Random and Site-Specific Mutagenesis of the Helicobacter pylori Ferric Uptake Regulator Provides Insight into Fur Structure-Function Relationships

Summary

The ferric uptake regulator (Fur) of Helicobacter pylori is a global regulator that is important for colonization and survival within the gastric mucosa. H. pylori Fur is unique in its ability to activate and repress gene expression in both the iron-bound (Fe-Fur) and apo forms (apo-Fur). In the current study we combined random and site-specific mutagenesis to identify amino acid residues important for both Fe-Fur and apo-Fur function. We identified 25 mutations that affected Fe-Fur repression and 23 mutations that affected apo-Fur repression, as determined by transcriptional analyses of the Fe-Fur target gene amiE, and the apo-Fur target gene, pfr. In addition, eight of these mutations also significantly affected levels of Fur in the cell. Based on regulatory phenotypes, we selected several representative mutations to characterize further. Of those selected, we purified the wildtype (HpFurWT) and three mutant Fur proteins (HpFurE5A, HpFurA92T, and HpFurH134Y), which represent mutations in the N-terminal extension, the regulatory metal binding site (S2) and the structural metal binding site (S3), respectively. Purified proteins were evaluated for secondary structure by circular dichroism spectroscopy, iron-binding by atomic absorption spectrophotometry, oligomerization in iron-substituted and apo conditions by in vitro cross-linking assays, and DNA binding to Fe-Fur and apo-Fur target sequences by fluorescence anisotropy. The results showed that the N-terminal, S2, and S3 regions play distinct roles in terms of Fur structure-function relationships. Overall, these studies provide novel information regarding the role of these residues in Fur function, and provide mechanistic insight into how H. pylori Fur regulates gene expression in both the iron-bound and apo forms of the protein.

Introduction

Helicobacter pylori is a Gram-negative microaerophile that chronically colonizes the gastric mucosa of over half of the world’s population (Blaser, 1998, Dunn et al., 1997). While infection always results in localized inflammation, only a subset of colonized individuals develops overt disease. In severe cases, infection may lead to gastric ulcers, gastric adenocarcinoma, or MALT lymphoma (Blaser, 1998, Dunn et al., 1997). One of the remarkable characteristics of this highly successful pathogen is the ability to survive in the hostile environment of the stomach. Although H. pylori normally colonizes within the mucus layer near the gastric epithelium, the bacteria are continuously faced with dramatic fluctuations in pH and changes in the availability of nutrients such as iron. As a result of living in this dynamic niche, H. pylori has evolved tightly regulated transcriptional networks that allow the organism to rapidly adapt to changes in the environment.

Despite the importance of regulating gene expression in response to environmental stimuli, the genome of H. pylori encodes for relatively few dedicated transcriptional regulators (Baltrus et al., 2009, Tomb et al., 1997). As a result, several of these regulatory proteins have evolved the ability to function in a capacity beyond what is typically seen for their counterparts in other bacterial species. One such protein is the ferric uptake regulator (Fur). Fur belongs to a large family of metalloregulatory proteins and is classically thought of as a repressor of iron uptake systems (reviewed in (Lee & Helmann, 2007)). Similar to Fur in other bacterial species, H. pylori Fur uses ferrous iron (Fe²⁺) as a co-factor and binds DNA sequences known as Fur boxes to repress transcription (Bereswill et al., 2000, Delany et al., 2001b, Pich et al., 2012). However, there are also examples where Fe-Fur acts as a positive regulator, both directly and indirectly (Alamuri et al., 2006, Delany, 2004, Delany et al., 2001b, Delany et al., 2003, Ernst et al., 2005a, Gancz et al., 2006, Gilbreath et al., 2012). Those studies highlight the fact that Fur is more than a repressor of iron uptake systems and instead functions as a global regulator. This global regulatory role is especially apparent in H. pylori, where Fur has been shown to regulate a large number of genes required for diverse functions, and has also been shown to respond not only to intracellular iron levels but also to changes in pH and levels of oxidative and nitrosative stressors(Cooksley et al., 2003, Qu et al., 2009). One way that H. pylori Fur is able to regulate so many genes under different environmental conditions is by repressing and activating gene expression in both the iron-bound (Fe-Fur) and apo (apo-Fur) forms of the protein (Bereswill et al., 2000, Carpenter et al., 2009, Delany, 2004, Delany et al., 2001a, Delany et al., 2001b, Delany et al., 2003, Ernst et al., 2005b, Gilbreath et al., 2012, Pich et al., 2012, van Vliet et al., 2003). This degree of functional diversity is so far unique to H. pylori Fur, although there have been reports of Fe-Fur activation in other bacteria, and recently apo-Fur repression has been described in Staphylococcus aureus and Campylobacter jejuni (Deng et al., 2012, Grabowska et al., 2011). However, H. pylori remains the only bacterial species where all four types of Fur regulation have been characterized in detail. Thus, a functional understanding of Fur regulation in this organism could provide valuable information that could expand our understanding of Fur regulation in other bacterial species.

Based on crystallographic data from Pseudomonas aeruginosa (Pohl et al., 2003), Vibrio cholerae (Sheikh & Taylor, 2009), and more recently H. pylori (Dian et al., 2011), progress has been made in understanding the important features of the fully metallated form of the protein (holo-Fur). The protein is made up of two major domains: the N-terminal domain, which spans amino acids 1-92 in H. pylori, contains a canonical winged helix-turned-helix motif (helices α2-α4 and sheets β1-β2) that facilitates interaction with target DNA, and a C-terminal domain (residues 93-150) that contains several metal coordinating residues and is responsible for dimerization of Fur protomers. A key feature of H. pylori Fur is an N-terminal 10 amino acid extension; this region is not seen in the well-studied organisms such as Escherichia coli, P. aeruginosa, or V. cholerae and has only been identified in one other bacterial pathogen to date, C. jejuni (Butcher et al., 2012). The structure of H. pylori Fur shows that a portion of this extended region forms helix α1, which stacks against the DNA binding domain and may stabilize this motif. In terms of the C-terminal domain, this region is made up of 3 antiparallel β strands (β3-β5) and helices α5-α6, which serve as the dimerization interface.

Also atypical compared to other holo-Fur structures, the H. pylori Fur protein contains three distinct metal binding sites (S1, S2, and S3). S1 is a structural zinc-binding site (Dian et al., 2011, Vitale et al., 2009) that is coordinated by four cysteine residues (C102, C105, C142, and C145) found in two CXXC motifs. Metallation at this site stabilizes the three β sheets in this region (β3-β5) and is necessary for dimerization (Vitale et al., 2009). S2 is the regulatory binding site; metallation at S2 is responsible for a conformational change in the DNA binding domain that “activates” the protein and facilitates interaction with Fe-Fur target DNA. As the S2 coordination spheres differ in each protomer of the Fur dimer, the exact residues that coordinate the metal ion differ between protomer A and protomer B. In one protomer, the metal ion in S2 is coordinated by H42, E90, H97, and H99; in the other protomer, the ion is coordinated by H42, E90, H97, H99, and E110. In contrast to S2, the S3 site has similar coordination spheres in each protomer of the Fur dimer. Located near the dimerization domain, S3 has a tetrahedral geometry and is coordinated by residues H96, D98, E117, and H134. Previous mutational analysis of the S2 and S3 sites, indicate that S2 is required for holo-Fur (metallated with zinc ions in that study) to bind DNA, but S3 is not required; however, metallation of S3 likely increases binding affinity (Dian et al., 2011).

Until very recently, the structural features of apo-Fur have largely remained a mystery. However, the recent publication of the C. jejuni apo-Fur structure (Butcher et al., 2012) has finally shed some light on how Fur is able to regulate gene expression in the absence of the iron cofactor. Because C. jejuni Fur is most similar to H. pylori Fur in terms of sequence, it is likely that the C. jejuni apo-Fur structure is reasonably similar to H. pylori apo-Fur (Figure S1). Much like its iron-bound counterparts, C. jejuni apo-Fur is a dimer and has a canonical winged helix DNA binding domain. However, when compared to the holo-Fur structures (Dian et al., 2011, Pohl et al., 2003, Sheikh & Taylor, 2009), the DNA binding domain is rotated 180 degrees. As a result of this rotation, helix α1 is positioned within the DNA binding region, and most likely directly interacts with DNA. In addition to the conformational changes in the N-terminal region of the protein, the metal coordination sites of C. jejuni apo-Fur also show some divergence from their metallated counterparts in holo-Fur. Similar to H. pylori holo-Fur, the S1 site in C. jejuni apo-Fur is occupied by a metal ion and is coordinated by two pairs of cysteine residues. Given the similarity of S1 in holo-Fur and apo-Fur, it is likely that this metal ion plays a role in maintaining the dimeric structure for both forms of Fur. The other occupied metal site in the apo-Fur structure is S3; while S3 in H. pylori holo-Fur is tetracoordinated by residues H96, D98, E117 and H134, in C. jejuni apo-Fur S3 is hexacoordinated with D101, D120, H137 and two water molecules (Butcher et al., 2012). These differences in S3 geometry arise from the rotation of the DNA-binding domain, which renders this site permissive to coordinating water molecules (Butcher et al., 2012). Consistent with the hypothesis that S2 is the regulatory or iron-sensing site within Fur, this site is not occupied in the apo- form of the protein.

In the current study, we utilized a combination of random and site-specific mutagenesis to identify residues that are important for iron-bound and apo-Fur function in H. pylori. We identified 23 mutations via random mutagenesis that affected regulation of the Fe-Fur target gene amiE and/or the apo-Fur target gene pfr. Of the five site-specific fur mutations we constructed (R3A, E5A, E8A, R13A, E117R), all were important for Fe-Fur (E5A, E117R) and/or apo-Fur (R3A, E5A, E8A, R13A) function. Based on the regulatory phenotypes observed in the mutant strains, we selected a subset of these mutations for biochemical studies. Of the five mutations we selected to further characterize, we successfully purified the WT (HpFurWT) and three mutant proteins (HpFurE5A, HpFurA92T, and HpFurH134Y). These proteins were evaluated for secondary structure content and overall folding, iron binding, oligomerization, and DNA binding in manganese-substituted and apo conditions. Combined, the results of these studies provide functional insight into Fe-Fur and apo-Fur regulation in H. pylori.

Results

Random mutagenesis and manganese selection

To identify important residues of H. pylori Fur, we opted to initially take a non-biased approach. Using an error-prone PCR-based strategy, we generated a library of ~10,000 random mutant strains; each H. pylori mutant strain contained ~1 nucleotide change per copy of fur, a frequency that was confirmed by sequencing a subset of fur genes from the library. We selected this mutation frequency because it was most likely to result in single amino acid changes in the protein. After creating the library, we used a combination of two methods to identify mutations that resulted in a regulatory phenotype. The first strategy was a manganese selection, which has been successfully used to identify key Fur residues in other bacterial species (Hantke, 1987, Lam et al., 1994, Loprasert et al., 2000). Because manganese acts as a redox stable iron analog, it is thought that strains grown in the presence of excess manganese sense an overabundance of “iron” in the environment. As a result, the bacteria down-regulate expression of iron uptake systems, which prevents iron from entering the cell. As iron is an essential element for almost all living organisms, shutting down iron uptake in response to the excess manganese in the environment is lethal. By plating the fur mutant library on plates that contained 4 mM manganese (manganese (II) chloride), we selected for mutations in fur that precluded down- regulation of the iron uptake systems and allowed the mutant bacteria to grow. Aliquots of the mutant library were diluted and plated for single colonies, and 115 manganese resistant (Mn^R) colonies were archived. From this pool of 115 mutant strains, we sequenced the fur gene to identify mutations that conferred single amino acid changes. We sequenced fur from 15–20 individual mutant strains at a time as a means to determine the saturation point of the mutagenesis. After analyzing the fur sequence from 83 Mn^R strains we did not detect any new mutations that translated into single amino acid changes and thus, we considered this to be the saturation point. From these 83 strains, we identified 19 single amino acid changes in 15 different residues that conferred a Mn^R phenotype. The majority of these residues were localized to the C-terminal domain of the Fur protein (Table 1, Figure 1); in fact, no single amino acid changes were detected within the region spanning amino acids 1-60. To ensure the Mn^R phenotype was linked to the mutation in fur, we next moved each mutation into a clean H. pylori strain background, re-plated on agar that contained manganese, and re-sequenced fur from the newly created strain. In addition, we sequenced the promoter region upstream of fur in each strain to ensure that there were no mutations that could affect fur expression. For all newly created strains, the Mn^R phenotype was transferred with the fur mutation, and there were no additional amino acid changes in the coding region or mutations in the fur promoter. Thus, we concluded that the ability of each strain to grow in the presence of manganese was linked to the fur mutation carried by that strain.

Table 1.

Summary of regulatory phenotypes observed by qRT-PCR

Group	Mutation	Regulatory phenotype
		amiE	pfr
Manganese Selection	S62T	INT	INT
	L80V	Δfur	INT
	R87Q	Δfur	WT
	Y89C	Δfur	INT
	E90Q	INT	INT*
	E90G	INT	INT*
	A92V	Δfur	INT*
	A92T	Δfur	INT*
	K94N	Δfur	INT*
	K94R	Δfur	INT*
	H96L	Δfur	Δfur
	I108N	INT	INT*
	E110D	Δfur	WT
	V123D	Δfur	Δfur
	I132N	Δfur	WT
	H134Q	Δfur	INT*
	H134Y	Δfur	INT*
	D135Y	Δfur	INT*
	M138T	Δfur	Δfur
Pfr screen	G40D	WT	WT
	I52T	Δfur	WT
	S60N	Δfur	INT
	I67F	Δfur	WT
	Y89H	Δfur	INT
Site-specific	R3A	WT	INT*
	E5A	INT	INT
	E8A	WT	INT
	R13A	WT	Δfur
	E117R	Δfur	INT

WT, phenotype similar to wildtype; Δfur, phenotype similar to fur deletion strain; INT, intermediate phenotype;

^*indicates hyper-repression after chelation

Figure 1.

Distribution of random and site-specific mutations across the coding region of H. pylori Fur. (A) Mutated residues obtained by site-specific mutagenesis are designated by yellow letters, mutated residues obtained by manganese selection are designated by cyan, and mutated residues obtained by the Pfr-based screen are designated by magenta. The gap in the secondary structure between residues 92 and 93 denotes the transition from the N-terminal region to the C-terminus (Dian et al., 2011). Metal coordinating residues are indicated by a grey circle (for S1), a grey triangle (for S2) or a grey diamond (for S3). Fur amino acids and secondary structure alignment image generated with Polyview-2D: http://polyview.cchmc.org/ (Porollo et al., 2004) using the H. pylori Fur crystal structure as a template. (B) 3-D representation of H. pylori Fur containing mutated residues. Color scheme for mutated residues is the same as in (A). Image was generated with Polyview 3D: http://polyview.cchmc.org/polyview3d.html (Porollo & Meller, 2007) using the H. pylori holo-Fur crystal structure as a template.

Screen for altered Pfr expression

Although the manganese selection identified multiple residues that appear to be important for Fur function, we considered the fact that performing such a selection under excess metal conditions could bias the results towards identifying mutations that alter metal binding and thus iron-bound repression. This notion was supported by the fact that many of the identified mutations clustered in the C-terminal region of Fur, which contains the vast majority of the metal coordinating residues. To decrease the potential bias of the selection, we also used an alternative strategy to identify mutations that affected apo-Fur repression. Based on previous results (Bereswill et al., 2000) which showed the apo-Fur repressed prokaryotic ferritin protein (Pfr) could easily be visualized by SDS-PAGE followed by Coomassie staining, we designed a visual screen to look for de-repression of the Pfr protein. A representative gel from this screening strategy is depicted in Figure S2. After screeing ~ 400 individual random fur mutant strains, we identified five strains that displayed altered Pfr levels and contained a single amino acid change in Fur (Table 1). These mutations were distributed across both the C-terminal and N-terminal domains of Fur (Figure 1). Once again, to ensure that the regulatory phenotype was linked to the mutations in fur, we moved each mutation into a clean H. pylori strain background, re-screened the newly created strains for altered Pfr levels, and re-sequenced fur and the fur promoter. For all five re-created strains, the regulatory phenotype was linked to the mutation in fur and there were no additional mutations in the coding or promoter regions.

Selection of residues for site-specific mutagenesis

Because analysis of the distribution of the mutant residues showed that some areas of Fur appeared to be under-represented in our phenotypic assays, we also constructed several site-specific mutations in fur. Specifically, we selected four residues in the N-terminal extension (residues 3, 5, 8, and 13) and mutated each of these residues to an alanine (R3A, E5A, E8A, and R13A) (Figure 1 and Table 1). In addition, based on the prediction that changing the charge and size of the side chain at this residue would prevent metallation at this site, we mutated the metal coordinating E117 residue to arginine (E117R). We also created a Cm^R marked control strain (WT_C) to ensure that genetic manipulation of the strains did not affect Fur regulation in some way. These mutations were generated using SOE PCR as described in the Materials and Methods section. Once again, after these site-specific mutant strains were created, we sequenced the fur coding sequence and promoter to ensure no additional mutations were present. Each of the site-specific mutant strains were tested for manganese resistance and altered Pfr levels. Although the E5A and E117R mutant strains displayed an intermediate level of resistance to manganese, none of the site-specific mutant strains displayed altered levels of Pfr by SDS-PAGE (data not shown). As a result of our combined mutagenesis approaches, we now possessed 29 fur mutations to study in further detail. Additionally, we included the WT, WT_C, and fur mutant strains as controls for our future analyses.

Transcriptional analysis of amiE regulation

To determine how the random and site-specific mutations affected iron-bound Fur (Fe-Fur) repression, we quantitated the levels of amiE transcript in the mutant strains using qRT-PCR before and after iron chelation; amiE is a well-characterized Fe-Fur repressed gene and has been previously used as a reliable readout of Fe-Fur regulation (Carpenter et al., 2010, Carpenter et al., 2007). Because amiE transcription is normally repressed by Fe-Fur when iron is replete (T₀) the addition of the chelating agent should result in the de-repression of amiE expression (T₆₀). Thus, we compared changes in the level of amiE transcript after iron chelation (T₆₀/T₀). As expected, we saw an increase in amiE transcription in the wildtype strain after chelation (Figure 2A). However, in the absence of the Fur protein (the Δfur strain), we saw no change in amiE transcription. Fifteen of the mutations identified via the manganese selection showed a phenotype similar to Δfur; they were not responsive to the change in iron availability. The other four Mn^R strains were responsive to iron, but to a lesser degree than the wildtype strain. Four of the five mutant strains obtained from the Pfr-based screen displayed a significant decrease in iron-responsiveness. However, only a single site-specific mutant strain, E117R, was significantly less responsive to iron chelation.

Figure 2.

Transcriptional regulation of amiE under iron replete and iron-depleted conditions as measured by qRT-PCR. (A) Fold increase in amiE transcript levels after chelation. For each strain, relative levels of transcript after chelation, where iron was depleted, were compared to levels prior to chelation, where iron was abundant (T₆₀/T_0). (B) Relative basal level of amiE transcript in each mutant strain under iron-replete conditions as compared to the wildtype strain (T₀/T₀). In both (A) and (B) the geometric mean of fold differences for at least three biologically independent experiments is shown as grey bars. Error bars indicate one standard deviation above and below the geometric mean. * indicates a statistically significant difference (P≤ 0.05, Student’s t test with a Holm-Sidak correction for multiple comparisons) when compared to the wildtype strain. WT, wildtype; Δfur, fur deletion strain, WT_C, Cm^R wildtype control strain.

Given that amiE is derepressed in the presence of iron (T₀) in the fur mutant, we also compared the basal levels of amiE expression in each of the mutant strains to that of the wildtype and Δfur strains. As shown in Figure 2B, all of the mutations identified in the manganese selection resulted in significant de-repression of amiE transcription even under iron replete conditions A similar trend was seen for the mutations obtained from the Pfr-based screen; the I52T, S60N, I67F, and Y89H mutations displayed significantly different basal amiE expression, whereas the other mutation (G40D) did not. Finally, E5A and E117R were the only single site- specific mutations that significantly de-repressed amiE expression in iron replete conditions. Based on all of the data shown in Figure 2, we classified each of the mutant strains into three possible categories in terms of Fe-Fur regulation: those with a wildtype phenotype, those with a Δfur phenotype, and those with an intermediate phenotype (Table 1): strains that showed a significant difference in regulation for the T₆₀/T₀ (Figure 2A) and the basal level (Figure 2B) comparisons were considered to be like Δfur, whereas strains that showed a significant difference in only one of the two comparisons were considered to have an intermediate (INT) phenotype. En masse, our results suggest that our mutagenesis strategy successfully identified Fur residues important for Fe-Fur dependent regulation.

Transcriptional analysis of pfr regulation

One of the unique characteristics of H. pylori Fur is the ability to repress gene expression in the absence of the regulatory iron cofactor. As such, we next sought to determine if any of our random and site-specific mutations altered regulation of the apo-Fur repressed gene pfr (Bereswill et al., 2000). To this end, we used qRT-PCR to compare the levels of pfr transcript before and after iron chelation. As an apo-Fur repressed gene, expression of pfr is decreased after chelation; thus, when iron is removed from the system, there is a decrease in pfr transcript levels. As shown in Figure 3A, in the wildtype strain chelation resulted in a decrease in pfr transcript from the T₀ to T₆₀ timepoint. As expected, this decrease was not seen in the Δfur strain. Of the 19 Mn^R mutant strains, only the H96L, V123D, and M138T mutations showed a significantly different fold decrease in response to chelation Of the five mutations identified in the Pfr-based screen, none resulted in a fold change significantly different than the wildtype strain. Of the site-specific mutations, only R13A resulted in a significantly lower fold change after chelation.

Figure 3.

Transcriptional regulation of pfr under iron replete and iron-depleted conditions as measured by qRT-PCR. (A) Fold decrease in pfr transcript levels after chelation. For each strain, relative levels of transcript after chelation, where iron was depleted, were compared to levels prior to chelation, where iron was abundant (T₆₀/T_0). (B) Relative post-chelation levels of pfr transcript (iron depleted conditions) in each mutant strain as compared to the wildtype strain (T₆₀/T₆₀). In both (A) and (B) the geometric mean of fold differences for three biologically independent experiments is shown as grey bars. Error bars indicate one standard deviation above and below the geometric mean. * indicates a statistically significant difference (P≤ 0.05, Student’s t test with Holm-Sidak correction for multiple comparisons) when compared to the wildtype strain. WT, wildtype; Δfur, fur deletion strain, WT_C, Cm^R wildtype control strain

In addition to the T₆₀/T₀ comparison, we also compared relative pfr transcript levels in each of the mutants to the wildtype strain after chelation (T₆₀), where expression should be repressed by apo-Fur. As shown in Figure 3B, in the fur deletion strain, pfr transcript levels were more abundant (de-repressed) relative to the wildtype strain. Based on this comparison, 16 Mn^R mutant strains displayed statistically significant differences in levels of pfr at the T₆₀ timepoint. Of the strains isolated in the Pfr-based screen, the S60N and Y89C mutants showed significant differences, and all five site-specific mutations resulted in significant differences in pfr levels at this time point.

Upon closer inspection of the pfr levels in the Mn^R strains, we noted that several mutations resulted in hyper-repression at this time point. Importantly, several of these strains carried mutations in or near a metal binding site. For example, the post-chelation levels of pfr transcript in the E90Q, E90G, A92T, A92V, K94N, K94R, H134Q, and H134Y mutant strains were significantly lower than the wildtype strain (Figure 3B), which suggests that apo-Fur is more abundant or more active in these strains. Similar to the amiE transcriptional analysis, we classified each of the mutant strains into four possible categories in terms of apo-Fur regulation: those with a wildtype phenotype, those with a hyper-repressed wildtype phenotype, those with a Δfur phenotype, and those with an intermediate phenotype (Table 1). Mutant strains were grouped as described for amiE regulation: strains that showed a significant difference in regulation for the T₆₀/T₀ (Figure 3A) and the post-chelation level (Figure 3B) comparisons were considered to be like Δfur, whereas strains that showed a significant difference in only one of the two comparisons were considered to have an intermediate (INT) phenotype. Taken together, our results suggest that our mutagenesis strategy successfully identified Fur residues important for apo-Fur dependent repression.

Fur Western blot analysis

One possible factor that could affect Fur-mediated transcriptional regulation is an increased or decreased amount of Fur protein in the cell. Given that Fur in H. pylori is autoregulatory, changes in the levels of Fur could occur as a result of altered transcriptional regulation at this promoter, or as the result of alterations in mutant Fur stability. To evaluate the levels of Fur in each of our mutant strains, we used an aliquot of liquid culture taken at the T₀ time point of the iron chelation experiment described above and performed Western blot analyses. For the majority of the mutant strains, the levels of Fur at this time point were similar to that of the wildtype strain (Figure 4). However, in eight of the mutant strains the level of Fur was either significantly greater than (I52T, S60N) or less than (G40N, L80V, K94R, K94N, I132N, M138T) the wildtype. We also noted that, although not statistically significant, the Fur levels in the S62T strain were also elevated, and appeared similar to Fur levels in the S60N mutant strain. Thus, in most cases the single amino acid change did not overwhelmingly affect levels of Fur.

Figure 4.

The relative levels of Fur in each mutant strain were measured by Western blot and compared to Fur levels in the wildtype. The geometric mean of fold difference values from three biologically independent experiments are shown by the grey bars, and the error bars indicate one standard deviation above and below the geometric mean. * indicates statistically significant differences (P≤ 0.05, Student’s t test with a Holm-Sidak correction for multiple comparisons); when compared to the wildtype strain; WT, wildtype; WT_C, Cm^R wildtype control strain

Purification of recombinant Fur proteins

In order to better understand at a molecular level how the mutations identified in this study affect the different aspects of Fur function, we selected five mutations to characterize in further detail. Based on the regulatory phenotypes described above, we selected the following mutations: E5A, which is located in the N-terminal extension and showed a intermediate (INT) phenotype for both amiE pfr regulation; A92T, which is located proximal to the S2 site and showed a Δfur-like phenotype for amiE and a hyper-repressive phenotype for pfr; E117R, which coordinates a metal ion in site S3 and showed a Δfur-like phenotype for amiE and an INT phenotype for pfr; V123D, which is located in the dimerization helix α5 and showed a Δfur-like regulatory phenotype for both amiE and pfr; H134Y, which is a metal coordinating residue in S3 and showed a Δfur-like phenotype for amiE and a hyper-repressive phenotype for pfr regulation. Recombinant expression strains were created for each of these mutations, and native (non-tagged) protein was purified. We were able to succesfully purify the HpFurWT, HpFurE5A, HpFurA92T, and HpFurH134Y proteins; however, we were unable to obtain soluble protein for the E117R and V123D mutations. Given that we did not see dramatic differences in the levels of Fur in these mutant strains (Figure 4), our inability to obtain soluble Fur for these mutants may be due to differences in protein solubility or folding in the heterologous E. coli strain. Because of this issue, E117R and V123D were not studied further.

Secondary structure content and thermal denaturation

Using circular dichroism (CD) spectroscopy, we determined the relative secondary structure content for the HpFurWT, HpFurE5A, HpFurA92T, and HpFurH134Y proteins. As shown in Figure 5A, the CD spectra of the HpFurWT, HpFurE5A, and HpFurH134Y proteins were highly similar, which indicated that the E5A and H134Y mutations did not dramatically alter the secondary structure content of these proteins. However, the spectra for the HpFurA92T protein did display a slight negative deviation at the lower end of the tested UV range. This shift highlighted the possibility that the HpFurA92T mutation altered the secondary structure content of the protein. To compare these changes in secondary structure content more quantitatively, we analyzed the relative differences in CD spectra using CDPro (Sreerama & Woody, 2000). In agreement with the raw CD secondary structure data, only the HpFurA92T mutant protein appeared to have notable differences in secondary structure content. This mutation resulted in a ~7% decrease in helical content, as well as a ~3% increase in both turn structure and unstructured regions. To further characterize any structural differences, we also performed thermal denaturation studies to determine whether or not these mutations altered gross protein folding. As shown in Figure 5B, the denaturation profiles for all three mutant proteins were highly similar to that of the WT protein and are indicative of a single unfolding event. Taken together, the CD data indicate that although the secondary structure content of the HpFurA92T protein is somewhat different than HpFurWT, none of the mutant proteins have notable folding defects. Thus, the regulatory phenotypes seen are most likely due to the effects of the mutations on some aspect of Fur function, rather than differences in overall protein fold.

Figure 5.

Secondary structure and thermal denaturation profiles obtained by circular dichroism (CD) spectroscopy. (A) CD spectra were obtained for HpFurWT and each mutant protein at room temperature from 190 nm to 250 nm. Each point represents an average of five accumulations. (B) Thermal denaturation studies were performed for HpFurWT and each mutant protein from 4°C to 100°C. Readings were taken at a wavelength of 222 nm.

Iron binding studies

To evaluate the effect of the E5A, A92T, and H134Y mutations on the ability of Fur to bind iron, we dialyzed each of the proteins against increasing amounts of iron (Fe²⁺) under anoxic conditions, and measured the amount of iron bound to each protein by atomic absorption spectrophotometry (Figure 6). As expected, based on previous iron binding studies (Carpenter et al., 2010), the HpFurWT protein bound three iron ions. Similar to the HpFurWT protein, the HpFurE5A mutant protein also bound three metal ions; however, this protein was able to bind three ions at approximately 8 μM iron, which was 2 μM lower than we observed for HpFurWT In contrast, both the HpFurA92T and HpFurH134Y mutant proteins displayed a clear reduction in iron binding. Even after dialysis against higher concentrations of iron (and up to 25 μM Fe²⁺, data not shown), binding never reached the level observed for the HpFurWT and HpFur E5A proteins. The fact that neither of these two mutant proteins ever averaged more than 1.5 iron ions per monomer suggests that these two mutations not only affected their respective metal binding sites (S2 adjacent to A92 and S3, which contains H134), but also may have had an effect on the adjacent metal binding sites as well. This phenomenon has also recently been reported for Fur in Bacillus subtilis (Ma et al., 2012).

Figure 6.

The ability of the HpFurWT and each mutant Fur protein to bind iron (Fe²⁺) as measured by atomic absorption spectrophotometry. Aliquots of each protein were dialyzed against increasing concentrations of Fe²⁺ (2.5 μM, 5 μM, 7.5 μM, or 10 μM) under anoxic conditions. Each data point represents the mean from at least three replicates and the error bars represent the standard deviation of the mean.

Oligomerization assays

Another critical facet of Fur regulation is the ability of the protein to form oligomers. As the structures of both holo-Fur and apo-Fur were reported as being dimeric, this characteristic is important for regulation by both forms of the protein (Butcher et al., 2012, Dian et al., 2011). To evaluate the ability of the HpFurE5A, HpFurA92T, and HpFurH134Y mutant proteins to form higher order structures in both the holo- and apo- forms, we employed in vitro cross-linking assays. Each protein was incubated in manganese-substituted (MnCl₂) or apo- binding buffer (Carpenter et al., 2010) with and without the crosslinking agent DSS; samples were separated by SDS-PAGE and protein complexes were visualized by SYPRO staining. As depicted in Figure 7A, in the presence of excess metal, all four proteins form higher order oligomers. Compared to the HpFurWT, none of the mutant proteins appeared to have a defect in oligomerization under this condition. Interestingly, in the absence of DSS, it appeared that H. pylori Fur preferentially exists as a mixture of monomers and oligomers that are larger than the dimer size. Thus, while the basic unit of Fur function may be the dimer, there is a propensity to form higher order oligomers under the conditions tested here. In contrast, by adding the crosslinking agent, we were able to trap additional oligomeric forms of Fur. In addition to monomers and higher order oligomers, we saw the presence of Fur dimers. These data indicate that in the presence of excess metal the equilibrium seems to be between monomers and oligomers. While the formation of these oligomers required the presence of metal, this process does not appear to be DNA dependent; the addition of excess Fe-Fur target DNA (the amiE promoter) did not visibly alter oligomerization in the presence of excess metal (data not shown).

Figure 7.

The ability of HpFurWT and each mutant protein to oligomerize as determined by in vitro cross-linking. Oligomerization in iron-substituted conditions (A) or apo (B) conditions were performed in the presence (+) or absence (−) of the crosslinking reagent DSS. Protein complexes were separated by SDS-PAGE and stained with SYPRO Ruby Red. Fur monomers, dimers, and > dimer complexes are indicated by arrows. Predicted molecular weights for Fur complexes are shown.

The recently published structure of C. jejuni apo-Fur (Butcher et al., 2012) gave the first crystallographic evidence that this form of the protein existed as a dimer. We next tested the HpFurWT and mutant Fur proteins for their ability to oligomerize in apo- conditions. At equilibrium, apo-Fur appears to exist as a mixture of monomeric and dimeric forms (Figure 7B). In contrast to the excess metal conditions, WT apo-Fur does not seem to form a significant number of oligomers larger than the dimer. Thus, it is likely that whatever region of Fur that is responsible for the formation of higher order oligomers (larger than dimer) is not accessible in the apo- form of the protein. Similar to the observation made during the manganese-substituted Fur oligomerization assays, the addition of target DNA (pfr promoter sequence) did not alter the oligomeric form of the protein (data not shown). Compared to the HpFurWT protein, the only mutant that displayed a difference in oligomerization in apo- conditions was the HpFurE5A protein. In the presence of the crosslinking agent, we were able to faintly detect the presence of a slghtly increased proportion of higher order oligomers that were larger than dimer size (Figure S3).

DNA binding assays

Finally, we evaluated the ability of the HpFurWT and mutant Fur proteins to bind DNA in manganese-substituted and apo- conditions using fluorescence anisotropy. A summary of the best-fit anisotropy data is shown in Table 2. Under manganese-substituted conditions, the HpFurWT protein bound the amiE oligo duplex with a lower affinity than two of the mutant proteins and with approximately the same affinity as the other mutant (Figure 8A). The order of binding affinity under these conditions was HpFurE5A > HpFurH134Y > HpFurA92T > HpFurWT, with K_d values of 2 ^± 1 nM, 6 ^± 1 nM, 15 ^± 2 nM, and 16 ^± 2 nM, respectively. Based on these results, it appears that each of these residues plays a role in holo-Fur interaction with DNA. To ensure that this interaction was specific, we also performed competition experiments using an unlabeled DNA duplex where the Fur box sequence was mutated to G-T repeats. For these experiments the HpFurWT and HpFurE5A proteins did not bind the scrambled amiE promoter DNA sequence to a significant degree, while the HpFurA92T and HpFurH134Y proteins did bind the scrambled template to some extent (Figure S4).

Table 2.

Summary of Fluorescence anisotropy data

Oligo Fragment	Fur protein	Kd (+ error)
amiE; manganese-substituted conditions	HpFurWT	16 (2) nM
	HpFurE5A	2 (1) nM
	HpFurA92T	15 (2) nM
	HpFurH134Y	6 (1) nM
pfr; apo conditions	HpFurWT	43 (2) nM
	HpFurE5A	19 (2) nM
	HpFurA92T	13 (1) nM
	HpFurH134Y	8 (1) nM

Each titration was performed using 50 nM labeled ligand (Ltot=50 nM); Kd values were rounded to the nearest whole number. R-squared values for all goodness-of-fit curves were ≥ 0.998.

Figure 8.

The ability of the HpFurWT and mutant proteins to bind DNA sequences from the amiE (A) and pfr (B) promoters as determined using fluorescence anisotropy. Forward titrations were performed using 50 nM of labeled oligo duplex in either iron-substituted (for amiE) or apo (for pfr) conditions. In both (A) and (B), the curves represent the best fit line and the concentration of Fur dimer (nM) is shown.

Next, we performed similar experiments using the high affinity Fur binding sequence from the pfr promoter and apo- binding buffer. Similar to the manganese-substituted DNA binding assays, the relative binding affinity of the mutant proteins was higher than HpFurWT (Table 2 and Figure 8B). Under apo conditions the order of binding affinity for the pfr promoter sequence was HpFurH134 > HpFurA92T > HpFurE5A > HpFurWT, with K_d values of 8 ^± 1 nM, 13 ^± 1 nM, 19 ^± 2 nM, and 43 ^± 2 nM, respectively. Once again, we performed competition experiments with an unlabeled DNA oligo duplex that had the putative apo-Fur binding site mutated to G-T repeats. Under apo conditions, none of the proteins bound to the unlabeled scrambled pfr promoter DNA, which suggests that the binding seen with the wildtype pfr promoter sequence was specific (data not shown). Consequently, the binding promiscuity observed for the HpFurA92T and HpFurH134Y (Figure S4) appears to be specific to the fully metallated form of the protein.

Discussion

As highlighted by its importance in vivo, Fur mediated regulation is a key component of stress adaptation and gene regulation in H. pylori (Bereswill et al., 2000, Carpenter et al., 2009, Cooksley et al., 2003, Delany et al., 2001b, Ernst et al., 2005a, Ernst et al., 2005b, Gancz et al., 2006, Pich et al., 2012, Qu et al., 2009, van Vliet et al., 2003). The degree of functional diversity displayed by this pleotropic regulator is thus far unique. Although Fe-Fur activation and apo-Fur regulation have also been reported in a few other bacteria (Delany, 2004, Deng et al., 2012, Grabowska et al., 2011), no other bacterial Fur proteins perform all four types of regulation: Fe-Fur repression, Fe-Fur activation, apo-Fur activation and apo-Fur repression. As a result, functional information about H. pylori Fur regulation is not only important for understanding regulation in this organism, but also for defining the range of functional capabilities of the Fur family. A previous site-specific mutagenesis study targeted residues that are conserved in several well-studied bacterial species that had been shown to be critical for Fur function in those organisms (Carpenter et al., 2010). However, unlike those studies, mutations in the same conserved residues in H. pylori Fur did not render the protein completely non-functional. These findings highlight the unique nature of H. pylori Fur, and suggested to us that additional methods were required to identify key functional residues in this protein.

In the current study, we employed random and site-specific mutagenesis to identify residues important for Fe-Fur and apo-Fur repression. We hypothesized that this type of functional information would provide insight into the contribution of the various regions of the protein to both types of regulation. Using manganese selection, we identified 19 mutations in 15 different residues that conferred a Mn^R phenotype (Figure 1 and Table 1). Most of these 15 residues were located in the C-terminal half of the protein, which contains the majority of the metal coordinating residues. Due to the role of these residues in metal coordination, it is perhaps not surprising that single amino acid changes in this region resulted in a strong Fe-Fur phenotype. However, it is interesting to note that in other manganese selection studies, the distribution of mutations imparting manganese resistance is quite different. For example, Lam et al., isolated 11 independent Mn^R strains of V. cholerae that contained mutations in 10 different amino acid residues (Lam et al., 1994). In contrast to our findings, those mutations were predominantly located in the N-terminal half of V. cholerae Fur, which contains the DNA binding domain of the protein. H. pylori Fur and V. cholerae Fur share 32% identity and 55% similarity (Bereswill et al., 1998), thus, the reason for this difference is not completely apparent at the amino acid sequence level. Instead, the difference is likely due to additional levels of regulation seen only with H. pylori Fur; the N-terminal domain of H. pylori Fur likely undergoes a fairly dramatic conformational change in order to function in the iron-bound or apo- form. Because of this requirement for “flexibility,” this region of the protein is likely more permissible to secondary structure perturbation or slight differences in conformation compared to other Fur species.

Despite the apparent plasticity within the N-terminal region of H. pylori Fur, by combining multiple mutagenesis strategies, we were able to identify important residues across virtually the entire coding region (Figure 1). Within the N-terminal extension, which is not seen in other Fur species, we specifically mutated residues R3, E5, E8 and R13 (Figure 1 and Figure 9). Of these four, only the E5A mutation affected both iron-bound and apo-Fur regulation (complete phenotype summarized in Table 3). Based on the structural description of the N-terminal extension (Dian et al., 2011), it is possible that this mutation may have shifted the hydrogen bonding network that links these N-terminal residues with residues that coordinate metal binding at S2; Dian et al., also proposed a similar hypothesis regarding the biological role for these residues (Dian et al., 2011). If this were the case, then we would anticipate that the resulting redistribution of hydrogen bonds could have potentially shifted the coordination sphere of S2, and increased the affinity for the regulatory metal ion. However, as the iron-binding measurements obtained here are not a true measure of binding affinity per se, further studies will be required to test this hypothesis. Consistent with this line of reasoning, a change in binding affinity could also explain the increased abundance of HpFurE5A higher order oligomers in the apo- cross-linking assay (Figure 7 and S3); given a higher affinity for metal, S2 would be less responsive to chelation. As a result, a portion of the E5A protein might remain as holo-Fur, which readily forms higher order oligomers. Within the bacterial cell, this shift in the equilibrium between Fe-Fur and apo-Fur, where more Fe-Fur is present even after metal chelation, could also explain the defect in apo-Fur repressor activity as measured by qRT-PCR (Figure 3B).

Figure 9.

Structural representation of H. pylori holo-Fur (A) and apo-Fur (B) dimers. Residues mutated in this study are highlighted as follows: N-terminal residues 3, 5, 8, and 13 are shown in green; residues 40, 52, 60, 63, and 67 within helices α2- α4 are shown in blue; β1-β2 residues 80, 87, and 89 are shown in magenta; S2 residues 90, 92, 94, 108, and 110 are highlighted in red; S3 residues 96, 117, 132, 134, 135, and 138 are yellow; residue 123 in α5 is highlighted in cyan. From left to right, the structures are rotated as indicated. The holo-Fur dimer in (A) is based on the structure reported by Dian et al., (Dian et al., 2011). The apo-Fur dimer in (B) was generated by modeling the H. pylori Fur amino acid sequence over the C. jejuni apo-Fur structure (Butcher et al., 2012). Metal binding sites S1, S2, and S3, as well as residues E5, A92, and H134 are labeled. Images were generated using the Polyview-3D server: http://polyview.cchmc.org/polyview3d.html (Porollo & Meller, 2007, Porollo et al., 2004).

Table 3.

Summary of mutant protein characterization

ASSAY	HpFurE5A	HpFurA92T	HpFurH134Y
amiE qRT	INT	Δfur	Δfur
pfr qRT	INT	INT*	INT*
amiE FA	>WT	WT	>WT
pfr FA	>WT	>WT	>WT
Iron binding	WT	<WT	<WT
MnCl2 oligomerization	WT	WT	WT
apo-oligomerization	WT	WT	WT

WT, phenotype similar to wildtype; Δfur, phenotype similar to fur deletion strain; INT, intermediate phenotype;

^*indicates hyper-repression at T60; >WT, greater than the wildtype protein; <WT, less than the wildtype protein. Summary of phenotypes is based on data in Table 1 and Figures 6, 7, and 8.

Interestingly, the relative binding affinity of HpFurE5A for the apo-Fur target sequence was also higher than the HpFurWT protein. Based on the structure of apo-Fur in C. jejuni, we propose a possible explanation for increased binding under these conditions. In this structure, helix α1is positioned within the “V” of the DNA binding domain. Given this position, it is possible that altering this residue directly affects how H. pylori apo-Fur interacts with target DNA and causes the protein to bind with higher affinity than wildtype apo-Fur. Of note, another H. pylori metalloregulator, the nickel-responsive regulator NikR, also contains a unique N-terminus that, when disrupted, affects DNA binding (Benanti & Chivers, 2007). Thus, it is possible that acquisition of additional N-terminal residues that function to mediate specific DNA interactions is a common mechanism for extending the function of regulatory proteins in H. pylori.

Based on these studies collectively, we propose that in Fe-Fur/holo-Fur, the E5 residue serves as an anchor point for establishing the S2-stabilizing hydrogen bond network. Conversely, in the apo form of Fur where helix α1 is positioned to interact with DNA, we propose that E5 serves as a key residue within this helix to facilitate interaction with DNA. The role for this extension in apo-Fur function is supported by the fact that all four mutations in this region affected pfr regulation to some degree. Thus, the unique N-terminal extension appears to play a crucial multi-factorial role in Fur function.

Adjacent to the N-terminal extension are helices α2-α4, which make up the winged helix- turn-helix motif that is involved in DNA binding (Dian et al., 2011). We identified a single mutation in this region using manganese selection (S62T), but identified four additional mutations (G40D, I52T, S60N, and I67F) using the Pfr-based screen (Figures 1 and 9). Despite being identified in the Pfr-based screen, the G40N mutation did not significantly alter amiE or pfr regulation at the transcriptional level. The I52T and I67F mutations showed similar phenotype in regulation of amiE; neither mutant strain was able to fully repress amiE under iron replete conditions. Based on the location of these residues, we propose that this defect could be the result of changes in DNA binding affinity for Fe-Fur target DNA. Compared to amiE regulation, the regulation of pfr in the I52T and I67F mutant strains more closely resembled that of the wildtype strain. The final two mutations located within the α2-α4 region are the S60N and S62T mutations. Despite the close proximity of these two residues and the fact that the levels of Fur were ~3–4 fold higher in these strains as compared to the wildtype (Figure 4), the regulatory phenotypes conferred by these two mutations are quite different. Because the basal levels of amiE transcript were increased in the S60N mutant strain and there was no change in these levels after chelation, we hypothesize that the S60N mutation alters the ability of Fe-Fur to bind to DNA and repress transcription. This line of reasoning is consistent with the increased levels of Fur protein in this strain; fur expression is autoregulated (repressed by Fe-Fur) so a lack of repression at this promoter would result in higher protein levels. Similarly, the S60N mutation also seems to prevent proper repression of pfr expression, as demonstrated by the increased levels of pfr transcript after chelation. Conversely, in the S62T mutant strain the basal levels of amiE transcript were lower than wildtype, but this strain showed a lower degree of de-repression after chelation. Thus, the S62T mutation may have increased binding affinity for the amiE promoter. While this explanation may seem to be contradicted by the higher levels of Fur protein seen in the S62T strain (i.e. fur should be repressed similar to amiE) it is possible that the location of S62T within the DNA binding helix α4 confers differences in binding affinity to one promoter but not to another. In addition, the nucleotide composition and location of the Fur binding sequence(s) in these two promoters is different, which can directly affect Fur-dependent regulation (Pich et al., 2012).

At the distal end of the DNA binding domain, we isolated mutations in sheets β1-β2 (Figure 9). These mutations (L80V, R87Q, Y89C, and Y89H) all displayed similar regulatory phenotypes and levels of Fur. The basal levels of amiE transcript in these strains were higher that the wildtype stain and none were responsive to chelation. Since these four strains were able to regulate pfr expression to nearly the same extent as the wildtype strain, it appears that these residues are most important for iron-bound regulation. Given the location of these two sheets near the hinge region that connects the DNA binding domain to the C-terminal domain of the protein, it is possible that these residues may be more important for maintaining the orientation of the DNA binding domain in Fe-Fur, but not necessarily for the apo form of the protein.

The S2 metal binding site senses intracellular iron levels, and the presence or absence of a metal ion in this site distinguishes between the holo- and apo forms of Fur. Using the manganese selection, we obtained mutations in several residues in this region (Figures 1 and 9). Given their location in relation to S2, it is perhaps not surprising that all eight of these mutations (E90Q, E90G, A92T, A92V, K94R, K94N, I108N, and E110D) affected Fur regulation in a similar manner. As a general theme, the basal level of amiE transcript in these eight mutant strains was increased compared to the wildtype strain, and metal chelation had little effect on transcript levels (Figure 2). Conversely, the mutated versions of Fur in these strains were able to repress pfr expression to a higher degree than wildtype. Comparison of the post-chelation pfr levels in this group of mutant strains revealed that mutations in this region of Fur have a tendency to confer a hyper-repressive phenotype. Clues to possible mechanisms for these regulatory defects were provided by the biochemical analysis of the purified HpFurA92T mutant protein as a representative of this mutant group (summarized in Table 3). The HpFurA92T mutant displayed a significant defect in iron-binding; the protein was only able to bind 1–1.5 metal ions per monomer. Although the A92 residue is not directly involved in coordinating the metal ion in S2, the close proximity of this residue to this site combined with the change in side chain (alanine to threonine) clearly affected the ability of metal to occupy this site. Furthermore, the atomic absorption data indicate that this mutation may affect metallation at S3 as well. This idea is supported by CD spectroscopy (Figure 5A), which detected several changes in overall secondary structure, and the oligomerization assays (Figure 7). As metallation at S1 is required for dimer formation (Dian et al., 2011, Vitale et al., 2009), the fact that the HpFurA92T protein was able to oligomerize in both the manganese-substituted and apo forms strongly suggests that the metal ion(s) detected by atomic absorption spectrophotometry occupy S1. On a cellular level, the dramatic reduction in iron binding also explains the constitutive de-repression of amiE as well as the hyper-repression of pfr that was observed by qRT-PCR (Figures 2 and 3). Because of the defect in iron binding, the vast majority of Fur in the cell at a given time exists as apo-Fur. As a result, amiE is not repressed even under iron replete conditions (T₀), and higher proportion of apo-Fur in the cell results in hyper-repression of pfr. Finally, the DNA binding assays provided additional clues as to how S2 and the residues proximal to this site affect Fur function. Based on the iron-binding and transcriptional data, we expected that the HpFurA92T protein would also be defective in binding DNA under manganese-substituted conditions. Instead, the HpFurA92T protein bound to the amiE oligo duplex with nearly identical affinity as the HpFurWT protein (Table 2). However, since this protein also bound to the Fur box scramble sequence to some degree (Figure S4), this interaction with the amiE promoter fragment is likely non-specific. Thus, this mutation seems to have rendered the HpFurA92T mutant protein in a “locked” conformation that allows for non-specific interaction with DNA. This promiscuous binding is specific for iron-substituted conditions, as the HpFurA92T protein did not bind the scrambled pfr promoter sequence in apo conditions (data not shown). Consistent with the hyper-repression of pfr observed via qRT-PCR, the binding affinity for the pfr oligo duplex was higher than the HpFurWT protein (Table 2). Taken together, the analysis of the mutations found in and around S2 indicate that in addition to the metal coordinating residues, the neighboring amino acids such as A92 and K94 are critical for maintaining the proper coordination sphere at S2, and may also indirectly impact metallation at S3.

S3 serves as a structural metal binding site and is occupied in both Fe-Fur and apo-Fur. We obtained several mutations in residues that lie within (H96L, E117R, H134Q, H134Y) or directly adjacent to this region (I132N, D135Y, and M138T; Figure 9). These six mutations in or near S3 shared similar regulatory phenotypes. Under iron replete conditions, the basal levels of amiE were higher in these mutant strains than the wildtype, and there was no major change in transcript levels after metal chelation. Thus, these strains had a Δfur-like phenotype for amiE regulation. As a group, these mutations had less of an effect on pfr regulation. Most of these strains displayed a wildtype-like fold decrease in pfr transcript after chelation (with the exception of M138T), and the relative levels of pfr after chelation were equal to (I132L) or significantly less than the wildtype strain (e.g. H96L, H134Q, H134Y, D135Y). These findings highlight the role of S3 in apo-Fur function, and suggest that these residues are involved in proper metal ion coordination in this form of the protein. The HpFurH134Y protein displayed a defect in the ability to respond to chelation. In vitro analysis of this mutant protein suggested a mechanistic basis for the regulatory phenotypes observed by qRT-PCR (summarized in Table 3). As H134 directly interacts with the S3 metal ion in holo-Fur, it was not surprising that this mutant protein displayed a significant defect in iron binding. Similar to the HpFurA92T mutant, HpFurH134Y only bound 1–1.5 iron ions per monomer (Figure 6). Once again, it appears that mutation of this residue affects metal binding at other sites. Given the lack of a distinct oligomerization defect in either manganese-substituted or apo conditions (Figure 7), it would appear that this effect on metal binding influences S2 rather than S1. Given the apparent requirement for metallation at S3 for both holo-Fur and apo-Fur, this defect in metal binding could explain the lack of amiE repression as well as the slight defect in pfr regulation. Based on the DNA binding assays performed in both manganese-substituted and apo conditions, the HpFurH134Y protein displayed a higher affinity for DNA than HpFurWT (Figure 8). Not only did this mutant protein bind to the amiE and pfr oligo duplexes with higher affinity, HpFurH134Y also bound to the scrambled amiE oligo duplex to some degree (Figure S4). These data suggest that HpFurH134Y may have a propensity to bind DNA non-specifically in manganese-substituted conditions. These findings are consistent with a previous study that reported non-specific DNA binding of a HpFurH134A Fur mutant (Carpenter et al., 2010). Thus, under manganese-substituted conditions, mutation of this residue may result in a form of the protein that is “locked” in a position that is permissive for DNA binding. While a relatively higher binding affinity was seen with the wildtype version of the pfr promoter oligo duplex, the promiscuous binding was not observed under apo conditions. In the context of in vivo regulation of the amiE gene, the higher binding affinity seen with the HpFurH134Y mutant in vitro cannot be clearly explained. However, it is important to consider the fact that this isolated in vitro assay may not necessarily translate directly to the way things work within a functional cell. For instance, the DNA binding assays do not take into account the multiple types of metal (Fe²⁺and Zn²⁺) that are present in vivo.

We obtained a mutation in one additional residue within the C-terminal domain, V123D, which is located within helix α5 (Figures 1 and 9). This helix serves as an interface for dimerization in both holo- and apo-Fur. As dimerization is important for both types of Fur regulation, it is perhaps not surprising that this mutation resulted in a regulatory phenotype for both amiE and pfr (Figures 2 and 3). Based on the location of this residue, we hypothesize that this mutation precludes proper dimerization; however, as we were unable to purify the HpFurV123D protein due to issues with solubility in the heterologous E. coli expression strain, we were unable to test this hypothesis in further detail.

One surprising outcome of this study was the fact that no S1 mutations were obtained in either the manganese selection or Pfr-based screen. Given the importance of this site in dimerization (presumably for both types of Fur), it seems that a mutation in this metal binding site would abrogate Fur function to a similar degree as mutations in either S2 or S3 (both of which were obtained by manganese selection). One possible explanation is that because of the essential nature of S1 to proper function, minor structural disruptions in this region may not exert a strong effect; thus, single amino acid substitutions may not be sufficient to completely eliminate metal binding in S1 and confer a Mn^R phenotype, or visual changes in Pfr protein levels.

En masse, the in vivo and in vitro studies detailed herein shed new light on functional features of Fe-Fur and apo-Fur regulation in H. pylori. When interpreted in the context of the H. pylori holo-Fur (Figure 9) and closely related C. jejuni apo-Fur structures (Figures 9 and S1), we can begin to piece together how specific residues or regions of Fur contribute to each type of regulation and how some of these residues contribute to different aspects of function in the iron-bound and apo forms of the protein. For example, as proposed by Dian et al., (Dian et al., 2011), the N-terminal extension apparently plays a role in facilitating metal coordination at S2; furthermore, when interpreted in the context of the C. jejuni apo-Fur structure (Figure S1), it is likely that this region is directly involved in binding to apo-Fur target DNA. Based on our analysis of the A92T and H134Y mutations, it also appears that maintenance of the S2 site is important for proper metallation at S3, and vice versa. In combination with this and other functional studies, we are now poised to address some of the many questions that still remain. For example, it is clear that the N-terminal extension contributes to metal coordination at S2; however, the evolutionary basis of this feature remains a mystery. Did this region evolve to facilitate apo-Fur interaction with DNA? Does the hydrogen bonding network that this region forms in Fe-Fur provide metal ion selectivity in vivo? While these and many other questions still remain, it is clear that Fur in H. pylori, and likely its close relatives, has many unique features that contribute to the regulatory diversity displayed by these organisms.

Materials and Methods

Strains and growth conditions

A complete list of bacterial strains and plasmids used in this study can be found in Table 4. H. pylori strains were maintained at −80°C in brain heart infusion broth that contained 10% fetal bovine serum and 20% glycerol. E. coli strains were maintained at −80°C in LB broth supplemented with 40% glycerol. H. pylori strains were cultured on horse blood agar (HBA) that contained 4% Columbia agar, 5% defibrinated horse blood (HemoStat Laboratories, Dixon, CA), 0.2% β-cyclodextrin, 10 μg/mL vancomycin, 8 μg/mL amphotericin B, 2.5 U/mL polymyxin B, and 5 μg/mL trimethoprim. Liquid H. pylori cultures were routinely grown in Brucella broth that contained 10% fetal bovine serum and 10 μg/mL vancomycin with shaking at 110 rpm. All H. pylori cultures were grown under microaerophilic conditions (5% oxygen, 10 % carbon dioxide, and 85% nitrogen) in gas evacuation jars. E. coli cultures were grown on LB agar plates or in LB broth with shaking at 220 rpm. Bacterial cultures were supplemented with appropriate antibiotics as necessary: ampicillin (Amp), 100 μg/mL; kanamycin (Kan), 50 μg/mL; chloramphenicol (Cm), 50 μg/mL for E. coli or 8 μg/mL for H. pylori.

Table 4.

Strains and plasmids used in this study

Strains and Plasmids
Strain	Description	Reference
E. coli
DSM431	BL21 (DE3) Rosetta/pLys Δfur (pDSM430) AmpR, KanR, CmR	(Carpenter et al., 2009)
DSM1197	BL21 (DE3) Rosetta/pLys Δfur (pDSM1191) AmpR, KanR, CmR	This study
DSM1199	BL21 (DE3) Rosetta/pLys Δfur (pDSM11193) AmpR, KanR, CmR	This study
DSM1200	BL21 (DE3) Rosetta/pLys Δfur (pDSM1194) AmpR, KanR, CmR	This study
H. pylori
DSM1	G27 WT	(Baltrus et al., 2009)
DSM300	G27 Δfur::cat CmR	(Gancz et al., 2006)
DSM948	G27 Fur R3A; CmR	This study
DSM949	G27 Fur E5A; CmR	This study
DSM950	G27 Fur E8A; CmR	This study
DSM951	G27 Fur R13A; CmR	This study
DSM953	G27 Fur E117R; CmR	This study
DSM954	G27 Fur A92V; CmR	This study
DSM955	G27 Fur A92T; CmR	This study
DSM956	G27 Fur K94N; CmR	This study
DSM957	G27 Fur K94R; CmR	This study
DSM958	G27 Fur H134Q; CmR	This study
DSM959	G27 Fur H134Y; CmR	This study
DSM960	G27 Fur E90Q; CmR	This study
DSM961	G27 Fur E90G; CmR	This study
DSM962	G27 Fur H96L; CmR	This study
DSM963	G27 Fur I108L; CmR	This study
DSM964	G27 Fur D135Y; CmR	This study
DSM965	G27 Fur S62T; CmR	This study
DSM966	G27 Fur M138T; CmR	This study
DSM967	G27 Fur R87Q; CmR	This study
DSM968	G27 Fur I132L; CmR	This study
DSM969	G27 Fur L80V; CmR	This study
DSM970	G27 Fur Y89C; CmR	This study
DSM971	G27 Fur V123D; CmR	This study
DSM972	G27 Fur WT_C; CmR	This study
DSM973	G27 Fur E110R; CmR	This study
DSM1133	G27 Fur S60N; CmR	This study
DSM1134	G27 Fur I52T; CmR	This study
DSM1135	G27 Fur Y89H; CmR	This study
DSM1136	G27 Fur I67F; CmR	This study
DSM1137	G27 Fur G40D; CmR	This study
Plasmid
pDSM430	pET21b::HpFurWT; AmpR	(Carpenter et al., 2009)
pDSM1191	pET21b::HpFurE5A; AmpR	This study
pDSM1193	pET21b::HpFurA92T; AmpR	This study
pDSM1194	pET21b::HpFurH134Y; AmpR	This study

Amp^R, ampicillin resistant; Kan^R kanamycin resistant; Cm^R chloramphenicol resistant

Random mutagenesis of the H. pylori fur gene

A library of random fur mutations was generated by error-prone PCR using the GeneMorph II Random Mutagenesis Kit (Stratagene) and primers upFur/ATG-F and Fur/cat-R (Table 5). Each PCR was performed in accordance with the manufacturer’s instructions so that the resulting amplicons contained 0–2 nucleotide changes per copy. The randomly mutated fur amplicons were used to create chloramphenicol resistant Cm^R allelic exchange substrates as follows: the promoter containing upstream region of fur was PCR amplified using the Fur-F and upFur/ATG-R primers (Table 5).; the cat cassette was amplified from strain DSM300 using the Fur/Cat-F and Cat/dnFur-R primers; the downstream region was amplified using the Cat dn/Fur-F and dnFur-R primers (Table 5). Amplicons that contained the fur promoter region, mutated coding sequence, cat cassette, and downstream region of fur were fused together using splicing by overlap extension (SOE) PCR. Briefly, in one SOE PCR, amplicons that contained the region upstream of the fur coding sequence were spliced to amplicons that contained the mutated fur ORF; in a second SOE reaction, the cat cassette was spliced to the region immediately downstream of fur. The spliced products from each of these two SOE reactions were subsequently spliced together in a third SOE reaction. The resulting linear allelic exchange substrates were naturally transformed into the wildtype H. pylori strain G27. Transformants were selected for on HBA plates that contained 8 μg/mL of chloramphenicol. To ensure we obtained the desired mutation rate, we sequenced the fur gene from several transformants and determined that each copy of fur did in fact contain between 0–2 randomly distributed nucleotide changes. Approximately 10,000 Cm^R transformants obtained through numerous independent transformations were pooled and used for the selection and screening procedures described below.

Table 5.

Oligonucleotides used in this study

Oligonucleotidea	Sequence (5′-3′)b	Reference
SOE PCR
Fur-F	AAGGCTCACTCTACCCTATTCC	This study
upFur/ATG-R	CCAAAGTTTCTAATCTTTTCATGCTGATATTTCCCTTATCCG	This study
upFur/ATG-F	CGGATAAGGGAAATATCAGCATGAAAAGATTAGAAACTTTGG	This study
Fur/cat-F	GAATGCCAAGAGAGTGAATATTAACGCACTACTCTCGACAGAGAG	This study
Fur/cat-R	CTCTCTGTCGAGAGTAGTGCGTTAATATTCACTCTCTTGGCATTC	This study
cat/dnFur-R	CTAAGCTTCTCCTTTAAAATTTTGATAGATTTATGATATAGTGG	This study
Cat/dnFur-F	CCACTATATCATAAATCTATCAAAATTTTAAAGGAGAAGCTTAG	This study
dnFur-R	GGTATTTCTGATGTGGATGG	This study
FurCF (XbaI)	TCTAGAAAGGCTCACTCTACCCTATT	(Carpenter et al., 2007)
FurCR (SalI)	GTCGACAAFACTTTCACCTGGAAACGC	(Carpenter et al., 2007)
R3A-F	GGGAAATATCAGCATGAAAGCCTTAGAAACTTTGG	This study
R3A-R	CCAAAGTTTCTAAGGCTTTCATGCTGATATTTCCC	This study
E5A-F	FCAGCATGAAAAGATTAGCTACTTTGGAATC	This study
E5A-R	GATTCCAAAGTAGCTAATCTTTTCATGCTG	This study
E8A-F	GATTAGAAACTTTGGCTTCTATTTTAGAGC	This study
E8A-R	GCTCTAAAATAGAAGCCAAAGTTTCTAATC	This study
R13A-F	CTATTTTAGAGGCTTTGAGGATG	This study
R13A-R	CATCCTCAAAGCCTCTAAAATAG	This study
E117R-F	CCTGAAATTCGGAACCGCCAGAATG	This study
E117R-R	CATTCTGGCGGTTCCGAATTTCAGG	This study
Real-time PCR
G27_16S RT-F	ATGGATGCTAGTTGTTGGAGGGCT	(Gilbreath et al., 2012)
G27_16S RT-R	TTAAACCACATGCTCCACCGCTTG	(Gilbreath et al., 2012)
amiE_RT-F	GCGATGATGGAAACTACCC	This study
amiE_RT-R	CCCACGCCATAGCTTTTAC	This study
pfr_RT-F	TGAGCATGAGTTCTTGGTGC	This study
pfr_RT-R	AGTCAATTGCACAGGCAC	This study
Fur expression
HPFur_expression-F (NdeI)	CATATGAAAAGATTAGAAACTTTGGAATCTATTT	(Carpenter et al., 2009)
HPFur_expression-R (XhoI)	CTCGAGTTATTAATATTCACTCTCTTGG	(Carpenter et al., 2009)
E5A-expression-F (NdeI)	GCTACCATATGAAAAGATTAGCTACTTTGGAATCCATTTT	This study
Fluorscence anisotropy
amiE-FA_LABEL	5FluorT/GCCCCAATAATCATAATGATTAA	This study
amiE-FA_COMPLEMENT	TTAATCATTATGATTATTGGGGCA	This study
amiE-FA_SCRAM-1	TGCCCCGTGTGTGTGTGTGTATTAA	This study
amiE-FA_SCRAM-2	TTAATACACACACACACACGGGGCA	This study
pfr-FA_LABEL	5FluorT/TCATTATCATTTATGCTAT	This study
pfr-FA_COMPLEMENT	ATAGCATAAATGATAATGAA	This study
pfr-FA_SCRAM-1	TTCATTAGTGTGTATGCTAT	This study
pfr-FA_SCRAM-2	ATAGCATACACACTAATGAA	This study

^aImportant restriction enzyme recognition sites shown in parentheses

^bRestriction sites are underlined, mutated codons are in bold italics, and start or stop codons are italicized

Site-specific Fur mutant strains

The R3A, E5A, E8A, R13A, and E117R mutations were created by SOE PCR using the mutation-specific SOE primers listed in Table 5. Briefly, the FurCF primer was used in combination with each of the mutation-specific reverse primers, and FurCR was used in combination with each of the mutation specific forward primers. These two fragments were spliced together and used to create Cm^R allelic exchange substrates by SOE PCR as described above. Each mutation-bearing substrate was naturally transformed into wildtype G27 and plated on HBA that contained 8 μg/mL chloramphenicol. Proper integration was confirmed by PCR and the presence of the correct mutation was confirmed by sequencing. The fur promoter region in each strain was also sequenced to ensure that there were no mutations present.

Manganese selection

Manganese selection has been previously used to identify residues that are important for Fur function (Hantke, 1987, Lam et al., 1994, Loprasert et al., 2000). Aliquots of the random fur mutant library were diluted and plated for single colonies on Brucella broth-based agar plates that contained 10% fetal bovine serum, 10 μg/mL vancomycin, 8 μg/mL amphotericin B, 2.5 U/mL polymyxin B, 5 μg/mL trimethoprim, and 4 mM MnCl₂.

Screen for altered Pfr expression

The H. pylori prokaryotic ferritin (Pfr) has been previously shown to be apo-Fur repressed (Bereswill et al., 2000). Based on the ability to easily visualize Pfr in a stained protein gel, we designed a Pfr-based screen to identify fur mutant strains that were deficient in apo- repression. Aliquots of the fur mutant strain library were diluted and plated on Brucella broth-based plates that contained 10% fetal bovine serum, 10 μg/mL vancomycin, 8 μg/mL amphotericin B, 2.5 U/mL polymyxin B, 5 μg/mL trimethoprim, and 60 μM of the iron chelator 2,2′-dipyridyl (dpp). This concentration of dpp is sufficient to create an iron-depleted condition without completely prohibiting bacterial growth (Carpenter et al., 2007). Single colonies were expanded on chelation plates and used to make whole-cell lysates. An equal volume of each lysate was separated via SDS-PAGE and analyzed after staining with Coomassie. The levels of Pfr in each of the random mutant strains were visually compared to both the wildtype and fur mutant strains of G27.

RNA isolation, cDNA synthesis, and quantitative real-time PCR

RNA was harvested from liquid cultures under iron-replete and iron depleted conditions. Briefly, an overnight culture of each strain was grown to exponential phase; one half of each culture was harvested by vacuum filtration (T₀) and the other half of the culture was treated with 200 uM dpp to remove iron from the system. After one hour of incubation, the iron-depleted culture was harvested (T₆₀). For each time-point, bacterial cells were snap frozen in liquid nitrogen and stored at −80°C until processed. RNA was purified as previously described (Carpenter et al., 2010). cDNA was reverse transcribed from 1 μg of total RNA using the Quantitect reverse transcription kit (Qiagen); for each RNA sample, a duplicate reaction that contained no reverse transcriptase enzyme was also performed. Levels of target transcripts were determined by quantitative real-time PCR (qRT-PCR) using the Quantitect SYBR Green qPCR kit (Qiagen). Each 20 μL reaction contained 1X SYBR green master mix, 1 μL of cDNA template, and 1.5 pmol each of a template specific forward and reverse primer. Cycling conditions included an initial activation step of 15 minutes at 95°C followed by 35 cycles of denaturation at 94°C for 15 seconds, annealing at 50°C for 20 seconds, and extension at 72°C for 30 seconds. For each primer set, post-run denaturation curves were performed to ensure that the reactions only resulted in a single amplicon. Relative gene expression was calculated as 2^−(ΔΔCt) using 16S as the internal reference gene. At least three biologically independent replicates were performed for each qRT-PCR experiment. During the course of these experiments, no overt growth defect was observed for any of the mutant strains.

Western blot analysis

H. pylori whole cell lysates were made from aliquots taken during the T₀ time-point of the RNA isolation experiments described above. One milliliter of each culture was pelleted and re-suspended in lysis buffer (Carpenter et al., 2010); total protein was quantitated for each lysate using the BCA kit (Pierce). Six micrograms of total protein was separated on 20% SDS-polyacrylamide gels, and Western blots were performed using polyclonal α-H. pylori Fur rabbit sera as previously described (Carpenter et al., 2010). Three biologically independent blots were performed for each strain.

Recombinant Fur expression strains and protein purification

The fur coding region from the A92T and H134Y mutant strains was PCR amplified using the HPFur_expression-F and HPFur_expression-R primers, digested with NdeI and XhoI, and ligated into the appropriately digested pDSM430 expression vector to create plasmids pDSM1193 and pDSM1194, respectively. The coding region from the E5A mutant strain was PCR amplified using the E5A-expression-F and HPFur_expression-R primers and cloned as described above to create plasmid pDSM1191. The presence of the appropriate mutation in each expression plasmid was confirmed by sequencing. The expression plasmids pDSM1191, pDSM1193 and pDSM1194 were each transformed into chemically competent E. coli Δfur BL21 (DE3) Rosetta/pLys cells (Carpenter et al., 2009) to create strains DSM1197, DSM1199, and DSM1200, respectively. Wildtype and each of the mutant Fur proteins was purified by FPLC as described previously (Carpenter et al., 2009). Briefly, Fur proteins were first passed through a HighTrap SP ion-exchange column using a gradient of 25 mM to 500 mM NaCl. Peak fractions were further purified by size-exclusion on a Sephacryl-200 column. Fur proteins were stored in 50 mM sodium phosphate, 500 mM NaCl, 25% glycerol, pH 8.0.

Circular dichroism (CD) spectroscopy

CD spectra of HpFurWT and mutant Fur proteins were analyzed on a Jasco-810 spectropolarimeter. Experiments were performed using a 1 mm path-length-cell at room temperature. Measurements were taken for wavelengths of 190–250 nm with a scan rate of 50 nm per minute. The spectra shown are an average of five accumulations. Samples contained 5μM protein in 25 mM sodium phosphate, pH 7.5. Thermal denaturation studies were performed using a 1 mm path-length-cell and 5 μM of each protein. Protein unfolding was monitored at 222 nm across a temperature range of 4–100°C.

Atomic absorption spectrophotometry

The ability of the HpFurWT and mutant Fur proteins to bind iron (Fe²⁺) was determined using graphite furnace atomic absorption mass spectrophotometry as previously described (Carpenter et al., 2010). Briefly, 2.5 μM of each protein was dialyzed against anoxic buffer that contained 50 mM ultrapure NaCl, 10 mM ultrapure sodium formate, the oxygen scavenging enzymes (Oxyrase, OB00-50), and increasing concentrations of ultrapure FeCl₂ • 4H₂O; the dialysis buffer was used at a pH of 7.0. The state of anaerobiosis was verified prior to dialysis using a Clarke- type oxygen electrode. The concentration of iron was measured using a Shimadzu AA-6701F spectrophotometer, and the amount of Fur-bound iron was calculated by subtracting the concentration of iron in protein free buffer from the concentration found in the dialysis bags. The number of iron molecules per monomer of Fur protein was calculated by dividing the amount of Fur-bound iron by the concentration of each Fur protein. Data shown represent at least three independent measurements.

Oligomerization assays

To evaluate any potential defects in oliogomerization, we assessed the ability of the Fur proteins to form higher order structures using in vitro cross-linking. Experiments were performed essentially as described previously (Carpenter et al., 2010) with minor changes. Briefly, 2 μg of each protein was combined with 1X apo- or MnCl₂ binding buffer (Carpenter et al., 2010), and 2 μL of 25 mM disuccinimidyl suberate (DSS, Sigma) and incubated for one hour at room temperature. For each reaction, a no-DSS control reaction was also performed. After incubation, 1X Laemmli buffer was added and samples were boiled for 5 minutes at 95°C. Reactions were separated on 12% SDS-polyacrylamide gels and stained using SYPRO Ruby Red (Invitrogen) according to the manufacturer’s recommendation. Each experiment was performed in triplicate.

Fluorescence anisotropy

Fluorescence anisotropy was used to measure the binding of HpFurWT and mutant Fur proteins to known binding sites in the Fe-Fur repressed amiE promoter and the apo-Fur repressed pfr promoter. Stock solutions (50 μM) of fluorescin-labeled promoters were prepared by annealing the labeled top-strand oligonucleotide with the unlabeled bottom strand oligonucleotide in 20 mM Tris-HCl, 100 mM NaCl, pH 8.0. Titrations involving the amiE Fur box were carried out in manganese-substituted binding buffer (30 mM Tris, 120 mM KCl, 16 mM DTT, 1 mM MnCl₂, 20% glycerol, 480 μg/mL Bovine Serum Albumin (BSA), and 0.1 mg/mL sheared salmon sperm DNA at pH 8.0) and titrations involving the pfr Fur box were conducted in apo-binding buffer (40 mM Tris, 150 mM KCl, 2 mM DTT, 600 μg/mL BSA, 200 μM EDTA, 24% glycerol, and 0.1 mg/mL sheared salmon sperm DNA at pH 8.0).

A cuvette containing 50 nM labeled DNA was incubated at 20 °C in sample chamber of an ISS PC-1 spectrofluorimeter (ISS, Inc., Champaign, IL) equipped with excitation and emission polarizers configured in the L-format. Anisotropy was recorded using an excitation wavelength of 495 nm and emission wavelength of 526 nm. Band passes for excitation and emission were set to 8 nm. Purified Fur protein was titrated stepwise into the labeled DNA duplexes until the change in measured anisotropy neared saturation. Each anisotropy value represents the average of 30 measurements recorded with an integration time of 3 seconds.

Anisotropy data (change in anisotropy) were analyzed assuming a 1:1 interaction of Fur dimer with the promoter DNA and Fur concentrations are represented in terms of [Fur₂]. A single-site binding model was fit to the data using the nonlinear fitting package in Mathematica (Wolfram Research, Inc). Errors in K_d are derived from the residual sum of squares from nonlinear fitting.

Fur protein homology modeling

The H. pylori Fur amino acid sequence was modeled against the C. jejuni apo-Fur structure (PDB 4ETS) using the SWISS MODEL server (http://swissmodel.expasy.org) (Schwede et al., 2003). Chains A and B were modeled separately and then combined into a dimer structure in PyMol. This dimeric structure was used for residue highlighting (Figure 9) and structural overlay images (Figure S1). Due to differences in the H. pylori and C. jejuni primary amino acid sequence, and symmetric differences between the two protomers of the C. jejuni apo-Fur structure, chain A of the modeled H. pylori apo-Fur structure contains residues 4-146, and chain B contains residues 9-149.

Statistical analysis

Differences in Fur protein levels and levels of amiE and pfr transcripts were compared using a Student’s t test on log₁₀ transformed fold changes values with a Holm-Sidak correction for multiple comparisons. For clarity of presentation, the data are shown in the original scale.

Supplementary Material

Supp Fig S1-S4

Figure S1. Overlay of the modeled H. pylori apo-Fur structure (blue) with C. jejuni apo-Fur (red) (Butcher et al., 2012). The H. pylori structure was modeled as described in the Materials and Methods. Merged structures and images were generated in PyMol. The metal coordinating residues in S1 and S3 are completely conserved between structures. Notably, the major regions appear to be fairly similar with the exception of the N-terminal extension.

Figure S2. Representative SDS PAGE gel of random mutant strains screened for altered Pfr levels. The approximate size of the Pfr band (19 kDa) is indicated by the arrow on the left; de-repression of Pfr can be seen in the Δfur, I52T and S60N strains. MW, molecular weight marker; G27, wildtype G27 strain; Δfur, fur deletion strain.

Figure S3. Densitometric analysis of Fur crosslinking. (A) Relative abundance of major Fur protein species in manganese-substituted conditions before the addition of the crosslinking reagent DSS. (B) Relative abundance of Fur protein species in manganese-substituted conditions after the addition of DSS. (C) Relative abundance of Fur protein species in apo conditions after the addition of DSS. The mean % of total each protein sample from 3 replicates of each protein sample is shown.

Figure S4. Non-specific DNA binding of HpFurWT, HpFurA92T, and HpFurH134Y proteins measured by fluorescence anisotropy. For the HpFurWT and each mutant protein we incubated 50 nM labeled amiE promoter DNA with a concentration of protein that resulted in ~70% binding saturation (concentration of each protein based on forward titrations depicted in Figure 8) under manganese-substituted conditions. The HpFur:labeled amiE DNA complexes were titrated with unlabeled amiE promoter DNA in which the Fur box had been scrambled to “GT” repeats. For each titration, the initial anisotropy value was set to 1 and subsequent anisotropy values are plotted as a fraction of the initial anisotropy value (Fraction bound). A decrease in anisotropy after the addition of unlabeled scrambed amiE promoter DNA indicates binding to the competing promoter DNA. At high concentrations of competing oligo, titration against the HpFurWT;labeled amiE DNA complex produced a minor reduction in anisotropy (black diamonds). Conversely, titrations against the HpFurA92T (green squares) and HpFurH134Y:labeled amiE promoter DNA complexes (red triangles) resulted in a more substantial decrease in anisotropy, indicating that the HpFurA92T and HpFurH134Y proteins bound non-specific DNA sequences much more readily that the HpFurWT protein.