The LysR-Type Transcription Factor HbrL Is A Global Regulator Of Iron Homeostasis And Porphyrin Synthesis In Rhodobacter capsulatus

Summary

The purple bacterium Rhodobacter capsulatus is unique among Rhodobacteriacae as it contains a putative Iron response regulator (Irr) but does not possess a copy of the Ferric uptake regulator (Fur). Interestingly, an in-frame deletion mutant of Irr shows no major role in iron homeostasis. Instead, we showed that the previously identified activator of heme gene expression HbrL is a crucial regulator of iron homeostasis. We demonstrated that an HbrL deletion strain is unable to grow in iron-limited medium in aerobic, semi-aerobic and photosynthetic conditions and that suppressor strains can be isolated with mutations in iron uptake genes. Gene expression studies revealed that HbrL is a transcriptional activator of multiple ferrous and ferric iron uptake systems in addition to a heme uptake system. Finally, HbrL activates the expression of numerous heme biosynthesis genes. Thus, HbrL has a central role in controlling the amount of iron transport in conjunction with the synthesis of its cognate tetrapyrrole heme.

Introduction

Even though iron is the fourth most abundant element of Earth's crust, its availability is severely restricted in oxic environments due to extremely low solubility of ferric iron Fe³⁺ (10⁻¹⁸M at pH 7) as compared to ferrous iron Fe²⁺ (0.1M at pH 7) (Andrews et al., 2003). In most cells, iron has important functions as an electron and gas carrier as well as a gas and redox sensor. For example, iron is frequently used as mono- or bi-nuclear cofactors in enzymes where they function as electron carriers or sensors of cellular redox. Iron is also inserted into protoporphyrin IX to form heme where it also functions as an important electron and gas carrier/gas sensor.

Free iron and free heme are both highly toxic compounds and thus rarely present in bacterial cells. Through Fenton's chemistry, free iron can generate hydroxyl radicals (OH^•) that induces the formation of alkyl radicals (R^•) from unsaturated fatty acids (RH) (Touati, 2000; Chiancone et al., 2004; Cornelis et al., 2011). Free heme toxicity has been known for decades but the mechanism whereby heme leads to toxicity remain unclear. Models of heme toxicity involve the releasing of iron, the destabilization of membranes, the exerting of peroxidase and/or mono-oxygenase activities that can damage DNA or proteins, or the synthesis of highly reactive alkoxyl (RO^•) and peroxyl (ROO^•) radicals (Graça-Souza et al., 2006; Anzaldi and Skaar, 2010). In addition, heme precursor porphyrins also produce reactive oxygen species via photochemical reactions (Nakahigashi et al., 1991; Yang et al., 1995). Because of these toxicity problems, cells must have appropriate regulatory mechanisms to tightly control the transport and storage of iron with the synthesis of heme. However, an understanding of regulatory networks involved in iron homeostasis is present for just a small group of prokaryotes (Andrews et al., 2003; Rudolph et al., 2006; Cornelis and Andrews, 2012). Smaller still is the number of studies dealing with the regulation of heme biosynthesis (Schobert and Jahn, 2002; Zappa et al., 2010). Prior studies have focused on individual pathways so little is known about how cells coordinate iron transport with biosynthesis of heme.

A correlation between iron availability with that of porphyrin and heme contents was initially observed in Corynebacterium diphteriae as early as the 1940s (Pappenheimer, 1947; Pappenheimer and Hendee, 1947; Rawlinson and Hale, 1948). Later, Lascelles published a series of studies in Rhodobacter sphaeroides and Novispirillum itersonii [formerly known as Spirillum itersonii and Aquaspirillum itersonii (Yoon et al., 2007)] where she showed that iron deficiency induces excretion of precursor porphyrins, resulting in a reduction of heme, cytochrome and bacteriochlorophyll content (Lascelles, 1956; Clark-Walker et al., 1967; Lascelles et al., 1969; Lascelles and Hatch, 1969).

In regards to regulating iron transport, the ferric uptake regulator Fur is a well-characterized regulator that is present in a broad range of species (Rudolph et al., 2006; Lee and Helmann, 2007; Carpenter et al., 2009; Cornelis and Andrews, 2012). The Fur paradigm consists of repression of iron regulon genes by the binding of Fe²⁺ containing Fur holoprotein to DNA promoters under iron replete conditions. During iron scarcity, apoprotein Fur is unable to bind DNA thereby allowing transcription of genes involved in iron uptake. This long standing model is now known to be simplistic as additional layers of complexity have recently been described including: i) Fur-dependent expression of a small RNA RhyB that equilibrates the ratio of essential over non-essential iron-using protein as a function of iron availability (Massé et al., 2007; Lee and Helmann, 2007); ii) description of repressor and activator activity of both apo-and holo-Fur (Lee and Helmann, 2007; Carpenter et al., 2009; Butcher et al., 2012); iii) Fur regulating the uptake of other metals(Hohle and O'Brian, 2009; Davies et al., 2011; Perry et al., 2012). This complex Fur-mediated metabolic shift from an iron-based biochemistry to an alternative one, for example using manganese, is known as the “iron-sparing response” (Massé and Gottesman, 2002; Smaldone et al., 2012).

Iron regulation in α-proteobacteria is of particular interest as the Fur paradigm is not strongly supported in this phylum. Indeed, some branches, namely the Rhizobiales and the Rhodobacterales, have invented other iron regulators. For example, the iron response regulator, Irr, is a member of the Fur superfamily that senses the intracellular level of iron through the binding of heme. Among the Rhizobiales, the Rhizobiacae have added an extra-level of regulation with the appearance of RirA that senses iron requirement as a function of the Fe-S level. In this phylla, the emergence of Irr and RirA as iron regulators seems to be correlated with a functional shift of Fur from an iron sensor to a manganese/zinc sensor, thus termed Mur/Zur respectively (Rodionov et al., 2006; Johnston et al., 2007; Lee and Helmann, 2007). A manganese sensing shift by Fur has been confirmed in several Rhizobiale species (Rhizobium leguminosarum, Sinorhizobium meliloti, B. japonicum, Brucella abortus and to some extent in Agrobacterium tumefaciens), suggesting that it has occurred in a broad range of α-proteobacteria (Wexler et al., 2003; Díaz-Mireles et al., 2004; Chao et al., 2004; Kitphati et al., 2007; Hohle and O'Brian, 2010; Menscher et al., 2012).

Photosynthesis generates heavy needs for iron as light-driven electron transport requires numerous heme-containing cytochromes along with a nuclear-iron containing reaction center and Fe-S containing Rieske protein (Vermeglio and Joliot, 1999). The synthesis of heme and the photosynthetic pigment bacteriochlorophyll also requires several iron dependent enzymes. For example, the enzyme coproporphyrinogen III oxidase in the common heme/bacteriochlorophyll trunk of the pathway requires an Fe-S cluster. Likewise the enzymes magnesium chelatase and dark-operative protochlorophyllide reductase in the bacteriochlorophyll branch also contain Fe-S clusters (Sirijovski et al., 2007; Sarma et al., 2008; Zappa et al., 2010). Thus, even at the post-transcriptional level, the process of photosynthesis is highly dependent on iron transport.

By lacking a fur gene, R. capsulatus represents an extreme trend of the α-proteobacteria regarding Fe/Mn regulation. In addition, R. capsulatus stands out as a unique member of this genus in regards to the diversity of its iron homeostasis apparatus (Zappa and Bauer, 2013). In this study, we show that an Irr homolog from R. capsulatus unexpectedly does not exert major control on iron homeostasis. Instead, we demonstrate that a previously described heme binding activator of heme gene expression, the LTTR termed HbrL, is also involved in regulating both iron transport and heme biosynthesis as a function of iron availability.

Results

Deletion of hbrL severely impairs growth of R. capsulatus in iron-depleted media

Inspection of the annotated SB1003 genome shows the presence of two ORFs listed as “transcriptional factor, Fur family”, corresponding to COG0735. Further analysis of their amino-acid sequences shows that ORF rcc02670 corresponds to an Irr representative rather than Fur. The other ORF, rcc01134, presumably encodes the zinc uptake regulator Zur as its located in a gene cluster containing zinc ABC transporter genes znuB and znuC, and a zinc-binding protein encoding znuA (Zappa and Bauer, 2013).

Given the absence of Fur, we addressed whether Irr may have a role in regulating iron homeostasis. Using the wild-type strain and isogenic mutants Δirr and Δzur, we assayed optical density of cultures grown for 24h in RCV* minimal defined media with 50μM FeSO₄ as well as increasing amounts of the iron chelator 2,2'-dipyridyl (DIP). Figure 1 shows the average optical density at 660nm of 6 biological replicate cultures grown for 24h. The WT strain coped with the presence of iron chelator with no visible sign of growth defect up to 100μM of DIP. Indeed, approximately 50% of its maximum cell density is retained in the presence of 150μM DIP. At 150μM and 200μM DIP, the Δirr and Δzur strains also showed a better growth than did the WT strain.

Figure 1.

Growth of R. capsulatus SB1003 in response to DIP. (A) Final OD₆₆₀ of various R. capsulatus strains after 24h incubation in RVC* medium containing 50μM FeSO₄ with increasing amount of DIP. Each value corresponds to the average of six biological replicates, grown in three different days. Error bars correspond to Standard Error of the Mean.

Hypothesizing that iron homeostasis could be linked to the regulation of heme, which binds iron, we tested a series of regulatory mutants known to control heme synthesis (Smart et al., 2004). The ΔhbrL mutant stood out as being very sensitive to iron-depletion. Indeed, DIP as low as 50μM was sufficient to reduce the ΔhbrL strain to a final cell density that is approximately 30% the maximal level reached by all other tested strains. It is clear from growth inhibition assays that Δirr and Δzur cells exhibit minimal growth effects in iron depleted medium as compared to ΔhbrL cells.

To further characterize the effect of iron limitation, we also assayed growth on iron-repleted and iron-depleted PY media under aerobic, semi-anaerobic and photosynthetic anaerobic conditions. As shown in Fig. 2, the WT strain easily coped with the presence of 100μM DIP as it exhibited no impaired growth in aerobic and semi-anaerobic conditions and only a slight delay under anaerobic photosynthetic growth. Such stronger inhibition in anaerobic conditions was expected as the absence of oxygen promotes the presence of Fe²⁺, which is preferentially chelated by DIP. Thus, in this experimental setting, iron is chelated more effectively and thus less available to the cells under anaerobic conditions. Like that of the parent strain, the Δirr mutant also exhibited good tolerance to iron-depletion under similar conditions (Fig. 2). This is contrasted by the ΔhbrL strain that exhibited growth impairment under all iron depleted conditions (Fig. 2).

Figure 2.

Growth kinetics of WT, Δirr, ΔhbrL and ΔhbrL^SUP1 in PY medium containing 50μM FeSO₄ and 50μM FeCl₃ in the absence (close symbols, solid lines) or presence (open symbols, dashed lines) of 100μM DIP. Growth is measured by following optical density at 660nm (OD₆₆₀). Error bars correspond to Standard Error of the Mean.

The iron stress phenotype of ΔhbrL was further characterized by checking semi-anaerobic growth kinetics on a wide range of DIP concentrations in the defined minimal medium RCV* (Fig. 3). Interestingly, the WT strain did not exhibit significant growth inhibition until DIP reached 200μM where it resulted in a slight reduction in growth. On the other hand, the ΔhbrL mutant responded to increasing DIP concentrations in a dose effect manner with its growth rate decreasing as DIP concentration increased. Given that DIP preferentially chelates Fe²⁺ with a 2:1 stoichiometry and that RCV* contains 50μM Fe, 100μM DIP should titrate all of the available iron if it were all fully reduced. However, semi-anaerobic conditions enable enough dissolved oxygen to maintain some extracellular Fe³⁺ (Fig. 2), so at 100μM DIP, the extracellular Fe²⁺ is likely all depleted with a fraction of Fe³⁺ remaining to partially feed these cells. At 200μM, where DIP is in large excess it is more likely that most Fe is depleted leading to a harsher intra-cellular iron stress. Such harsh iron depletion could generate a general stress response to these cells that goes beyond the unavailability of iron.

Figure 3.

Growth of wild type and ΔhbrL mutant strains under varying levels of DIP.

Suppressors of the Δ hbrL DIP phenotype contain multiple mutations in iron homeostasis genes

Continued growth of ΔhbrL in the presence of 250μM DIP beyond 48h consistently generated rapidly growing suppressors of the ΔhbrL phenotype. Two suppressors, ΔhbrL^SUP-1 and ΔhbrL^SUP-2, exhibiting wild type ability to grow in the presence of DIP were studied (Fig. 1 and Fig. 2). To obtain a better understanding of what genes may be regulated by HbrL, we undertook comparative genomic sequence analysis of these suppressors. Interestingly, both suppressors exhibit multiple mutations in iron uptake genes (Table 1). ΔhbrL^SUP-1 exhibits two single nucleotide polymorphisms (SNP) in ORF rcc00101 that codes for a ABC transporter permease protein located in a 22kb region of the genome that appears dedicated to iron uptake (Table 1). This region codes for genes involved in Fe²⁺ uptake Feo (rcc00090-00093), heme uptake (rcc00094-00098), an ABC transporter (rcc00099-00103) and a ferrichrome uptake (rcc00105-00108'00111) system. ΔhbrL^SUP-1 also contains insertion deletion polymorphisms (InDel) in ORF rcc01439 and rcc01443 that are both part of a ferric enterobactin uptake system (Table 1). Finally, ΔhbrL^SUP-1 also contains one InDel in the Fe²⁺ permease EfeU (Table 1).

Table 1.

Mutations related to iron homeostasis in ΔhbrL suppressors

Summary	Δ hbrL SUP1 3	Δ hbrL SUP2 3
Average coverage	176.1X	229.8X
Total SNP1: intragenic	29 (28)	132 (65)
intergenic	6 (4)	66 (24)
Total InDel2: intragenic	42 (37)	79 (30)
intergenic	7 (6)	55 (29)
SNP in iron regulon: intragenic	2	8
intergenic		2
InDel in iron regulon: intragenic	3	7
intergenic		4

Strain	Locus	Description	Position	Mutation type	Nucleotide change	Coding change	Frequency (%)	Coverage
Δ hbrLSUP−1	rcc00101	ABC transporter, permease protein	108670	SNP	C->A	Gly65Cys	100	4
			108673	SNP	A->T	Phe64Ile	100	4
	rcc01439	ABC transporter, ATP-binding/permeas e protein	1552051	Insertion, 1bp	A	Frameshift at Arg23	60	5
	rcc01443	Ferric enterobactin tranport system, permease protein FepD-2	1557629	Deletion, 1bp	ΔC	Frameshift at Ser283, introduce a stop codon at 285	100	4
	rcc03065	Ferrous iron permease EfeU	3258964	Insertion, 1bp	C	Frameshift at Thr260	100	3
Δ hbrLSUP−2	rcc00097	Hemin transport protein HmuS	103837	SNP	G->C	Ala222Gly	35	54
	rcc00105	Ferrichrome ABC transporter, ATP-binding protein FhuC-1	112764	SNP	A->C	Ala48Ser	34	86
			112771	Deletion, 1bp	ΔC	Frameshift at Arg45	37	92
			112777	Insertion, 1bp	G	Frameshift at Ser44	35	97
			112794	SNP	A->C	Ala38Ser	38	91
			112803	SNP	C->G	Pro35Arg	38	92
	rcc01046	Iron siderophore/cob alamin ABC transporter, permease protein	1125878	Insertion, 1bp	C	Frameshift at Arg115	87	90
			1125880	Insertion, 1bp	C	Frameshift at Arg115	87	90
	rcc01342	Iron siderophore/cob alamin ABC transporter, periplasmic protein	1444970	Insertion, 1bp	C	101bp from ATG	34	70
			1444996	SNP	A->G	75bp from ATG	54	87
	rcc01343	Iron siderophore/cob alamin ABC transporter, permease protein	1446480	Deletion, 1bp	AG	Frameshift at Ala127	57	74
			1446685	Insertion, 1bp	G	Frameshift at* 195	43	89
	rcc01445	tonB-dependent receptor	1560953	SNP	G->T	Met410Arg	45	150
			1560955	SNP	G->C	Leu411Val	45	149
			1561778	SNP	T->C	Val685Ala	71	99
	rcc02028	FeoA family protein	2185842	Insertion, 1bp	C	390bp from ATG	75	4
			2185996	SNP	A->G	236bp from ATG	100	2
			2185996	Deletion, 1bp	ΔA	236bp from ATG	33	3
			2185999	Insertion, 1bp	A	233bp from ATG	50	6
	rcc03065	Ferrous iron permease EfeU	3258964	Insertion, 1bp	C	Frameshift at Thr260	84	126

¹SNP denotes single nucleotide polymorphism

²InDel denotes insertion or deletion polymorphism

³total number of mutations is indicated and the corresponding number of gene concerned is in parenthesis. Frequency: percentage of nucleotide change out of the total reads. Coverage: number of sequencing reads of the correspond base position.

The second sequenced suppressor, ΔhbrL^SUP-2, exhibits a SNP mutation in ORF rcc000097 that codes for a subunit of the heme uptake ABC transporter HmuRSTUV as well as additional mutations in seven iron uptake systems (Table 1): three SNP and two InDel mutations in rcc00105 that code for ATP binding subunit in a ferrichrome ABC transporter, two InDel in rcc001046 that codes for a permease subunit in an iron siderophore/cobalamine ABC transporter, one SNP and one InDel upstream of rcc01342 and two InDel in rcc01343 that code for subunits in a second copy of an iron siderophore/cobalamine ABC transporter, three SNP in a enterobactin TonB-dependent receptor (Table 1). Ferrous iron uptake is also affected in ΔhbrL^SUP-2 as one SNP and three InDels upstream of a Feo operon (rc02028-02029) and one InDel in the ferrous iron permease EfeU (Table 1). Interestingly, both independently derived suppressors ΔhbrL^SUP-1 and ΔhbrL^SUP-2 share the same insertion of a cytosine after the adenine 778 in the efeU gene. Overall, the presence of numerous mutations in iron uptake systems in ΔhbrL suppressors, coupled with the strong phenotype of the ΔhbrL mutant under iron depletion, suggests that HbrL may be a key regulator of iron homeostasis in R. capsulatus.

HbrL controls expression of multiple genes involved in iron uptake

To test whether HbrL regulates iron uptake genes, we designed primers to undertake quantitative reverse transcriptase-PCR (qRT-PCR) analysis of the expression of 14 genes associated with iron homeostasis. The tested genes coded for Fe²⁺ uptake systems Feo1, Feo2 and Efe, several Fe³⁺ uptake systems including seven TonB-dependent outer membrane siderophore receptors and one ferric iron ABC transporter; the Fief ferrous iron efflux pump; and a Bfr bacterioferritin iron storage gene.

In one assay, we exposed cells to 100μM of DIP, which is a concentration that does not significantly impact growth of WT cells and only partially impacts growth of ΔhbrL cells (Fig. 3) and then assayed Fe gene expression after two hours of DIP exposure (Fig. 4). In WT cells there is a significant increase observed for transcription of all assayed Fe²⁺ and Fe³⁺ uptake genes (Fig. 4, open bars) (table S3, P(H1)<0.05). Moreover, DIP induced activation of Fe²⁺ and Fe³⁺ uptake genes is not observed in the ΔhbrL strain, which exhibited an “iron-blind” phenotype at 100μM DIP (Fig. 4, solid bars) (table S3, P(H1)<0.05). In a second assay, we exposed cells to 200μM of DIP, which slightly affects WT growth and very severely affects ΔhbrL growth (Fig. 3). In this assay, WT cells shows even higher expression of all assayed Fe²⁺ and Fe³⁺ uptake genes than it did under 100μM of DIP (Fig. 5, open bars). Although the ΔhbrL mutant seems to exhibit a failure in its ability to increase Fe²⁺ and Fe³⁺ uptake gene expression, statistical analysis does not confirm a significant effect of HbrL (Fig. 5, solid bars) (table S4, P(H1)>0.05). The effect of 200μM of DIP on Fe²⁺and Fe³⁺ uptake expression in the ΔhbrL mutant is tempered by the fact that these cells are very severely limited for growth so there is also the likely possibility that additional transcriptional factors could affect their expression in response to excessive stress conditions.

Figure 4.

Expression assay of genes involved in iron homeostasis under mild iron stress with 100μM DIP. Gene expression is represented as expression ratios using the WT strain in the absence of DIP as the arbitrary reference, normalized by rpoZ. (A) Genes associated with the uptake of ferrous iron. (B) Genes associated with uptake of ferric ion. (C) Genes associated with heme uptake, iron efflux and storage. White bars: WT; gray bars: ΔhbrL; t(h): time of exposure to DIP in hours; the dotted line represents the gene expression reference, i.e. transcription change = 1. Each value was calculated from analysis of RNA preparations from six biological replicates grown on three different days.

Figure 5.

Expression assay of genes involved in iron homeostasis under high iron stress with 200μM DIP. Legend identical as in fig. 4.

In contrast to the expression of Fe²⁺ and Fe³⁺ import transporters that appears to be under the control of HbrL, fieF which codes for a efflux ferrous iron pump and bfR that codes for the iron storage protein bacterioferritin showed stable transcription in the presence of DIP with no significant HbrL-dependency (Fig. 4-5, table S3-S4).

Iron depletion induces transcription of heme uptake and porphyrin synthesis genes, in the presence of HbrL

Genes coding for enzymes involved in the eight steps leading to heme synthesis, along with hmuR that codes for the TonB-dependent outer membrane receptor of the HmuRSTUV heme uptake system, were also tested for expression using qRT-PCR. Mild iron stress (100μM DIP) induced the increased expression of most hem genes by 1.75- to 8-fold in the WT strain (Fig. 6, open bars) (table S5, P(H1)<0.05). Unlike the iron uptake genes for which increasing the DIP concentration from 100 to 200μM intensified gene expression even more, the transcription of hem genes still increased but to a lesser extent at higher DIP concentrations relative to expression in the presence of 100μM DIP (Fig. 7, open bars) (table S6, P(H1)<0.05). Indeed, all hem genes, with the exception of hemD, showed an activation level ~2-fold lower under harsh iron stress than under mild iron stress. Thus, the moderate activation of hemH observed at 100μM DIP is further reduced at 200μM and not significant anymore (Fig. 6-7, tables S5-S6). In addition, iron-mediated transcriptional changes of hem genes were not observed in the ΔhbrL mutant as seen with iron uptake gene expression at 100μM DIP (Fig. 6, solid bars) (table S5, P(H1)<0.05). However, unlike the iron transporters, this HbrL-mediated regulation is conserved at the higher 200μM concentration of DIP (Fig. 7, solid bars) (table S6, P(H1)<0.05). Overall, these expression assays highlight that HbrL is a crucial actor in the upregulation of hem genes in the absence of iron.

Figure 6.

Expression assay of genes involved in heme synthesis under mild iron stress, Legend identical as in fig. 4.

Figure 7.

Expression assay of genes involved in heme synthesis under high iron stress, i.e. 200μM DIP. Legend identical as in fig. 4.

In addition to de novo heme synthesis, heme can presumably be imported using the HmuRSTUV heme uptake system. qRT-PCR analysis of genes involved in this uptake system show that they are highly responsive to iron availability as hmuR expression was activated ~8- to ~450-fold in the presence of DIP at 100μM and 200μM, respectively (Fig. 4-5, open bars) (tables S3-S4). In both DIP concentrations, this operon is also significantly activated in the ΔhbrL mutant (100μM DIP: P(H1)=0.033; 200μM DIP, P(H1)=0.001). Nevertheless, activation is also significantly different between WT and ΔhbrL after 2h of exposure to DIP, at both 100μM and 200μM (P(H1)=0.041 and P(H1)=0.025, respectively). Thus, the absence of hbrL drastically impaired this transcription increase, indicating that HbrL also has an important role in heme uptake along with an unidentified transcriptional regulatory mechanism.

We addressed whether this heme uptake system is functional and an important contributor to the observed phenotype by addressing the growth of cells in the presence of hemin and in the presence of increasing DIP concentrations (Fig. 8). In the absence of hemin, the WT strain exhibits severe growth inhibition at DIP concentrations above 150μM. However, in the presence of exogenous 10μM or 50μM hemin, growth is restored even at DIP levels as high as 200μM (Fig. 8). This indicates that the WT strain is capable to use hemin as an iron source to cope with the absence of available cationic iron. In contrast, the ΔhbrL mutant displays impaired growth in the presence of heme buffer, which contains ethanol and sodium bicarbonate (as seen by its low OD₆₆₀ when no DIP is added). The addition of hemin seems to improve its growth slightly at the lower DIP concentrations. Nevertheless, it does not rescue the growth of the ΔhbrL mutant in a manner comparable to what is observed for the WT. Such growth impairment is consistent with the previously observed defective activation of the heme uptake operon in the ΔhbrL mutant (Fig. 8).

Figure 8.

Supplementation of iron-limited RCV* with exogenous hemin. WT, ΔhbrL were grown in RCV* containing various DIP (0 to 250μM, X axis) and hemin (circle: 0, square: 10μM, triangle: 50μM) concentrations. Optical density at 660nm was read after 18 to 24 hours of incubation. Each value corresponds to the average of three biological replicates, grown in three different days. Error bars correspond to Standard Error of the Mean.

Discussion

In a previous study, we demonstrated that HbrL can function as an activator of hemA, and hemN2 (hemZ) expression in the absence of exogenous heme (Smart and Bauer, 2006). In this study, we further show an HbrL-dependent increase in transcription of all hem genes when cells are grown in iron limiting growth medium (Fig. 9). It is worth noting that the least activated gene was hemH that was only slightly (~1.7) activated. This gene encodes the ferrochelatase, the last enzyme of the heme biosynthesis pathway that inserts iron into protoporphyrin IX to form heme. Such a property highlights the involvement of HbrL in the synthesis of precursor porphyrins rather than the final product heme. Indeed, early literature showed that iron depletion induces an increase of porphyrin synthesis while overall heme content decreases (Pappenheimer, 1947; Pappenheimer and Hendee, 1947; Rawlinson and Hale, 1948; Lascelles, 1956; Clark-Walker et al., 1967; Lascelles et al., 1969; Lascelles and Hatch, 1969). While decreases in heme levels during iron scarcity make sense, accumulation of precursor porphyrins that are known to be potentially toxic is not intuitive and a biological interpretation remains to be found. Nevertheless, this study confirms this trend at the mRNA level, i.e. Fe limitation promotes the accumulation of mRNA related to heme synthesis genes with a favorable ratio to precursor porphyrin genes. Such an imbalance in hem gene expression could lead to heme precursor porphyrin predominance.

Figure 9.

HbrL, a central regulator of iron import and heme synthesis. Iron scarcity induces a HbrL-mediated synthesis of ferric iron (ferrisiderophore uptake) and ferrous iron (Feo- and EfeUOB-type) transporters. It also activates the uptake of heme as well as its biosynthesis. Finally, heme exerts a repression on HbrL activity (Smart and Bauer, 2006).

Another finding of this study is that Fur-family mutants, Δirr and Δzur exhibited better tolerance to iron depletion than did the WT strain. Even though we cannot yet provide definitive roles for Δirr and Δzur in this species from this work, the phenotype of the Δirr strain could be a result of oxidative stress. Indeed, iron and oxidative stress are closely related and Irr has been proven to respond to this type of stress, including in the Rhodobacter genus (Martínez et al., 2006; Peuser et al., 2012). Regarding Δzur, assuming it does function as a zinc uptake regulator, one can imagine that an excess of DIP may generate non-specific Zn-chelates that would modify the mechanism of Zn homeostasis.

The vast majority of α-proteobacteria have at least one copy of the fur/mur gene but it appears that it has little involvement is regulating iron level in this group of organisms (Rodionov et al., 2006; Johnston et al., 2007). Occurrence of Irr in the Rhizobiales and Rhodobacterales is likely to be correlated with a functional shift of Fur from an iron regulator to a manganese regulator, as already confirmed in some Rhizobiale species (Wexler et al., 2003; Díaz-Mireles et al., 2004; Chao et al., 2004; Kitphati et al., 2007; Hohle and O'Brian, 2010; Menscher et al., 2012). Few studies have been done concerning the Rhodobacterale phylum but it appeared that Fur and Irr in R. sphaeroides have little effect on iron homeostasis (Peuser et al., 2011; Peuser et al., 2012). Moreover, genome screening of α-proteobacteria revealed that only Mesorhizobium loti and R. capsulatus are lacking the fur gene. These two species are also the only one showing the mntR gene, encoding a manganese uptake regulator, which was most likely acquired by horizontal transfer (Rodionov et al., 2006; Johnston et al., 2007). Thus, because of the Fur-based manganese regulation observed in a large part of α-proteobacteria and moderate phenotype of Irr in iron response in R. sphaeroides and R. capsulatus, one can speculate that the use of a new regulator of iron homeostasis could be wide-spread in Rhodobacterales. Our study points toward the heme binding LTTR HbrL for that role. Overall, R. capsulatus would represent an extreme example of this evolutionary trend, as it has also traded fur/mur for mntR.

Analysis of the R. capsulatus genome indicates that this species has three Fe²⁺ uptake systems comprised of two Feo and one EfeUOB systems. Occurrence of two Feo systems in the same organism is not rare and in some cases one of the Feo systems actually can become specialized in Mn²⁺ uptake (Andrews et al., 2003; Dashper et al., 2005; Cartron et al., 2006) or Fe²⁺ uptake for a specific task (Rong et al., 2012). Both Feo systems in R. capsulatus are likely to be involved in Fe²⁺ uptake as expressions of both are up-regulated under iron limitation in the presence of HbrL. Interestingly, under high iron depletion stress (200μM DIP), feo2 is actually repressed. R. capsulatus also contains an EfeUOB system that corresponds to a Fe²⁺ uptake system. The mechanism of uptake by EfeUOB is not yet settled as, in addition to be involved in iron transport, it has been reported that EfeUOB can extract Fe from heme (Grosse et al., 2006; Cao et al., 2007; Létoffé et al., 2009). According to the annotation, in R. capsulatus SB1003, EfeU and EfeO are fused into a single two-module peptide, that to our knowledge, occurs in only one other system (Rajasekaran et al., 2010). In order to confirm a frame shift observed in both ΔhbrL^SUP-1 and ΔhbrL^SUP-2, we resequenced this region of the genome and found out that this regions was mis-sequenced. Specifically, the insertion is present in ΔhbrL^SUP-1 and ΔhbrL^SUP-2, strains as well as in the WT and ΔhbrL strains. Interestingly, this frameshift restores the more common configuration of a tripartite efeUOB operon. We speculate that the EfeUOB system may also be involved in extracting Fe from heme for use as an iron source as there is no classic heme oxygenase identified in the R. capsulatus genome. Our data also shows that HbrL is crucial in upregulating heme uptake and the putative deferrochelatase system EfeUOB when iron is scarce. Also, growth inhibition of the wild type strain in iron-depleted conditions can be alleviated by supplementing with heme and that this is not the case for the ΔhbrL mutant. Thus, HbrL could be involved in the use of heme iron, both at the heme uptake and in the iron extraction level. This conclusion must, however, be taken with the caveat that a deferrochelatase role of EfeUOB is still being debated (Dailey et al., 2011). Another candidate for heme iron usage is HmuS, which encoded in the hmuRSTUV operon. Orthologues of this protein have been shown recently to be involved in heme degradation (Zappa and Bauer, 2013; Roe et al., 2013).

R. capsulatus also contains genes coding for iron storage protein bacterioferritins as well as a gene coding for an Fe²⁺ efflux pump. Absence of significant transcriptional response to iron stress of the tested bacterioferritin and iron efflux pump illustrate a possible strategy to deal with iron toxicity in the cell: maintaining a basal level of both efflux and storage to avoid excess of free iron while modulating the import of iron as a function of its availability. Both of the tested genes are not part of the HbrL regulon indicating that HbrL appears dedicated to increase the concentration of cellular iron and is not involved in regulating iron storage or export.

The absence of a fur gene in the sequenced R. capsulatus genome (Zappa and Bauer, 2013), coupled with the absence of a strong phenotype of the Δirr strain, provides a good indication that HbrL is a major regulator for the control and coupling of heme synthesis with iron homeostasis in this species. However it should be noted that the presented data does not address whether HbrL directly or indirectly regulates heme synthesis and iron transport genes, as this would require extensive DNA binding studies that is beyond the scope of this study. For example, other LTTRs have been found to exert both direct gene expression control and indirect control via an interaction with another transcription factor (Ghrist et al., 2001; Kovacikova et al., 2004; Dangel and Tabita, 2009). Nevertheless, one important observation is that HbrL is itself capable of binding heme as a cofactor (Smart and Bauer, 2006). This latter point indicates that reduction of cellular heme content, as would occur when iron becomes limiting, likely causes apo-HbrL to activate iron transport and heme gene expression. Conversely, when heme is in excess, heme-HbrL likely either does not bind to this class of promoters or alternatively binds but does not appropriately interact with RNA polymerase to stimulate transcription. Such heme driven activity of HbrL could occur within both cases of a direct or indirect regulation process. Future biochemical studies on the mode of activation by HbrL, and the role of heme in this process, will help differentiate these possibilities.

In B. japonicum, the O'Brian group has unraveled many aspects of the master regulator Irr that is shown to connect iron availability with iron transport and heme biosynthesis (Hamza et al., 1998; Qi and O'Brian, 2002; Small et al., 2009; Small and O'Brian, 2011). In our study, we show that a homolog of Irr does not appear to have a role in controlling iron homeostasis in R. capsulatus. Instead, HbrL occupies a central role in coordinating iron uptake with synthesis of porphyrin/heme (Fig. 9). As is the case where Irr/RirA has highlighted the limitation of the Fur paradigm, these results with HbrL suggest that the Irr/RirA model could also be limited within the α-proteobacteria. A BLAST analysis using the HbrL sequence reveals numerous LTTR with identity score in the 45-55% range in the proteobacteria phylum, especially of the α class (Altschul et al., 1990). Whether HbrL homologous exert such a crucial role in other species remains to be elucidated, as well as how HbrL exerts its regulation at the molecular level. At the very least, these results suggest that different species have evolved several different mechanisms to control the import and export of iron with the control of heme synthesis.

Experimental procedures

Strains, media and growth conditions

R. capsulatus SB1003 was used as the parent strain in this study. Isogenic mutant strains ΔhbrL, Δirr and Δzur were generated by clean removal of ORFs rcc03370, rcc02670 and rcc01134, respectively (table S1). The general procedure was based on a two-step in-frame deletion protocol as previously described (Swem et al., 2005). Briefly, the deletion plasmids pΔhbrL, pΔirr and pΔzur were constructed by inserting the ~0.5kb upstream and downstream of hbrL, irr and zur respectively into the suicide vector pZJD29a (table S1). While the inserts of pΔhbrL and pΔirr were designed by ligating the two 0.5kb fragments, the insert of pΔzur was generated by fusing the fragments by crossover PCR (Senanayake and Brian, 1995). Primers used to construct the suicide plasmids pΔhbrL, pΔirr and pΔzur are presented in table S2. Complementation of the ΔhbrL strain was not performed because the hbrL gene appears to be at the end of a large operon (>15kbp) that makes effective complementation problematic. HbrL also exhibits an alternative start codon that is yet to be defined.

Strains were grown routinely on PY agar (3g.L⁻¹ peptone, 3g.L⁻¹ yeast extract, 15g.L⁻¹ agar), PYS (3g.L⁻¹ peptone, 3g.L⁻¹ yeast extract, 2mM MgSO₄, 2mM CaCl₂) and RCV* a modified RCV where the carbon source consists of 6g.L⁻¹ D-glucose, 5g.L⁻¹ sodium pyruvate, instead of malate (Weaver et al., 1975). The RCV* medium was adjusted at pH 6.8 and contained, otherwise stated, 50μM FeSO₄. Iron chelator 2,2’-dipyridyl (DIP) was used to generate iron-depleted media at various concentrations. A stock solution of hemin was prepared in heme buffer (50% ethanol vol:vol, 50mM sodium bicarbonate) at 1mM for the heme supplementation experiments. Heme buffer was added to the medium such that the cells were exposed to a constant 5% (vol:vol) of heme buffer irrespective of hemin concentration. Aerobic growth condition consisted in maintaining medium volume at 10% of the maximal flask volume and incubating in a rotary shaker at 200 rpm and 34°C. Likewise, for semi-anaerobic conditions, flasks were filled to 80% of there maximum volume and shaken at 100 rpm. Finally, light-anaerobic (photosynthetic) condition was set by using screw-cap tubes filled to the top and incubated in a ventilated room at 30°C under 80W incandescent light bulbs.

Isolation of suppressor strains ΔhbrL^SUP-1 and ΔhbrL^SUP-2

Prolonged incubation of ΔhbrL in RCV* containing 250μM DIP in semi-anaerobic conditions enabled the generation of ΔhbrL suppressor phenotype. Suppressor ΔhbrL^SUP-1 was isolated by first enriching the DIP-resistant phenotype in diluting the culture in RCV* containing 250μM DIP. Then, liquid culture was plated for colony isolation on PY agar. Isolates were then grown on 250μM DIP containing RCV* to check if the ΔhbrL suppressor phenotype was stable. Only one colony gave such a phenotype so it was stored as suppressor strain ΔhbrL^SUP-1. Suppressor ΔhbrL^SUP-2 consists of a ΔhbrL culture in RCV* containing 250μM DIP that was stored without further isolation. For both suppressor strains, single colonies were used to grow the cell and prepare genomic DNA for sequencing.

Genome sequencing

Genomes of strains ΔhbrL, ΔhbrL^SUP-1, ΔhbrL^SUP-2 were sequenced using the high-throughput DNA sequencing Illumina HiSeq2000 with 100bp single read protocol. The preparation of genomic DNA library for the sequencing was performed according to standard protocol. Briefly, genomic DNA was sheared by nebulization (Invitrogen) and repaired by blunting enzyme mix (NEB). Fragments were then A-tailed using Klenow fragment (NEB) to which adaptors were ligated. The library was finally amplified, assayed and sequenced. Strains ΔhbrL and ΔhbrL^SUP-1 were sequenced using the standard protocol while strain ΔhbrL^SUP-2 was prepared by barcoding the libraries and multiplexing the sequencing reactions.

Gene expression studies

WT and ΔhbrL strains were initially grown for 24h at 34°C in 3mL of RCV* shaken at 100rpm. The semi-anaerobic cultures were then diluted 100-fold into a 40mL final volume of RCV* in 50mL flasks and shaken at 100rpm for 24h at 34°C. Finally, using 50mL flasks, 40mL of RCV* were inoculated to a final OD₆₆₀ of 0.05 and incubated at 34°C shaking at 100rpm until OD₆₆₀ reached 0.1. Then, DIP was added to a final concentration of 100μM or 200μM. Cell samples were harvested right before the addition of DIP and after 2h of incubation (34°C, 100 rpm) in its presence. Cell pellets were stored at −80°C until further use. RNA samples were prepared using the RNA Isolate Mini kit (Bioline) followed by a DNAse treatment using the turbo DNA-free kit (Life Technologies) and storage at −80°C. One-step quantitative RT-PCR was used to determine gene expression levels using the Brilliant SYBR®Green qRT-PCR kit (Agilent Technologies) or the SensiFAST™ SYBR Hi-ROX One-Step kit (Bioline), according to the manufacturer's recommendations using 2ng of RNA per 25μL reaction. The reactions were performed on a StepOnePlus Real-Time PCR system (Life Technologies). Gene expression ratios were calculated using the Paffl method (Pfaffl, 2001) using primers specific to the target genes and rpoZ (rcc03318) that was used as the reference gene (table S2). Each gene expression value was obtained from 6 independent biological replicates, grown on three separate days. Ct were calculated from duplicated qRT-PCR reactions. Statistical analyses of the expression ratios were performed with the Relative Expression Software Tool (REST) (www.gene-quantification.info). The latter determine if differences between samples are significant using a hypothesis test based on randomization of samples to calculate a P(H1) value (Pfaffl et al., 2002). Two thousand random reallocations of samples were performed and sample difference was considered statistically significant when P(H1) < 0.05.