Review of Identification of the DNA binding sites of PerA, the transcriptional activator of the
bfp and per operons in enteropathogenic Escherichia coli

BIO 4470
Spring 2004

Enteropathogenic Escherichia coli (EPEC) are a group of virulent Gram Negative bacilli that attach to the epithelial lining of the human intestinal tract. EPEC carry an EAF plasmid containing genes for the expression of their primary virulence factor, a bundle forming pilus (BFP). The BFP, a long filament structure that extends from the outer membrane of the cell, attaches to surrounding bacteria to form clusters. The presence of these clusters on the surface of intestinal epithelial cells destroys these cells and degrades the membranes of the intestines. The result is severe bloody diarrhea. Current methods of treatment for the effects of EPEC are continued fluid replacement and the application of broad-spectrum antibiotics, but this often is not sufficient. EPEC are transmitted fecally-orally, and the diarrhea is often life threatening - even fatal - among infants and the elderly in developing countries with poor sewage management and inadequate medicine supplies.

The results of a previous study (1) provided a starting point for Ibarra et al from which they designed their experiments involving PerA. The study found that PerA is directly involved with the expression of the bundle forming pilus in EPEC. The virulence of the bacteria, and consequently the severity of the infection and resulting diarrhea, was linked to the PerA activator by mutating the PerA coding sequence, as well as the coding sequence for a structural protein of the bundle forming pilus, and observing changes in pathogenicity amongst human volunteers (1). Ibarra et al sought to build upon this research and identify the nucleotide sequence within the bfp operon to which PerA binds. Furthermore, Ibarra et al questioned the form in which PerA interacts with DNA. The ultimate goal of these scientists was to learn as much as possible about PerA. Their initial expectations were that they would locate the PerA binding site and identify the quaternary structure of PerA. They succeeded in this and more. Understanding the relationship between PerA and the bfp operon - the operon that controls expression of bundle forming pilus proteins - is important to understanding the epidemiology of EPEC.

Because PerA is difficult to purify by itself, the perA sequence was cloned into the plasmid pMALC2xa to fuse PerA with a maltose binding protein. The resulting fusion protein was purified via an amylose column and Western blotting (Fig 1) and tested to determine its binding affinity to the proposed DNA fragments (Fig 2). Then, a number of electrophoretic mobility shift assays (EMSAs) were performed on a sequence spanning the perA and bfpA regions to determine the general location of the MBP-PerA binding sites (Fig 3). Once the general binding sites were located - as well as that of a secondary bfpA binding site - DNase I footprinting analysis combined the suspected regions with MBP-PerA and DNase I to pinpoint the sequences bound by the MBP-PerA (Fig 4). With the knowledge of this protein-protected region of perA and bfpA, Ibarra et al created DNA probes with these nucleotide sequences and alternately deleted and added base pairs to the ends of the probes. EMSAs (Fig 5) show which alterations reduced and/or eliminated binding of the DNA to MBP-PerA. Figure 6 shows the consensus sequence in the binding sites of perA and bfpA. Next, to identify the sequence of the second bfpA binding site, additional DNase I footprinting analysis, dsDNA EMSAs, and bfpA-cat (chloramphenicol-acetyl transferase) transcription fusions were performed on the suspected region (Fig 7). Finally, with the binding sites identified, Ibarra et al set out to answer one more question - does PerA act as a dimer or monomer? By creating a PerA-LexA fusion protein, the authors assessed the drop in activity when PerA formed a dimer (Fig 8).

Since purified PerA is chemically unstable (2), the first step in this series of experiments was to stabilize the protein by fusing it with the gene of another protein. Using the pMALC2xa plasmid, the perA gene was inserted at the end of malE, a gene on the plasmid that encodes a maltose-binding protein (MBP). The result (called the pMAL-T2 plasmid) was mixed into a batch of BL21 EPEC to transform the plasmids into the cells. The malE and perA genes were transcribed and translated together, forming an MBP-PerA fusion protein. Purification of the MBP-PerA, the steps of which are shown in Figure 1, was accomplished by first lysing the bacteria, extracting the protein material, and running it through a column containing amylose and maltose, to which the MBP portion of the fusion proteins bound while other proteins flowed through. Figure 1A is a photo of a polyacrylamide gel electrophoresis (PAGE). Lane 1 of the PAGE (Fig 1A) was loaded with lysed E. coli; the lane is dark and overloaded with cellular debris. Lane 2 was loaded with the cell material - mostly proteins - that remained after centrifugation and sonication; the multiple bands on the gel show that this mixture contains a variety of proteins. Lane 3 was loaded with MBP-PerA fusion that had been purified by passing through an amylose binding column, which bound the MBP portions of the fusion proteins. Lane 3 shows two distinct bands on the gel. The higher, heavier band represents the MBP-PerA fusions isolated from the amylose column. The slightly lower, lighter band represents those fusion proteins that did not express the entire PerA domain. Figure 1A clearly shows that only one type of protein, the MBP-PerA, was recovered from the E. coli cells. This protein will be used for later experiments.

To further purify the MBP-PerA, Ibarra et al also performed a Western blot. The MBP-PerA were run on a PAGE, transferred to a nitrocellulose filter, and washed with rabbit-formed primary antibodies directed against the MBP-PerA fusion. The fusions were washed again with a secondary antibody, formed by goats and directed against the rabbit antibodies. The bands in Figure 1B show the labeled protein-antibody structures. Lane 1 is the negative control and proves that cells not containing the pMAL-T2 fusion plasmid have nothing to which the antibodies can attach. Lane 2 shows the protein-antibody complex that formed from non-purified MBP-PerA proteins extracted through centrifugation and sonication. Lane 3 shows the results of the Western blot performed on the MBP-PerA column-purified proteins; the slightly cleaner band shows that these are specifically MBP-PerA fusions, whereas the products in Lane 2 might possibly have contained other proteins.

Figures 1C and 1D verify the role of PerA in the expression of BfpA, a major structural protein for the bundle forming pilus. Figure 1C shows the results of a Western blot performed to isolate BfpA from four different sources. Lane 1 shows that healthy, wild type EPEC strains (the strain used was E2348/69) produce BfpA. Lane 2 shows that a cured form of the EPEC, one that does not contain pEAF or the bfp genes, will not produce BfpA. Lane 3 shows that an insertion of the kanamycin-resistance gene into the perA gene on pEAF will prevent the expression of BfpA. Lane 4 shows that such a mutation will be restored when pEAF is paired with a pMAL-T2 plasmid.

While Figure 1C demonstrates the role of PerA in BfpA expression, Figure 1D uses two of the same criteria - a kanamycin insertion into pEAF and the complementation of the mutated pEAF with the pMAL-T2 , which contains a functional copy of the perA gene. The top left quadrant (Fig 1D) is the negative control; it shows human cells with no bacteria present. With no bacteria present, there is no attachment. The lower left quadrant shows the same human cells when exposed to EPEC containing an inactivated perA; there is no attachment. The upper right quadrant is the positive control; the human cells are exposed to virulent, wild type EPEC, and as such, the bacteria attach successfully to the host cells. Finally, the lower right quadrant shows the successful attachment of EPEC, containing an insertional mutation in the pEAF perA gene, when accompanied by a pMAL-T2 plasmid.

The experiments presented in Figure 1 confirm two things. First, the purification of MBP-PerA proved that PerA was a viable tool to work with. Second, by creating pMAL-T2, and complementing it with mutated EPEC, they were able to confirm that PerA is indeed involved in the expression of BfpA and related attachment structures.

Once MBP-PerA was purified, Ibarra et al questioned whether this new fusion protein still had the same DNA binding properties as normal PerA. First, DNA segments were taken from upstream regions of both the bfpA and perA genes and amplified using polymerase chain reaction. This broad region was suspected of containing the PerA binding site, and by repeatedly heating and cooling - with appropriate primers, buffers, and Taq Polymerase - they were able to get numerous copies of each segment. Figure 2 shows the results of an EMSA when the amplified DNA was mixed with purified MBP-PerA. Figure 2A shows that in Lane 6, the MBP-PerA reached a concentration at which it could bind all or most of the available DNA fragments. Figure 2B shows the same concept for the perA gene. These results indicate that PerA still functions while bound to MBP. If PerA had been rendered inactive, then another method of purification would have to have been found in order to study the PerA-DNA relationship.

After stabilizing PerA by creating an MBP-PerA fusion and proving that the PerA domain of the protein still worked, Ibarra et al began an experiment to determine the actual PerA binding site. In order to identify the PerA binding site on the bfpA regulatory region, overlapping DNA fragments of 200-bp were taken from both upstream and downstream areas of the bfpA transcription start site. Each fragment was amplified by PCR, and the products were mixed with MBP-PerA. Each mixture was run on a PAGE, and the results are shown in Figure 3A. The bands higher on the gel represent the complexes formed between the DNA fragments and the MBP-PerA. The bands near the bottom of the gel represent free DNA that was not bound by the MBP-PerA. Each sample was run alongside a negative control containing the DNA fragments but lacking MBP-PerA. No bands were expected in these lanes, and none appeared. The high bands in Lanes 4-11 (+) represent protein-DNA complexes, indicating that each of these segments contained a complete PerA binding site. Lanes 1, 2, 3, and 12 did not form complexes, presumably because they did not contain complete PerA binding sites. Surprisingly, Lane 7 shows two bands of protein-DNA complexes, revealing the presence of a second PerA binding site in the bfpA regulatory region. Figure 3B illustrates these various DNA fragments and how, when overlapped, they provide a map for the PerA binding sites. The results of these EMSAs significantly narrow the range of possible binding sites. Having started with an overall sequence of 636 base pairs, Ibarra et al managed to determine that both binding sites were located somewhere in a 200-bp long region spanning -94 to +106 of the bfpA gene (Fig 3B). To determine the perA binding site, similar EMSAs were performed as shown in Figure 3, but the data for those particular experiments were not included in the article.

Previous EMSAs (Fig 3) certainly narrowed the region containing the PerA binding sites, but the exact sequence was still unknown. To narrow the possibilities even further, Ibarra et al performed protection assays on the suspected DNA region (Fig 4). The desired sequences for both bfpA and perA (determined in the above experiment) were mixed with MBP-PerA fusion proteins and then subjected to the endonuclease activity of DNase I. The MBP-PerA bound to the PerA binding sites within the larger fragments. DNase I then spliced the exposed DNA. DNase I did not destroy the sequences that were covered by MBP-PerA. Figure 4A shows the electrophoretic gel used to locate the region of DNA protected by MBP-PerA within the sequence taken from bfpA. Figure 4B shows the same for the sequence taken from perA. In both gels, Lane 1 was the negative control; it contained DNA and DNase I, but no MBP-PerA, and thus the DNA was not protected from the enzyme. In both gels, Lanes 2 - 6 illustrate the varying degrees of protection provided by MBP-PerA. As the concentration of MBP-PerA increased, represented at the top of each gel by the rising arrow, the amount of protected DNA increased. The black bars on the right side of both gels indicate that the most protected regions lie between -83 to -56 and -81 to -47 for bfpA and perA, respectively. The identification of protein protected regions allowed Ibarra et al to progress to the next step, which was to design double stranded, oligonucleotide probes to test the binding capacity of MBP-PerA to the desired sequences.

Using the sequences determined in the above EMSAs, Ibarra et al designed a number of double stranded DNA probes of varying lengths, containing various portions of the above-identified protected regions for both bfpA and perA. Each of these probes was mixed with MBP-PerA and run on several EMSAs. The results are shown in Figure 5A and 5B. The experiment used an unrelated probe of similar size as a negative control; it was run through the gel without the addition of MBP-PerA. The PBSA45 sequence spanned the entire protected region of the bfpA region, with additional nucleotides on the 5` and 3` ends. As seen in Figure 5A, when MBP-PerA was combined with the PBSA45 probe, normal PerA binding occurred, creating a band of protein and DNA. Normal binding also occurred with the PBSA56, which contained the above PBSA45 probe with an 11 base pair addition to one end. However, deletions at the 3` end of PBSA45 did not allow binding, and no bands of protein-DNA complexes were formed on the gel. Subsequent probes were created, removing several base pairs at a time from the 5` end of PBSA45; PerA binding was reduced but not eliminated. Mutations in the probes - represented in Figure 5A by Mut1, Mut2, and Mut3 - reduced PerA binding, and combinations of these mutations - Mut1-2 and Mut1-2-3 - completely prevented PerA binding to the DNA probes. Figure 5B shows the same methods employed for the identification of the perA binding site. The PBST45 probe spanned the protected region of the perA regulatory region, and PBST56 contained the same sequence as PBST45 with a 10-bp addition to the 3` end. Both showed successful MBP-PerA binding. Tables 2 and 3 list the nucleotide sequences of the probes used in these experiments. The manipulation of these probes and the evaluation of PerA binding efficiency via assays concluded that the PerA binding site for bfpA is between -84 and -45, and the PerA binding site for perA is between -75 and -40.

With the binding sites of both bfpA and perA identified, Ibarra et al then proposed a consensus binding sequence for PerA (Fig 6). Both binding sequences are compared in Figure 6A, with the shaded area identifying the portion the two sequences have most in common, and the proposed consensus sequence listed beneath them. Figure 6B compares the perA binding site with both bfpA binding sites. This consensus sequence will enable future researchers to quickly locate the PerA binding site within a line of genetic code.

With the primary bfpA binding site identified, Ibarra et al then turned to mapping the unexpected second bfpA site pictured in Figure 3. The second bfpA binding site underwent a series of experiments similar to the one described above (Fig 7). After the second bfpA site was discovered in Figure 3, those fragments too were bound to MBP-PerA and treated with DNase I. Figure 7A is the result of the protection assay PAGE. Lane 1 was the control, as in footprinting analyses previously described; it contained no MBP-PerA. The black bar along the right side indicates that the protected region of the second bfpA binding site spans approximately +45 to +100. Figure B shows the result of the EMSA conducted with two different probes. The first probe, PBSA2-56, was mixed with increasing concentrations of MBP-PerA and run on an electrophoretic gel. As the MBP-PerA concentration increased, so did the concentration of protein-DNA complexes. The second probe used, PBSA56, contained the proven binding sequence for the first bfpA binding site. This probe was used as a positive control, because its binding capacity had already been established in the experiments surrounding the primary bfpA binding site. Compared to the PBSA56 probe, the PBSA2-56 probe demonstrated a slightly less efficient capacity for binding to PerA. Finally, Figure 7C charts the results of a bfpA-cat fusion. Two segments from the bfpA regulatory region - one of which lacked the complete second bfpA binding site - were cloned into CAT plasmids (pCAT201 and pCAT+27), in front of the cat gene. The plasmids were then transformed into the wild type EPEC strain (E2348/69) and cured strain (JPN15), and the expression of CAT was measured. Comparing the results pf pCAT 201 and pCAT+27, Ibarra et al deduced that the second bfpA binding site wasn't critical for activation of the bfp operon. Not only did this series of experiments define the range of the second bfpA binding site, it illustrated the PerA binding efficiency of the site, and also provided some information on the function of the site.

Finally, Figure 8 represents the last in this series of experiments, and the last key point in Ibarra et al's article. One question the researchers had at the beginning of this study was regarding the form in which PerA interacts with its DNA binding sites. Was it a monomer, a dimer, or a multimer? To answer this question, the genes for PerA and LexA were fused into the plasmid pM1660PerA, and the proteins were translated and expressed as a LexA-PerA fusion protein. LexA is a known dimer that helps other proteins group their subunits into dimers. As controls, LexA was also fused with several other proteins that also formed dimers, to confirm the ability of LexA to help proteins form dimers. When examining the LexA-PerA efficiency of suppressing B-galactosidase activity, as is the function of the dimers used as controls, Figure 8 shows that PerA does not suppress B-galactosidase activity, and is therefore most likely a monomer.

From this labor-intensive research project, the scientific community learned several important things about EPEC virulence. First, PerA, the regulatory protein for both the bfp and per operons, can be successfully fused with a maltose binding protein to increase stability, without losing its DNA binding affinity. Second, the PerA binding site for bfpA is between -84 and -45, and the PerA binding site for perA is between -75 and -40. Third, there exists within the bfp operon a second bfpA site to which PerA binds. Fourth, comparison of the three binding sites has produced a possible consensus sequence for all PerA binding sites. Finally, PerA regulates the bfp and per operons as a monomer. Combined, these facts provide a wealth of knowledge into the pathogenicity of EPEC. Undeniably, the more that is known about this class of organisms and its mechanisms, the more likely it is that more efficient methods of control and/or treatment will be developed by the scientific community.

This was an extremely challenging, yet ultimately interesting article to review. It required four or five readings before the jargon and the experiments started to make sense. At first, reading the article yielded little insight or understanding as to the discoveries that were made, but the process of scrutinizing each figure and writing about each one in detail, revealed a glimpse of the bigger picture, of the significance of the results. Perhaps the most difficult part of writing this review was trying to understand portions of the figures that the text or legends did not address. The most enjoyable aspect was being able to sit down and write, and see if I understood the material. I may not understand it all, but I certainly comprehend far more about the experiments and reasons behind this research project than when first presented with the assignment.

REFERENCES

1. Bieber, D., S.W.Ramer, C.Y. Wu, W.J. Murray, T. Tobe, R. Fernandez, and G.K. Schoolnik. 1998. Type IV pili, transient bacterial aggregates, and virulence of enteropathogenic Escherichia coli. Science. 280:21142118.

2. Ibarra, J. Antonio, Miryam I. Villalba, and Jose Luis Puente. 2003. Identification of the DNA binding sites of PerA, the transcriptional activator of the bfp and per operons in enteropathogenic Escherichia coli. J. Bacteriol. 185(9):2835-2847.