|Home | About | Journals | Submit | Contact Us | Français|
The fission yeast Schizosaccharomyces pombe excretes and accumulates the hydroxamate-type siderophore ferrichrome. The sib1+ and sib2+ genes encode, respectively, a siderophore synthetase and an l-ornithine N5-oxygenase that participate in ferrichrome biosynthesis. In the present report, we demonstrate that sib1+ and sib2+ are repressed by the GATA-type transcriptional repressor Fep1 in response to high levels of iron. We further found that the loss of Fep1 results in increased ferrichrome production. We showed that a sib1Δ sib2Δ mutant strain exhibits a severe growth defect on iron-poor media. We determined that two metabolic pathways are involved in biosynthesis of ornithine, an obligatory precursor of ferrichrome. Ornithine is produced by hydrolysis of arginine by the Car1 and Car3 proteins. Although car3+ was constitutively expressed, car1+ transcription levels were repressed upon exposure to iron, with a concomitant decrease of Car1 arginase activity. Ornithine is also generated by transformation of glutamate, which itself is produced by two separate biosynthetic pathways which are transcriptionally regulated by iron in an opposite fashion. In one pathway, the glutamate dehydrogenase Gdh1, which produces glutamate from 2-ketoglutarate, was repressed under iron-replete conditions in a Fep1-dependent manner. The other pathway involves two coupled enzymes, glutamine synthetase Gln1 and Fe-S cluster-containing glutamate synthase Glt1, which were both repressed under iron-limiting conditions but were expressed under iron-replete conditions. Collectively, these results indicate that under conditions of iron deprivation, yeast remodels metabolic pathways linked to ferrichrome synthesis in order to limit iron utilization without compromising siderophore production and its ability to sequester iron from the environment.
Iron is vital for all eukaryotes. The ability of this metal to exist in two redox states confers properties that make it an essential participant at the active center of several enzymes that are involved in critical cellular processes, such as amino acid biosynthesis, energy production, and lipid metabolism. Paradoxically, the properties that make iron essential can also make it toxic under certain conditions. Excess iron has the ability to generate toxic reactive oxygen species that can damage cellular components (15). Consequently, cells have developed tightly regulated homeostatic mechanisms in order to optimize iron uptake while keeping its reactivity under tight control. Despite the fact that iron is one of the most abundant elements on earth, its bioavailability remains highly limited at physiological pH owing to its oxidation into insoluble ferric hydroxides under atmospheric oxygen conditions (4). To overcome this life-threatening issue, the fission yeast Schizosaccharomyces pombe, as well as most fungi, uses two high-affinity iron acquisition systems that involve either a reductive or a nonreductive mechanism (20).
In S. pombe, the reductive iron acquisition system relies on the ferrireductase Frp1, whose role is to solubilize extracellular iron by reducing its ferric form (Fe3+) to its ferrous form (Fe2+) (42). The subsequent transport of Fe2+ into cells is predominantly mediated by an oxidase-permease complex consisting of Fio1 and Fip1 (2). The nonreductive iron uptake system involves the acquisition of iron-loaded siderophores. Produced by most microorganisms, siderophores are small organic molecules that bind ferric iron with very high affinity and specificity (32). Fission yeast biosynthesizes, accumulates, and excretes only one type of siderophore, which is designated ferrichrome (45). Once loaded with iron, Fe3+-ferrichrome is captured and internalized by S. pombe via the cell surface transporters Str1 and Str2 (36). According to the general biosynthetic pathway of fungal hydroxamate siderophores proposed by Plattner and Diekmann (39), the first step in the biosynthesis of ferrichrome resides in the N5 hydroxylation of ornithine by ornithine-N5-oxygenase. The homology of sequences of the ornithine-N5-oxygenases characterized to date, such as Sid1 (from Ustilago maydis) and SidA (from Aspergillus nidulans), suggest that a similar enzymatic reaction is catalyzed by the product of the sib2+ gene in fission yeast (13). The subsequent step in the biosynthesis of ferrichrome is the formation of the hydroxamate group. Newly synthesized molecules of N5-hydroxyornithine are acylated by an N5-transacetylase, which is predicted to be encoded by the SPBC17G9.06c locus in S. pombe. Ultimately, the hydroxamate group is processed by nonribosomal peptide synthetases (NRPSs) (11, 53). In S. pombe, the NRPS Sib1 catalyzes the last step of ferrichrome biosynthesis (40, 46).
Iron transport in fission yeast is primarily regulated at the transcriptional level by the iron-sensing GATA-type repressor Fep1 (20). When iron is abundant, Fep1 binds to GATA-type cis-acting elements (A/T)GATA(A/T) and downregulates frp1+, fio1+, fip1+, str1+, str2+, str3+, and abc3+ gene expression (18, 35, 36, 41). The strength of the Fep1-mediated transcriptional repression is maximized when target gene promoters harbor the modified GATA-type sequence ATC(A/T)GATA(A/T) (41). Fep1 orthologs in U. maydis (Urbs1) and Histoplasma capsulatum (Sre1) are also known to bind to this motif (1, 7). Another member of the Fep1 regulon is php4+, a gene encoding the iron-responsive CCAAT-binding subunit Php4 (29). During iron starvation, Php4 coordinates the S. pombe iron-sparing response by repressing genes that encode components of iron-requiring metabolic pathways, such as the tricarboxylic acid (TCA) cycle, the electron transport chain, and the iron-sulfur cluster biogenesis machinery (30). Similarly, the Php4 ortholog in A. nidulans, denoted HapX, regulates the expression of genes encoding iron-using proteins (17). At the molecular level, Php4 associates with its target genes by recognition of the CCAAT-binding complex, which is composed of Php2, Php3, and Php5 (28, 29). The Php2/3/5 heterotrimer binds CCAAT cis-acting elements, whereas Php4 lacks DNA-binding activity. It has been demonstrated that the gene encoding the transcriptional repressor Fep1 is regulated by Php4, creating a reciprocal regulatory loop between both iron sensors (30). Therefore, Php4 and Fep1 act as key regulators of iron homeostasis in fission yeast by controlling iron utilization and iron acquisition.
The assembly of ferrichrome requires ornithine, a nonproteinogenic amino acid whose synthesis is tightly associated with nitrogen metabolism. In yeast, more than 80% of cellular nitrogen derives from the incorporation of ammonia into glutamate, a precursor of ornithine (24). Two metabolic branches mediate assimilation of ammonia into glutamate in fission yeast (Fig. (Fig.1).1). One is mediated by the glutamate dehydrogenase Gdh1, while the other consists of a coordinated reaction involving glutamine synthetase (Gln1) and glutamate synthase (Glt1) (37, 38). In the Gln1/Glt1 branch, Gln1 first catalyzes the ATP-dependent incorporation of ammonia into glutamine. The newly acquired amino group of glutamine is then transferred to a 2-ketoglutarate acceptor by Glt1. This reaction ultimately generates two molecules of glutamate. This ammonia assimilation pathway is iron dependent, since Glt1 requires the incorporation of an iron-sulfur cluster for activity (9). Glutamate generated by ammonia assimilation pathways serves in numerous biological processes, including the biosynthesis of arginine, polyamines, and hydroxamate siderophores. An alternative metabolic pathway that produces ornithine originates from the urea cycle. In this pathway, arginase catalyzes the degradation of arginine into urea and ornithine. According to complementation studies using a Saccharomyces cerevisiae arginase null strain (car1Δ), a similar reaction is mediated by the product of the car1+ gene in S. pombe (49).
Here we report that ornithine and ferrichrome are critical to S. pombe survival under conditions of iron deficiency. We further show that the metabolic pathways related to ornithine and ferrichrome synthesis are regulated at the transcriptional level by Fep1 and Php4 as a function of iron availability. Our findings for S. pombe also constitute the first reported case of a microbial organism expressing two genetically distinct arginases, specifically Car1 and Car3. In addition, these arginases were found to be differentially modulated by the cellular iron status.
All S. pombe strains used in this study are described in Table Table1.1. Yeast strains were grown in yeast extract medium containing 0.5% yeast extract and 3% glucose (YE) or in Edinburgh minimal medium (EMM) that was supplemented with 225 mg/liter of adenine, histidine, leucine, uracil, lysine, and arginine, unless otherwise indicated. In the case of liquid cultures, unless otherwise stated, cells were seeded to an A600 of 0.5, grown to exponential phase (A600 of ~1.0), and then treated with either 250 μM 2,2′-dipyridyl (Dip) or 100 μM FeCl3 or left untreated for 90 min. In the case of growth assays on solid medium, cells were spotted on the above-mentioned EMM (control) or EMM containing 75 μM Dip or a combination of 75 μM Dip and 1 μM ferrichrome (FC) unless otherwise stated.
Total RNA was extracted by a hot phenol method (8) and quantified spectrophotometrically. For RNase protection assays, 15 μg of RNA per reaction was used, as described previously (3). DNA templates for antisense riboprobes (Table (Table2)2) were cloned into the BamHI and EcoRI sites of the pBluescript SK vector. The resultant constructs were linearized with BamHI for subsequent antisense RNA labeling with [α-32P]UTP and the T7 RNA polymerase. act1+ mRNA was probed as an internal control for normalization during quantification of the RNase protection products.
S. pombe strains were grown in YE medium supplemented with 225 mg/liter of adenine, histidine, leucine, uracil, and lysine. All cultures were grown to stationary phase. FC was extracted as described previously (16, 31), with the following modifications. Cells (~1 × 109 cells) were resuspended in 250 μl of a saturated ammonium sulfate solution containing FeCl3 (500 μM). Glass beads (250 μl) and a 1:1 phenol-chloroform mixture (500 μl) were added prior to 45 s of lysis at full speed in a FastPrep apparatus. The samples were then centrifuged, and the organic layers were collected. The latter fraction was then diluted with 3 volumes of diethyl ether and vortex mixed with water (150 μl). The aqueous layer containing FC was collected by centrifugation, washed with 1 volume of diethyl ether, and lyophilized. Dried samples were resuspended in 4 μl of water and spotted on Whatman PE SIL G/UV silica gel thin-layer chromatography (TLC) sheets (Whatman, Maidstone, England). The plates were developed in 80% aqueous methanol, and FC was revealed by its reddish color.
Glutamate dehydrogenase (GDH) and glutamine synthetase (GS) activities were assayed as described previously (37), with some modifications. At early logarithmic phase, 50-ml cultures of S. pombe were incubated for 6 h with 250 μM Dip. Cell pellets were resuspended in 300 μl of extraction buffers of the following compositions. GDH buffer contained 67 mM phosphate (KH2PO4-Na2HPO4, pH 7.5), 2 mM EDTA, 2 mM α-ketoglutarate, and 0.2% (wt/vol) β-mercaptoethanol, whereas GS buffer contained 50 mM Tris-HCl, 1 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), and a cocktail of protease inhibitors (P-8340; Sigma-Aldrich), pH 7.5. Glass beads were added to the cell suspensions prior to their mechanical disruption with a FastPrep instrument. Samples were centrifuged for 60 min at 13,000 rpm, and the protein concentration in the supernatants was determined using the Bradford assay (5). Total proteins (100 μg/reaction mixture) were added to GDH and GS enzymatic reaction mixtures. To determine GDH activity, the oxidation rate of NADPH was followed spectrophotometrically at 23°C over 30 s at 340 nm. GDH specific activity was defined as the number of μmoles of NADPH oxidized per min per mg of protein. GS activity was determined by following glutamyl-transferase activity at 40°C over 5 min. l-γ-Glutamyl hydroxamate formation was measured spectrophotometrically at 500 nm. Specific glutamyl transferase activity was given as the number of μmoles of l-γ-glutamyl hydroxamate produced per min per mg of protein. Arginase activity was assayed essentially as described previously (6, 34). Cells were grown in arginine-supplemented rich medium until exponential phase and treated for 6 h with either 250 μM Dip or 100 μM FeCl3. Total protein extraction was performed in CAR buffer (50 mM Tris-HCl, 1 mM MnCl2, 1 mM DTT, 1 mM PMSF, 1× protease inhibitor cocktail, pH 9.5). After mechanical disruption, 10 μg of protein was added to 200-μl reaction mixtures containing 20 μl of l-arginine (0.8 M) and CAR buffer. After a 60-min incubation at 45°C, the enzymatic reactions were stopped by addition of 150 μl of an H2O-sulfuric acid-phosphoric acid (2:1:3) mixture. Urea detection reagent (9% ISPF [1-phenyl-1,2-propanedione-2-oxime or α-isonitrosopropiophenone] in 100% ethanol) was then added to each sample, and the reaction mixture was heated at 100°C for 60 min. After cooling to room temperature in the dark over a period of at least 30 min, urea production was measured spectrophotometrically at 540 nm. Arginase specific activity was defined as the number of μmoles of urea produced per min per mg of protein.
When ornithine serves as a precursor, both the first and the last steps of ferrichrome biosynthesis in fission yeast are mediated by the products of the sib2+ and sib1+ genes, respectively (Fig. (Fig.1)1) (14). Genome profiling studies have revealed that the S. pombe genes sib1+ and sib2+ are induced in response to iron deficiency but are repressed upon addition of iron (30, 43). In contrast, two other reports claim that the ferrichrome biosynthetic process in fission yeast is not differentially regulated as a function of iron availability (45, 46). Thus, the regulatory role of iron with respect to sib1+ and sib2+ gene expression and ferrichrome biosynthesis remains to be clarified. To address this point, we sought to determine whether sib1+ and sib2+ transcript levels were modulated by iron and whether the iron-responsive GATA repressor Fep1 was involved in this process. In order to accomplish this, the promoter regions of both genes were analyzed so as to identify any putative GATA cis-acting elements that could serve as Fep1-binding sites. sib1+ and sib2+ were found to be located close to one another on chromosome 1 of S. pombe and to be separated by a common promoter region 1,497 bp long. Four GATA boxes ([A/T]GATA[A/T]) were identified within this shared promoter region (Fig. (Fig.22 A). We noticed that one of the GATA boxes aligned perfectly with an extended GATA-type sequence, ATC(A/T)GATA(A/T), that is known to mediate the strongest Fep1-dependent repression response (41). This DNA element was located at position −462 relative to the translational initiator codon of sib1+ and at position −1036 with respect to that of sib2+ (Fig. (Fig.2A).2A). Using an RNase protection protocol (29), we monitored sib1+ and sib2+ transcript levels in cells grown in either the absence or the presence of iron or the iron chelator Dip. Both genes were found to be readily expressed in the presence of Dip, whereas their expression strongly decreased under basal or iron-replete conditions (Fig. (Fig.2B).2B). sib1+ and sib2+ mRNA levels were reduced 5.2- and 6.5-fold, respectively, under high-iron conditions compared to levels under iron-limiting conditions (Fig. (Fig.2C).2C). To investigate the role of Fep1, a similar experiment was performed using an isogenic fep1Δ strain. In this case, sib1+ and sib2+ transcript levels in both untreated and iron-treated fep1Δ cells were derepressed and were close to the levels observed in iron-starved wild-type cells (Fig. (Fig.2B).2B). These results confirm the predominant role of Fep1 in the iron-mediated repression of sib1+ and sib2+ expression. Interestingly, in the absence of Fep1, the levels of derepression of sib1+ and sib2+ were slightly higher (~1.5-fold) than those observed under the same iron-depleted conditions in the wild-type strain. These results strongly suggest the presence of a putative unidentified regulator that is more efficient in inducing or maintaining sib1+ and sib2+ gene expression in the absence of Fep1 under iron-limiting conditions.
Based on the results of iron- and Fep1-dependent expression profiles of sib1+ and sib2+ depicted in Fig. Fig.2,2, we reasoned that ferrichrome biosynthesis could be modulated by both iron availability and the iron-responsive GATA factor Fep1. Because sib1+ and sib2+ gene expression was derepressed in a fep1Δ strain (Fig. (Fig.2),2), we sought to determine whether the amount of ferrichrome would increase in the absence of Fep1. fep1Δ cells and isogenic wild-type cells were grown in yeast extract medium for 48 h at 30°C. Ferrichrome was then isolated from whole cells rather than culture medium, since most of the fission yeast siderophore produced accumulates intracellularly (45). Extracts from both strains were analyzed by thin-layer chromatography (TLC). A weak but reproducible ferrichrome signal was observed in the wild-type strain (Fig. (Fig.33 A, lane 1). Remarkably, the ferrichrome signal was stronger in fep1Δ cells, especially compared to wild-type cells (Fig. (Fig.3A,3A, lane 4 and lane 1, respectively). This result indicated that ferrichrome synthesis was upregulated in the absence of Fep1, supporting a regulatory role for Fep1 in the ferrichrome biosynthetic pathway. The specificity of ferrichrome signals detected in both wild-type (fep1+) and fep1Δ cell extracts was confirmed by analyzing extract preparations from cells unable to produce ferrichrome, sib1Δ sib2Δ double and sib1Δ sib2Δ fep1Δ triple mutant strains. Results showed that these mutants did not produce ferrichrome (Fig. (Fig.3A,3A, lanes 2 and 5, respectively).
To determine whether the amount of ferrichrome would increase in iron-starved wild-type cells, ferrichrome was isolated from cells that were grown under low iron versus elevated levels of iron. We determined that the ferrichrome signal was much stronger in extract preparations from cells that had been grown under low-iron conditions (Fig. (Fig.3B).3B). In contrast, we found no significant amount of ferrichrome in cells grown under iron-replete conditions (Fig. (Fig.3B3B).
Although the capacity of S. pombe cells to excrete ferrichrome has been described by Schrettl et al. (45), it has not yet been determined whether ferrichrome excretion by S. pombe significantly contributes to its capacity to acquire exogenous iron and consequently to its ability to grow under iron-limiting conditions. Therefore, we tested whether ferrichrome-deficient cells (sib1Δ sib2Δ) could grow on a solid medium containing Dip (75 μM). As shown in Fig. Fig.3C,3C, sib1Δ sib2Δ double mutant cells exhibited a severe growth defect when grown in iron-depleted medium compared to the isogenic wild-type cells. Importantly, this growth phenotype was fully rescued by supplementing the medium with ferrichrome (1 μM) (Fig. (Fig.3C).3C). Identical results were also observed for the sib1Δ sib2Δ fep1Δ triple mutant strain (data not shown). This finding indicates that ferrichrome biosynthesis contributes significantly to iron assimilation when S. pombe is subjected to iron-deprived conditions.
Ornithine is the precursor of ferrichrome. It is in part produced by the arginase-mediated degradation of arginine, a reaction that also generates urea (Fig. (Fig.1).1). Previous DNA microarray experiments have shown that the S. pombe arginase-encoding gene car1+ is downregulated 2.7-fold in cells treated with exogenous iron compared to expression in cells treated with the iron chelator Dip (30). However, the molecular determinants responsible for this iron-dependent expression profile have not yet been reported. Consequently, the promoter region of car1+ was examined, and six consensus GATA boxes were found (Fig. (Fig.44 A). Two of these harbored the 5′-ATC(T/A)GATA(T/A)-3′ element known to enhance the iron-mediated repression by Fep1 in S. pombe (Fig. (Fig.4A)4A) (41). In fission yeast, the open-reading frame car3+ (SPAC3H1.07) encodes a protein that exhibits high homology with arginase Car1 (87.9% identity and 93.5% similarity). However, there was no 5′-ATC(T/A)GATA(T/A)-3′ element within the car3+ promoter region (data not shown). This observation was in agreement with our previous genome profiling data, suggesting that the expression of car3+ was not modulated by cellular iron levels (30). To further study both arginase genes as a function of iron availability, the car1+ and car3+ transcript levels were analyzed in cells treated with 250 μM Dip, 100 μM FeCl3, or left untreated. A representative RNase protection assay revealed that car1+ steady-state mRNA levels were repressed 3.1-fold in iron-treated wild-type cells in comparison with levels in those treated with Dip (Fig. (Fig.4B).4B). Furthermore, when experiments were conducted under the same conditions using a fep1Δ strain, car1+ transcript levels were still slightly downregulated, 1.5-fold (Fig. 4B and C), suggesting that Fep1 was not the only determinant involved in the iron-dependent regulation of car1+. In contrast, car3+ mRNA levels did not vary significantly in response to changes in iron concentrations or with respect to either the presence or the absence of Fep1 (Fig. 4D and E). These results indicate that both arginase genes in S. pombe are differentially regulated by the iron status of the cell.
The S. pombe arginase-dependent production of ornithine cannot be predicted solely according to the car1+ and car3+ expression profiles. We therefore measured arginase activities of Car1 and Car3 in whole-cell extracts of isogenic wild-type, car1Δ, car3Δ, and car1Δ car3Δ strains grown under low- or high-iron conditions by monitoring urea production. Regardless of the iron status, all extracts possessed arginase activity except in the case of car1Δ car3Δ cell lysate. These observations provided evidence for the functionality of both arginases in S. pombe (Fig. (Fig.55 A). In car3Δ cell extracts, we found that Car1 specific activity was decreased 3.9-fold in cells grown in iron-containing medium compared to that in iron-starved cells (0.54 U/mg, as opposed to 0.14 U/mg). The modulation of Car1 activity by iron reflected the importance of iron-dependent regulation of car1+ gene expression (3.1-fold) (Fig. (Fig.5C).5C). As predicted from the constitutive expression of car3+ (Fig. (Fig.4D),4D), Car3 activity assayed in car1Δ cell lysates was not affected by the cellular iron status (Fig. (Fig.5A).5A). Surprisingly and despite a 3.9-fold upregulation of Car1 activity during iron deficiency, total arginase activity in wild-type cell extracts did not vary significantly in response to conditions where Dip or FeCl3 was present. In fact, total arginase activity in wild-type cell extracts (Dip or Fe) did not match the sum of Car1 and Car3 arginase activities determined separately in car1Δ and car3Δ cell lysates. We therefore hypothesized that the absence of one of the two arginases may induce the expression of its paralog and upregulate its activity. To address this point, we investigated the car1+ expression profile in a car3Δ deletion strain, since car3+ appeared to be the most expressed and active arginase (Fig. (Fig.4D4D and and5A).5A). Results showed that the disruption of car3+ (car3Δ) led to a marked upregulation (~5-fold) of the car1+ transcript levels under all conditions tested (Fig. (Fig.5C).5C). This result suggested the existence of a homeostatic regulatory mechanism that modulates car1+ expression in response to a lack of Car3-mediated arginase activity. At this point, however, it is unclear whether such a mechanism operates on car3+ since its expression did not vary significantly in a car1Δ deletion strain (data not shown). Thus, owing to a constitutive induction of car1+ in the absence of Car3, the activity of Car1 observed in car3Δ cell extracts (Fig. (Fig.5A)5A) had most likely been overestimated. Nevertheless, the magnitude of the iron-mediated regulation of car1+ gene expression (3.3-fold) and Car1 activity (3.9-fold) in the absence of Car3 (Fig. (Fig.5)5) was similar to that observed for car1+ transcript levels in wild-type cells (3.1-fold) (Fig. 4B and C). With respect to the contribution of Fep1 to this regulatory process, the results obtained with the car3Δ fep1Δ double disruption strain revealed that the Fep1-dependent repression of car1+ was almost negligible in the absence of Car3 (Fig. 5C and D). Under either basal or high-iron conditions, the car1+ transcript levels were similar in the car3Δ and car3Δ fep1Δ strains. Accordingly, we found no significant difference with respect to the iron-mediated repression of Car1 activity in extracts of car3Δ and car3Δ fep1Δ cells (Fig. (Fig.5B).5B). Collectively, these results demonstrate for the first time that S. pombe expresses two functional paralogous arginases (Car1 and Car3) and that Car1 activity is regulated as a function of iron availability. Furthermore, these results indicate that regardless of the contribution of Fep1 to the iron regulation of car1+, factors that remain undefined participate in the iron-dependent modulation of Car1 arginase activity.
Glutamate biosynthesis depends on two nitrogen-utilization pathways in S. pombe (Fig. (Fig.1).1). The first pathway involves reductive amination of 2-ketoglutarate by the glutamate dehydrogenase Gdh1 (37). The second pathway involves two coupled enzymatic reactions driven by the glutamine synthetase Gln1 and the Fe-S cluster-containing glutamate synthase Glt1, respectively (Fig. (Fig.1)1) (38). Genome profiling data have suggested that both pathways are transcriptionally regulated by iron, although in an opposite fashion (30). For example, the transcript levels of gdh1+ are upregulated 3.2-fold under conditions of iron starvation, whereas the levels of gln1+ and glt1+ are repressed 3.0- and 4.4-fold, respectively, under these conditions (30).
Based on the findings of the iron-regulated expression pattern of gdh1+ and its metabolic relationship with ferrichrome biosynthesis, we hypothesized that gdh1+ transcription was controlled by the GATA-type transcription factor Fep1. The promoter region of the gdh1+ gene contains 10 putative GATA elements [(A/T)GATA(A/T)] (Fig. (Fig.66 A). One of these (position −571 relative to the A of the gdh1+ initiator codon) harbors an ATCTGATAA motif that corresponds to a strong Fep1-responsive GATA element [ATC(A/T)GATA(A/T)] (41). On the other hand, the gln1+ and glt1+ promoter regions contain several CCAAT sequences that may be recognized by the CCAAT-binding factor, which is composed of Php2, Php3, Php5, and Php4 (Fig. (Fig.6A).6A). Interestingly, the CCAAT boxes were found to be highly clustered within the gln1+ and glt1+ promoter regions. It must be added that the biological relevance of this observation has not been investigated. To validate microarray data (30) in an independent manner, we examined the steady-state transcript levels of gdh1+, gln1+, and glt1+ in wild-type cells as a function of iron availability. Using RNase protection assays, the expression of gdh1+ was found to be repressed 2.1-fold in untreated and iron-treated cells in comparison with that in iron-starved cells (Fig. 6B and C). This regulation of gdh1+ by iron was solely dependent on the iron-responsive repressor Fep1, because gdh1+ mRNA levels became constitutively upregulated in a fep1Δ strain (Fig. 6B and C). On the other hand, under Fe-limiting conditions, gln1+ and glt1+ mRNA levels were repressed 3.1- and 10.0-fold, respectively, compared to their basal levels of expression in untreated cells (Fig. 6D to G). To examine the role of Php4 in the regulation of both genes, similar experiments were conducted using a php4Δ strain. Results showed that gln1+ and glt1+ mRNA levels were invariably expressed in the absence of Php4 (Fig. 6E and G). These findings supported the interpretation of an essential role for Php4 in the iron limitation-dependent repression of gln1+ and glt1+ expression.
We next measured glutamate dehydrogenase (Gdh) and glutamine synthetase (Gln) activities to confirm that both the Gdh- and the Gln1/Glt-dependent glutamate synthesis pathways were modulated by the cellular iron status, as suggested from transcriptional data (Fig. (Fig.6).6). S. pombe cells were grown to mid-logarithmic phase and subjected to iron-deprived or iron-replete conditions. Gdh activity was followed spectrophotometrically as a function of NADPH oxidation (37). Total extracts of iron-starved wild-type cells exhibited a 1.7-fold increase in Gdh activity (0.098 U/mg) compared to lysates of iron-treated cells (0.058 U/mg) (Fig. (Fig.77 A). In contrast, Gdh1 activity was upregulated and unresponsive to the cellular iron status when measured in the cell extracts derived from a fep1Δ mutant strain (Fig. (Fig.7A).7A). These results indicate that the Gdh-mediated glutamate synthesis pathway was modulated by cellular iron levels. Furthermore, the results presented in Fig. Fig.7A7A showed that Fep1 was required for downregulation of Gdh activity in the presence of iron. Subsequently, we sought to determine whether Gln1 activity was regulated by iron in an opposite fashion to Gdh activity. Based on the glutamyl transferase reaction, Gln activity was measured as described previously (37). Total lysates prepared from wild-type cells grown under low-iron conditions contained 2.0-fold less Gln activity than lysates derived from iron-treated cells (0.98 U/mg as opposed to 2.00 U/mg) (Fig. (Fig.7B).7B). In contrast, lysates prepared from an isogenic php4Δ mutant strain exhibited constitutively elevated Gln activity levels that were not repressed under conditions of iron depletion (Fig. (Fig.7B).7B). The results indicated that Php4 must be active to allow the repression of Gln activity under conditions of iron deprivation. Attempts to monitor the Glt activity spectrophotometrically (37) did not permit any conclusions with respect to Glt activity in either the presence or the absence of iron or Php4, since the control samples showed very high background levels of NADH oxidase activity (data not shown). Taken together, these results demonstrate that the nitrogen-dependent synthesis of glutamate (a mitochondrial precursor of ornithine) is an iron-regulated metabolic process in the fission yeast S. pombe. Under conditions of iron starvation, Php4 represses the Gln1/Glt1 pathway, whereas the concomitant inhibition of Fep1 leads to an activation of the Gdh1-dependent pathway.
Ornithine is the precursor of hydroxamate siderophore-like ferrichrome (39). In yeast, although it is known that ornithine biosynthesis can originate from either mitochondrial glutamate or cytosolic arginine (10), the relative contribution of each ornithine biosynthetic pathway to ferrichrome assembly has not yet been ascertained. Based on the observation that S. pombe sib1Δ sib2Δ (ferrichrome-deficient) cells were hypersensitive to iron starvation (Fig. (Fig.3B),3B), we sought to determine whether cells defective in each or both ornithine biosynthetic pathways were phenocopies of sib1Δ sib2Δ cells. The disruption of the mitochondrial ornithine synthesis pathway was accomplished by inactivation of arg1+, which encodes an acetylornithine aminotransferase essential to mitochondrial ornithine production (23). The car1Δ car3Δ double mutant strain described above (Fig. (Fig.5A)5A) was used for inhibition of cytosolic ornithine synthesis. Mitochondrial and cytosolic ornithine synthesis was abolished by inactivation of the arg1+ gene in the car1Δ car3Δ double mutant strain. The arginine auxotrophy of the arg1Δ and arg1Δ car1Δ car3Δ strains was validated by their inability to grow on arginine-free minimal medium (Fig. (Fig.88 A). arg1Δ mutant cells, which are defective only in mitochondrial ornithine synthesis, exhibited weak growth on medium containing the iron chelator Dip compared to that of wild-type cells (Fig. (Fig.8A).8A). In contrast, cells harboring a car1Δ car3Δ double deletion (which are defective only in cytosolic ornithine synthesis) grew at the same level as the wild-type strain grown in the presence of Dip. When both mitochondrial and cytosolic ornithine biosynthetic pathways were disrupted using an arg1Δ car1Δ car3Δ triple mutant, cell growth was defective in medium containing Dip (75 μM). The iron starvation-dependent growth defect was reversed by addition of purified Fe3+-ferrichrome to the growth medium (Fig. (Fig.8A).8A). The result confirmed the metabolic connection between ornithine and ferrichrome synthesis (Fig. (Fig.8B).8B). As controls, yeast strains of the indicated relevant genotypes (wild type, arg1Δ, car1Δ car3Δ, and arg1Δ car1Δ car3Δ) were grown on complete EMM (Fig. (Fig.8A,8A, control). Collectively, these results indicated that under low-iron conditions, mitochondrial ornithine was preferentially used for ferrichrome production, although siderophore synthesis also occurred in the cytosol. Nevertheless, the cytosolic Car1- and Car3-dependent supply of ornithine was not negligible, since the loss of both ornithine biosynthetic pathways (arg1Δ car1Δ car3Δ) made the cells sensitive to iron deprivation (Fig. (Fig.8).8). Overall, the data revealed that a global ornithine shortage results in an acute sensitivity to iron deficiency, a phenotype that can be attributed to a lack of ferrichrome production.
Studies of yeast have revealed that the biosynthesis of siderophores is positively regulated during iron deficiency (14, 19, 33, 50). Previous reports had suggested a constitutive production of ferrichrome in S. pombe (45, 46). However, recent genome profiling data have shown that sib1+ and sib2+, two S. pombe ferrichrome biosynthetic genes, are induced in response to iron deprivation (30, 43). Microarray data have further indicated that additional metabolic pathways related to ferrichrome production were down- or upregulated under conditions of iron starvation. Here we have investigated the iron-dependent expression profiles of sib1+, sib2+, car1+, car3+, gdh1+, glt1+, and gln1+, whose gene products are linked to ferrichrome biosynthesis by way of its ornithine precursor. We have also examined the relative contributions of some of these genes to ferrichrome biosynthesis and their contributions to S. pombe growth resistance under low-iron conditions.
We found that sib1+ and sib2+ gene expression was transcriptionally regulated by the GATA-type iron-responsive repressor Fep1 as a function of iron availability. Under iron-replete conditions, the transcript levels of both sib1+ and sib2+ were repressed in a Fep1-dependent manner. The loss of Fep1 (fep1Δ) resulted in an elevated and constitutive transcriptional expression of the sib1+ and sib2+ genes. Consistent with these results, ferrichrome production was enhanced in extracts of fep1Δ cells grown in an iron-sufficient medium. In the current study, the cell-permeating iron chelator Dip was used to investigate gene expression profiles under conditions of iron deprivation. This chelator creates an intensive cellular iron shortage, which results in the release of Fep1 from its promoters (18) and the derepression of target genes. The use of mild iron deprivation conditions and the lack of a fep1Δ mutant strain, as done in previous studies (45, 46), may explain why those authors did not observe that siderophore biosynthesis was regulated at the mRNA level.
One intriguing observation was an additional increase in the steady-state mRNA levels of sib1+, sib2+, and car1+ in iron-starved fep1Δ cells. The data suggest that a putative trans-acting regulator operates with a strong efficacy in the absence of Fep1 in response to iron deficiency. Because similar results have been reported for A. nidulans (for the Fep1 ortholog SreA), Candida albicans (Sfu1), and Aspergillus fumigatus (SreA) (21, 44), the identification of the putative transcriptional activator that allows the expression of the genes required for iron acquisition under low-iron conditions merits further study since it may be conserved among many filamentous fungal species. One phenotype associated with a sib1Δ sib2Δ double deletion is hypersensitivity to iron starvation, indicating that ferrichrome plays an important role in iron assimilation in fission yeast. Since ferrichrome produced by S. pombe is found intracellularly and extracellularly (45), the hypersensitivity of sib1Δ sib2Δ cells to iron deprivation may be a consequence of either a defective iron uptake process, an impaired ability to store iron in a usable form, or an inability to deliver iron from one intracellular compartment to another (51). On the other hand, we have determined that in the presence of 100 μM iron, the wild-type strain, as well as the sib1Δ sib2Δ, fep1Δ, and sib1Δ sib2Δ fep1Δ disruption strains, exhibited 100% survival of cells (data not shown). Under these conditions, we observed that the sib1Δ sib2Δ double mutant exhibited increased sensitivity to H2O2 (1 mM) with 41% growth inhibition. The fep1Δ mutant, which is known to overaccumulate iron intracellularly (35, 41), was also sensitive to H2O2, exhibiting 62% growth inhibition. The magnitude of H2O2 sensitivity was further increased for the sib1Δ sib2Δ fep1Δ triple mutant strain, with 90% growth inhibition (data not shown). Thus, when environmental iron levels are elevated, ferrichrome appears to play a physiological function in protecting yeast cells against oxidative stress toxicity. In support of these experimental observations, similar data for U. maydis have been previously reported by the Leong laboratory (22). Future experiments will be needed to clarify this issue.
Ferrichrome biosynthesis requires ornithine (39), which is in part synthesized from the arginase-dependent degradation of arginine. Here we have presented evidence that S. pombe expresses two functional paralogous arginases, Car1 and Car3. The saprophytic fungus Neurospora crassa is the only other microbial organism known to express two arginases (25). However, in contrast to the case with N. crassa, which expresses both enzymes from the same locus through alternative initiator codons, car1+ and car3+ are genetically distinct in S. pombe. Car1 activity was found to be downregulated in response to high concentrations of iron, whereas Car3 activity appears to be constitutive. This Fe-mediated control of Car1 activity is most likely due to its iron starvation-dependent transcriptional expression. Although Fep1 is involved in this regulatory process, we found that an undefined second iron-mediated regulatory mechanism was also involved in the absence of Car3 (car3Δ) (Fig. (Fig.5).5). Alternatively to a direct regulation by iron availability, the induction of car1+ under low-iron conditions may be linked to the rising need for ornithine, which is triggered by the upregulation of ferrichrome synthesis. Interestingly, S. pombe genome profiling studies have indicated that the differential regulation of car1+ and car3+ is not strictly modulated by either iron or ornithine availability. Oxidative stress, nitrogen source starvation, and entry into meiosis have been shown to induce the transcription of the car1+ gene (8, 26, 52). A similar situation occurs in N. crassa, since both arginases are differentially expressed according to metabolic or developmental determinants (e.g., upon the availability of nitrogen sources or entry into conidial germination) (48). According the results presented here, the transcriptional activation of car1+ under low-iron conditions does not seem to have a significant effect on total arginase activity. However, since Car1 and Car3 are almost identical and since arginases are known to self-assemble as multimeric enzymes (12), the biological impact of car1+ induction may reside in the quaternary structure of these arginase multimers. Although the nature of arginase complexes in fission yeast remains to be elucidated, one could speculate that iron deprivation alters the Car1/Car3 subunit ratio of these complexes, creating specialized arginase oligomeric structures that may deliver ornithine more efficiently to the ferrichrome synthesis machinery. If urea production in our experiments had been altered by urease activity as a function of iron availability, this would have been detectable for each genetic background assayed. Clearly, however, it was not the case (Fig. (Fig.5A).5A). We found that urea production in the wild-type strain and in the car1Δ strain was not modulated by iron, whereas it was regulated by iron in the car3Δ strain. Urea accumulation was seen only when Car1 activity was upregulated in response to iron deficiency (in the car3Δ strain). Consistently, the marked increase of Car1 activity (detected through urea accumulation) was detected in a manner that parallels the magnitude of car1+ gene induction.
In addition to its arginase-dependent biosynthetic pathway, ornithine is produced from glutamate (Fig. (Fig.1).1). This study unveils a Php4- and Fep1-dependent remodeling of the nitrogen-dependent biosynthetic glutamate pathways in response to iron deficiency. The optimization of cellular iron utilization is coordinated by Php4, which shuts off metabolic branches that consume iron during periods of iron deprivation (29, 30). Here we determined that Php4 represses the iron-dependent and ATP-requiring Glt1/Gln1 pathway. Hence, considering that glt1+ encodes a Fe-S cluster-containing enzyme, it is likely that S. pombe prevents its futile expression when iron levels are low. Similarly, a repression of glutamate synthase (Glt1) expression has been observed in the budding yeast Saccharomyces cerevisiae (47). However, the molecular determinants that trigger this iron-sparing response are still unknown. The glutamine synthetase Gln1 catalyzes the ATP-dependent formation of glutamine, which is the substrate of the glutamate synthase Glt1. It would therefore be homeostatically relevant to repress gln1+ together with glt1+ to avoid a superfluous production of glutamine. Furthermore, this would prevent unnecessary ATP consumption, especially considering that ATP is made through the iron-requiring oxidative phosphorylation pathway, a cellular process already known to be negatively regulated by Php4 (30). Consistent with the iron starvation-mediated repression of the Gln1/Glt1 pathway, the inactivation of Fep1 by iron deprivation resulted in the induction of the Gdh1 pathway. Increased Gdh1 activity would therefore favor an iron-independent glutamate synthesis and complement the loss of glutamate production inherent to the Php4-dependent repression of the Gln1/Glt1 pathway. A glutamate shortage during iron deficiency could be deleterious to S. pombe, since it would compromise both ornithine synthesis and ferrichrome production. This is underscored by the observation that an arg1Δ mutant strain, which cannot convert glutamate into ornithine, is hypersensitive to iron deprivation (Fig. (Fig.88).
We showed that cells devoid of measurable arginase activity (car1Δ car3Δ) were nearly insensitive to iron starvation in comparison to cells that were unable to synthesize ornithine (car1Δ car3Δ arg1Δ). These results indicate that the biosynthesis of ornithine from glutamate can sustain both ferrichrome production and cell survival without any ornithine supply from the arginases of the urea cycle. It further implies that the ornithine produced from glutamate in the mitochondria is exported and made available to the cytosolic ferrichrome biosynthesis machinery (Fig. (Fig.8).8). Consistently, in both A. nidulans and A. fumigatus, iron deprivation results in the induction of the mitochondrial ornithine carrier AmcA along with siderophore production (33, 44). It has been proposed that the upregulation of AmcA ensures an adequate mitochondrial ornithine delivery to the siderophore synthesis machinery (33). BLAST analyses suggest that AmcA is conserved in several fungal species (data not shown). However, protein similarity searches revealed that S. pombe lacks a mitochondrial ornithine carrier orthologous to AmcA. Because mitochondrial ornithine largely contributes to survival of S. pombe during iron starvation, it is likely that an unidentified ornithine carrier mediates the mitochondrial ornithine efflux required to ensure ferrichrome biosynthesis when fission yeast is forced to grow under conditions of iron deficiency.
We are grateful to Gilles Dupuis and William Home for critically reading the manuscript. We thank Francis Gaudreault for excellent technical assistance and the members of Éric Massé's laboratory for advice on ferrichrome extraction and TLC analysis.
A.M. (Ph.D. student) and S.L. (senior investigator) are recipients of scholarships from the Fonds de la Recherche en Santé du Québec (FRSQ). This study was supported by a Natural Sciences and Engineering Research Council of Canada grant (MOP-238238-2010) to S.L.
Published ahead of print on 30 April 2010.