic DNA sequences in Escherichia coli. Mol Microbiol 10: 113122.
  • Pérez-Martín, J., and de Lorenzo, V. (1995) Integration host factor suppresses promiscuous activation of the sigma 54-dependent promoter Pu of Pseudomonas putida. Proc Natl Acad Sci USA 92: 72777281.
  • Pérez-Martín, J., and de Lorenzo, V. (1996a) In vitro activities of an N-terminal truncated form of XylR, a σ54-dependent transcriptional activator of Pseudomas putida. J Mol Biol 258: 575587.
  • Pérez-Martín, J., and de Lorenzo, V. (1996b) Physical and functional analysis of the prokaryotic enhancer of the σ54-promoters of the TOL plasmid of Pseudomonas putida. J Mol Biol 258: 562574.
  • Pérez-Martín, J., and de Lorenzo, V. (1997) Clues and consequences of DNA bending in transcription. Annu Rev Microbiol 51: 593628.
  • Pérez-Martín, J., Rojo, F., and de Lorenzo, V. (1994a) Promoters responsive to DNA bending: a common theme in prokaryotic gene expression. Microbiol Rev 58: 268290.
  • Pérez-Martín, J., Timmis, K.N., and Lorenzo, V. (1994b) Co-regulation by bent DNA. Functional substitutions of the integration host factor site at σ54-dependent promoter Pi of the upper-TOL operon by intrinsically curved sequences. J Biol Chem 269: 2265722662.
  • Ramos, J.L., Marques, S., and Timmis, K.N. (1997) Transcriptional control of the Pseudomonas TOL plasmid catabolic operons is achieved through an interplay of host factors and plasmid-encoded regulators. Annu Rev Microbiol 51: 341373.
  • Ramos-Gonzalez, M.I., Campos, M.J., Ramos, J.L., and Espinosa-Urgel, M. (2006) Characterization of the Pseudomonas putida mobile genetic element ISPpu10: an occupant of repetitive extragenic palindromic sequences. J Bacteriol 188: 3744.
  • Rice, P.A. (1997) Making DNA do a U-turn: IHF and related proteins. Curr Opin Struct Biol 7: 8693.
  • Rice, P.A., Yang, S., Mizuuchi, K., and Nash, H.A. (1996) Crystal structure of an IHF-DNA complex: a protein-induced DNA U-turn. Cell 87: 12951306.
  • Schmid, M.B. (1990) More than just ‘histone-like’ proteins. Cell 63: 451453.
  • Ussery, D., Larsen, T.S., Wilkes, K.T., Friis, C., Worning, P., Krogh, A., and Brunak, S. (2001) Genome organisation and chromatin structure in Escherichia coli. Biochimie 83: 201212.
  • Valls, M., and de Lorenzo, V. (2003) Transient XylR binding to the UAS of the Pseudomonas putida σ54-promoter Pu revealed with high intensity UV footprinting in vivo. Nucleic Acids Res 31: 69266934.
  • Valls, M., Buckle, M., and de Lorenzo, V. (2002) In vivo UV laser footprinting of the Pseudomonas putida σ54-Pu promoter reveals that integration host factor couples transcriptional activity to growth phase. J Biol Chem 277: 21692175.
  • Vriend, G. (1990) WHAT IF: a molecular modeling and drug design program. J Mol Graph 8: 5256.
  • Wang, S., Cosstick, R., Gardner, J.F., and Gumport, R.I. (1995) The specific binding of Escherichia coli integration host factor involves both major and minor grooves of DNA. Biochemistry 34: 1308213090.
  • Weisberg, R.A., Freundlich, M., Friedman, D., Gardner, J., Goosen, N., Nash, H., et al. (1996) Nomenclature of the genes encoding IHF. Mol Microbiol 19: 642.
  • Yang, C.C., and Nash, H.A. (1989) The interaction of E. coli IHF protein with its specific binding sites. Cell 57: 869880.
  • Yang, S.W., and Nash, H.A. (1995) Comparison of protein binding to DNA in vivo and in vitro: defining an effective intracellular target. EMBO J 14: 62926300.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information
FilenameFormatSizeDescription
MMI_7835_sm_SuppInfor.pdf124KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.