Reversible Oxidation of a Conserved Methionine in the Nuclear Export Sequence Determines Subcellular Distribution and Activity of the Fungal Nitrate Regulator NirA

The assimilation of nitrate, a most important soil nitrogen source, is tightly regulated in microorganisms and plants. In Aspergillus nidulans, during the transcriptional activation process of nitrate assimilatory genes, the interaction between the pathway-specific transcription factor NirA and the exportin KapK/CRM1 is disrupted, and this leads to rapid nuclear accumulation and transcriptional activity of NirA. In this work by mass spectrometry, we found that in the absence of nitrate, when NirA is inactive and predominantly cytosolic, methionine 169 in the nuclear export sequence (NES) is oxidized to methionine sulfoxide (Metox169). This oxidation depends on FmoB, a flavin-containing monooxygenase which in vitro uses methionine and cysteine, but not glutathione, as oxidation substrates. The function of FmoB cannot be replaced by alternative Fmo proteins present in A. nidulans. Exposure of A. nidulans cells to nitrate led to rapid reduction of NirA-Metox169 to Met169; this reduction being independent from thioredoxin and classical methionine sulfoxide reductases. Replacement of Met169 by isoleucine, a sterically similar but not oxidizable residue, led to partial loss of NirA activity and insensitivity to FmoB-mediated nuclear export. In contrast, replacement of Met169 by alanine transformed the protein into a permanently nuclear and active transcription factor. Co-immunoprecipitation analysis of NirA-KapK interactions and subcellular localization studies of NirA mutants lacking different parts of the protein provided evidence that Met169 oxidation leads to a change in NirA conformation. Based on these results we propose that in the presence of nitrate the activation domain is exposed, but the NES is masked by a central portion of the protein (termed nitrate responsive domain, NiRD), thus restricting active NirA molecules to the nucleus. In the absence of nitrate, Met169 in the NES is oxidized by an FmoB-dependent process leading to loss of protection by the NiRD, NES exposure, and relocation of the inactive NirA to the cytosol.


Introduction
Nitrate is an important nitrogen source for fungi in natural environments. Most species of this kingdom possess an efficient enzymatic and regulatory system that allows conversion of nitrate to nitrite and further to ammonium, which is then incorporated into amino acids and other metabolites [1,2]. Nitrate represents the major soluble nitrogen form in soils and, besides serving as nutrient, also influences plant development [3][4][5], virulence of phytopathogenic fungi [6,7] and the production of fungal secondary metabolites [8,9]. Thus, elucidation of the molecular mechanisms underlying nitrate signalling in A. nidulans may serve as a model for other nitrate assimilating eukaryotes such as algae and plants. Marchive and colleagues [10] have shown that NLP7, the Arabidopsis thaliana nitrate-responsive transcription factor shuttles between the cytosol and the nucleus in response to nitrate availability in a similar way to NirA in A. nidulans.
Activation of the nitrate-specific transcription factor NirA in the ascomycete Aspergillus nidulans is a process which involves both nuclear retention of NirA and its conversion to a functional activator [11]. We previously found that intracellular nitrate or nitrite leads to disruption of the interaction between the nuclear export sequence (NES) of NirA and the specific exportin KapK, the CRM1 homologue in A. nidulans [12][13][14][15][16]. As a result NirA accumulates in the nucleus within less than a minute after the addition of nitrate (see S1 Video), and is subsequently able to bind to the UAS (upstream activating sequences) of genes involved in nitrate assimilation [17]. NirA target genes are only activated when nitrate is present and, at the same time, the intracellular concentration of glutamine, a crucial intermediate in nitrogen assimilation, is low [18]. NirA acts synergistically with the GATA-factor and glutamine sensor AreA to recruit chromatin acetylation and nucleosome remodelling activities [19][20][21][22]. NirA-AreA synergism leads to a rapid transcriptional activation of around 100 genes, among them those required for nitrate reduction and incorporation of the resulting ammonium into glutamate and glutamine. Upstream of these genes NirA and AreA binding sites are present. Genes involved in nitric oxide metabolism are also induced by nitrate but, interestingly, this process does only require NirA, but not AreA [18].
Our previous work established that nuclear accumulation, resulting from leptomycin B (LMB)-mediated inactivation of KapK, is not sufficient to activate NirA [11]. Thus, nitrateinduced activation of NirA involves at least two steps, i.e. release of KapK interaction resulting in nuclear accumulation and acquisition of transcriptional activation competence. In the nirA c 1 allele glycine 167 within the NES of NirA is replaced by a valine (G167V). This mutation not only leads to permanent, nitrate-independent NirA nuclear retention but also to constitutive binding to its DNA target-sequences and transcriptional activation of the cognate regulated genes. Conversely, replacement of consensus leucines in the NES required for KapK interaction also leads to constitutive nuclear accumulation but the protein is not functional [11]. These findings suggest that the NES region of NirA not only determines the subcellular distribution of the transcription factor but also its ability to act as activator protein. The rapid response of NirA nuclear accumulation to nitrate induction prompted us to investigate posttranslational modifications (PTMs) as possible regulatory switches between cytosolic-inactive and nuclear-active NirA. To identify possible PTMs associated with this activation-inactivation cycle we affinity-purified NirA from non-induced and short-term (3 minutes) nitrate-induced cells and analysed the resulting proteins by mass spectrometry (MS). In this article we report that the oxidation status of a conserved methionine within the NirA NES is correlated with the subcellular distribution and transcriptional competence of the protein. We investigated the cognate oxido-reduction switch and based on genetic and biochemical data, we propose a role of an hitherto uncharacterised central domain of NirA which, interacting with the C-terminal activation domain, determines the accessibility of the NES in response to the oxido-reduction switch.

Genetic constructs used in this work
The constitutive expression of the nirA gene is too low to allow biochemical analyses and cell localisation studies using GFP fusions. In previous work, expression was driven by constitutive (gpdA) or inducible ERE or alcA promoters [11,17]. Overexpression does not alter the response of NirA to regulatory signals [22] and thus we used these constructs in the work presented here. Western blots of the different NirA-GFP (expressed from gpdA or ERE promoters) or FLAG-NirA (expressed from the alcA promoter) constructs are shown in S5A and S5B

Met169 is present as its sulfoxide in the absence of nitrate
The NirA NES includes a highly conserved methionine (M169). This residue is present in 57 out of 68 known or predicted NirA orthologs, the only substitution detected being isoleucine, a residue of similar length and hydrophobicity, but not containing an oxidizable sulphur atom.
An extended alignment report can be found in B-Link BLAST results for NirA orthologs following http://www.ncbi.nlm.nih.gov/protein/259489709 (note: A. oryzae protein XP001824047 appears in this alignment based on the conserved binuclear Zn cluster domain but it is not a NirA orthologue). The conservation of M169 in the export sequences of NirA orthologues and the fact that the G167V exchange in NirA c 1 leads to permanently nuclear and active NirA [11] suggests that the region bearing the NES significantly contributes to the regulatory characteristics and nitrate-responsiveness of NirA.
We analysed FLAG-tagged NirA obtained by DNA affinity-purification from cells grown on non-inducing (NI, 3 mM arginine) or nitrate inducing (IND, 10mM NaNO 3 ) conditions by tandem mass spectrometry. In the absence of NO 3 -, the NES of NirA is modified by oxidation of the conserved methionine (Met169) to methionine sulfoxide (Met ox 169). When cells were exposed to nitrate for five minutes Met ox 169 could not be detected any longer (Fig 1 and  Table 1). This rapid response of A. nidulans cells to the inducer coincides with the swift nuclear accumulation of NirA (S1 Video). No Met169 oxidation could be detected in the NirA c 1 protein ( Table 1 and S6 Fig). These MS results strongly suggested a functional link between the oxidation status of M169 and intracellular localisation of NirA.
To exclude that methionine oxidation was an artefact occurring during extract preparation we included serum albumin in all samples before cell disruption, and checked the oxidation status of methionines. No significant oxidation of albumin methionine was detected demonstrating that the NirA NES Met169 oxidation isolated from NI cells condition was conditionspecific. Absence of Met ox 169 in induced wild type and non-induced nirA c 1 mutant cells  Methionine can form two diastereomeric sulfoxides which are individually reduced to methionine by MsrA (Met ox -S reductase) and MsrB (Met ox -R reductase), respectively [23,24]. In a previous work we studied this thioredoxin-dependent pathway in A. nidulans [25] and the availability of a double mrsAΔmrsBΔ and a trxAΔ (thioredoxin-deficient) strain allowed us to test directly if Met ox 169 reduction of NirA depends on MsrA, MsrB or thioredoxin [25,26]. No significant requirement of these proteins for Met ox 169 reduction or NirA activity was found (see Table 1 and S1 Fig). MS analysis of FLAG-NirA in the mrsAΔmrsBΔ strain revealed a mixed pattern of Met169 and Met ox 169 under both NI and IND conditions and we noticed that in these proteins also other methionines (e.g. M719 in the NirA C-terminus) appear in the sulfoxide form. Moreover, the thioredoxin-deficient trxAΔ strain showed no reduction in NirA activity as tested by growth tests [25] and niiA transcription in the deletion strain (S1C Fig).
These data indicate that absence of MsrA and MsrB leads to non-specific protein damage [25] but Met ox 169 reduction still occurs. The partial oxidative damage of NirA could explain the slightly reduced transcriptional activation capacity of NirA found in the msrA msrB double deletion strain (S1B Fig). Because FLAG-NirA carries reduced M169 in the msrA msrB double deletion background it remains to be determined which activities would be required for the very rapid NirA-NES Met ox 169 reduction associated with activation of NirA by nitrate.
As methionine oxidation could generally affect NirA protein stability under non-induced conditions, we tested the amounts of FLAG-NirA in crude extracts obtained from argininegrown (NI) or NO 3 grown (IND) cells. We found that NirA levels were basically equal under both conditions and this result indicates that overall stability of NirA is not influenced by the Met169 oxidation state (Fig 2C).
Met ox 169 formation and NirA nuclear export depends on FmoB, a flavincontaining monooxygenase Flavin-containing monooxygenases (FMOs) have been shown to oxidize peptide-bound methionines [27]. In the A. nidulans genome at least seven putative FMO-encoding genes were  the deletion strains. Met169 status and subcellular localization of NirA-GFP were unchanged in fmoAΔ cells but Met ox 169 was not detected any longer in the fmoBΔ strain under any conditions (Table 1 and S6 Fig). We subsequently tested subcellular localization of FmoB by fusing fmoB to gfp. Unfortunately, this fusion, expressed from the weak homologous fmoB promoter, did not result in clear GFP labelling of cellular compartments and we thus expressed the GFP-tagged version of FmoB from the strong gpdA promoter. The resulting fusion protein was found to localize all over the cell without particular accumulation in any compartment ( Fig 3A). Confocal microscopy of NirA-GFP however, showed that the transcription factor was constitutively nuclear in the fmoBΔ background (Fig 2A). This result is consistent with M169 oxidation in the NES under NI conditions having a specific role for NirA localization and function. However, despite the constitutive nuclear localisation of NirA, the fmoBΔ background does not lead to constitutive expression of NirA-dependent genes ( Fig 2B). Induction by nitrate, on the other hand, occurred as in the wild type and may even be slightly enhanced (+70%) in the fmoB mutant ( Fig 2B). These results confirm that the oxidation status of NirA-NES is determined specifically by FmoB, one out of seven flavin-containing monooxygenases predicted in the A. nidulans genome. These data also confirm that the NES oxidation status controls the nuclear-cytosolic shuttling process but not necessarily the transcriptional activity of NirA.
FmoB can oxidize free methionine and cysteine but does not oxidise Met169 in a synthetic NES peptide We next determined the NADPH-dependent enzymatic properties of FmoB and expressed the complete fmoB gene in E. coli (Fig 3B and 3C). In the absence of any substrate, FmoB displays a low NADPH-oxidase activity leading to gradual decrease in NADPH levels of roughly 30% over 12 minutes. Our assays and subsequent HPLC analysis demonstrated that recombinant FmoB is a functional methionine oxidase as it is able to convert free methionine to the sulfoxide form (Met ox ). However, the rate appears to be low since only 100 nmoles Met ox were formed during 5 minutes at a concentration of 5 mM free methionine ( Fig 3D). The same low activity was observed with methimazole, which is another previously characterized FMO substrate [28]. NADPH oxidation in the presence of methimazole was hardly increased over the NADPH consumption background without substrate and increasing the amounts of substrates from 0.25 mM to 2 mM only marginally increased the NADPH consumption ( Fig 3E). In contrast, other thiols such as free cysteine, dithiothreitol (DTT) or β-mercaptoethanol were favourable substrates for FmoB under our experimental conditions but, surprisingly, glutathione with its peptide-bound cysteine was not ( Fig 3F). We next tested if recombinant FmoB is able to oxidize Met169 in a synthetic peptide carrying the NirA-NES. We used an 18 amino acid peptide containing the NirA-NES (DQFESELAGKMSNLVLDG) as substrate but could not detect any enzyme-dependent NADPH oxidation. These biochemical data suggest that FmoB is either not a peptide methionine oxidase or that it requires additional NirA sequences to recognize the NES and to oxidize Met169 in NirA. Alternatively or additionally, FmoB function may diminish the pool of reduced aliphatic thiols (such as free cysteines) and thereby indirectly promote Met169 oxidation. Taken together, our biochemical and genetic characterization suggests that FmoB is a functional thiol oxidase directly or indirectly required for Met169 oxidation and relocation of NirA to the cytosol.
FmoB is involved in forcing NirA nuclear exclusion in nitrate-induced cells N-octylamine (abbreviated NOC in the text or "o" in figure labellings) is a compound which accelerates some FMO-mediated processes [29]. We thus tested whether purified FmoB responds to NOC. Addition of 1mM or 2 mM NOC to the in vitro reaction mixture noticeably increases the NADPH-oxidase activity of the enzyme, whereas higher concentrations were inhibitory ( Fig 3G). When we tested NOC effects in vivo we found a striking effect. NirA-GFP was rapidly exported from the nucleus after addition of 10 mM NOC to NO 3 induced wild type cells and this NOC-triggered exclusion did not occur in fmoBΔ cells (Fig 2A). To exclude the possibility that the observed cytosolic staining was simply a consequence of NirA degradation and subsequent random distribution of GFP within the cells we tested NirA stability after NOC treatment by Western blotting. Extracts prepared under denaturing conditions (TCA preparation to avoid non-specific degradation during processing) from FLAG-NirA cells grown under NI and IND condition in the presence and absence of NOC showed that NirA was not degraded during treatment with this chemical (Fig 2C). We also wanted to exclude the possibility that NOC treatment leads to a defect in nuclear integrity and subsequent random GFP distribution. The results demonstrate that this is not the case as NOC did not lead to cytoplasmic localization of NirA-GFP in a fmoBΔ strain (Fig 2A), nor did it lead to loss of nuclear positioning of a histone H1-mRFP fusion protein (strain hhoA-mrfp, Fig 4B).
The KapK exportin and Met169 in the NirA NES are required for the FmoB-triggered nuclear exclusion The above results suggest that the FmoB-triggered nuclear exclusion of NirA depends on active export process. In this case, the loss of NirA nuclear accumulation under IND conditions should be KapK dependent. To verify this, we treated the induced cells with NOC and simultaneously inactivated KapK1 with leptomycin B (LMB). Under these treatments the effect of NOC was lost and NirA-GFP remained nuclear ( Fig 4A). This finding confirmed that the FmoB-dependent nuclear exclusion of NirA-GFP triggered by NOC is an active process, which depends on the function of the exportin KapK. As expected, the loss of NirA nuclear accumulation negatively impacted on NirA transcriptional activity. In the wild type background NOC treatment led to a significantly lower level of the niaD transcript (ca. 60% reduction) whereas reduction of transcriptional activity is only marginal (ca. 10% reduction) in NOC-treated fmoBΔ cells (Fig 2B).
Although the above results suggest that the NOC-triggered effect is related to Met169 oxidation we wanted to directly determine if NOC treatment leads to Met ox 169 formation and the resulting NirA cytoplasmic localization and loss of transcriptional activating properties. To this end we tried to affinity purify FLAG-NirA from NOC-treated cells but, we were not able to obtain sufficient amounts of FLAG-NirA for mass spectrometry. As an alternative to test the involvement of Met169 oxidation in the export process, we mutated M169 to isoleucine (Nir-A M169I ) which is functionally and sterically similar but cannot be oxidized. Fig 4C shows that NOC treatment (INDo) of this strain did not lead to NirA M169I -GFP nuclear exclusion. This result establishes that Met169 is the only amino acid targeted by the FmoB-dependent process that leads to nuclear exclusion of NirA under nitrate induced conditions in the presence of NOC.
FmoB activity assays with different substrates. NADPH levels are expressed as absorbance (OD 340nm ) and are plotted against time. All samples received 2 μM FmoB and 200 μM NADPH. After 120 seconds pre-incubation time, samples received additionally 2 mM of putative substrates and kinetics of OD 340nm reduction indicates enzyme activities on the tested substrates. Cysteine, DTT and β-mercaptoethanol are apparently preferred substrates for FmoB under these conditions. (G) FmoB NADPH-oxidase activity is enhanced by n-octylamine. NADPH levels are expressed as absorbance (OD 340nm ) and are plotted against time. All samples received 0.5 μM FmoB and 200 μM NADPH and in addition indicated concentrations of n-octylamine. A specific substrate was omitted from this assay. Stimulation of FmoB activity apparently is optimal at concentrations between 1 mM and 2 mM, higher concentrations apparently inhibit the enzyme activity in these conditions. doi:10.1371/journal.pgen.1005297.g003 Met169 substitution by isoleucine results in a partial loss-of-function phenotype Given the vital role Met169 plays in NirA regulation, we tested if isoleucine, a sterically and biochemically similar residue which cannot be oxidized, could mimic reduced methionine in active NirA. We thus replaced nirA + with nirA M169I and tested the subcellular distribution of NirA M169I -GFP under conventional growth conditions (in the absence of NOC). We found that NirA M169I -GFP subcellular distribution was indistinguishable from wild type NirA-GFP, i.e. cytosolic in NI and nuclear in IND conditions ( Fig 4C) If, as hypothesized earlier, Met ox 169 has a regulatory role, some phenotype (either a gain-of-function or a loss-of-.function phenotype) should become apparent. To test NirA M169I function we constructed a strain in which the nirA M169I gene was driven by the estrogen-responsive promoter which is a hybrid sequence composed of roughly 100 bp of the nirA endogenous promoter fused to three estrogen-responsive elements and a 93 bp random sequence (for details on promoter composition see [30]). This [(3xERE)-RS-nirA] construct was chosen to be able to study the function of the mutated NirA M169I variant at near physiological levels (using part of the nirA promoter) but with the possibility to increase the protein concentration by ERE-mediated estrogen induction. Fig 4D shows the Western analysis testing protein levels of the NirA-GFP wild type (ERE p -nirA-gfp) and NirA M169I -GFP variant (ERE p -nirA M169I -gfp). In the absence of estrogen the proteins were barely detectable but were highly abundant in the presence of estrogen (DES).
At levels comparable with those obtained with the nirA physiological promoter (in the absence of DES) the NirA M169I substitution results in a clear growth phenotype on nitrate as sole nitrogen source, a phenotype which is not seen at higher expression levels of NirA (in the presence of DES), implying that that the M169I substitution is a hypomorphic mutation impairing but not completely abolishing NirA function. This effect is still evident in a strain expressing NirA M169I from the slightly stronger alcA promoter (S4B Fig, de-repressed with 0.2% fructose), as we found a roughly 50% reduction in NirA M169I activity under nitrateinduced conditions (alcA p -FnirA M169I ) compared to the wild type control (alcA p -FnirA).

Met169 substitution by alanine results in a NirA constitutive phenotype
We next exchanged M169 to alanine, a small and hydrophobic residue that does not occur in the NES of any known or predicted NirA orthologues. NirA M169A -GFP showed permanent, nitrate-independent nuclear accumulation and as for the NirA M169I -GFP mutation, NOC treatment was ineffective (Fig 2A, last panel). To test the activity of the protein a FLAG-tagged version of NirA M169A was expressed at moderate levels from the de-repressed alcA promoter (see nirA expression levels of two independent alcA p -FnirA M169A transformants and corresponding protein levels of one strain in S5 Fig). Notably, the M169A variant resulted in a gain-of-function phenotype as this NES variant was constitutively active and promoted inducer-independent transcription of the NirA target gene niaD (S3 and S5 Figs, lanes alcA p -FnirA M169A ). Thus, a nirA M169A substitution showed the same phenotype as the constitutively active nirA c 1 allele (carrying a nirA G167V substitution) and together these data support the view that the NES not only dictates NirA subcellular distribution but also its transcriptional competence.

NES accessibility and the interaction with KapK depends on the Met169 oxidation state
We next investigated to which extent oxidation or the substitution of Met169 by other amino acids influences the interaction between the NES and KapK. Structural work established that the interaction of Crm1 (KapK homolog in S. cerevisiae) with nuclear export sequences requires hydrophobic residues [12,13]. Methionine with its hydrophobic side chain conforms to this rule but oxidation to its sulfoxide (Met ox ) makes it more hydrophilic and polar [31]. It seems contradictory that, under conditions, in which NirA is effectively exported from the nucleus and NES-KapK interaction should be strong, oxidation of Met169 occurs, thus weakening precisely this interaction. We thus tested different NirA-NES synthetic peptides for in vitro interaction with S-tagged KapK present in protein extracts. These were prepared under native, non-denaturing conditions from KapK_S-tag cells grown on arginine medium (NI conditions). The 18 amino acids synthetic peptides spanning the NirA-NES (underlined; D 159 QFESELAGKMSNLVLDG 176 ) comprised either the conserved M169 (bold) in non-oxidized (Met169) or oxidized (Met ox 169) form or a M169A substitution. One additional peptide which carried two mutations (L172A/L174A) known to interrupt the NES-Kap interaction was also tested [11]. The biotinylated peptides were immobilized separately on streptavidin resin and incubated with cell-free extracts containing S-tag-KapK. After washing, the columns were eluted and the amount of S-tag KapK was determined by Western blotting with an anti-S-tag antibody ( Fig 5A). In agreement with data above, we saw no interaction in the assay when the M169A or L172A/L174A peptides were used but a clear interaction with the M169 peptide (lane "M" in Fig 5A). When the NES peptide containing Met ox 169 was tested the interaction with KapK was considerably weaker as documented by the much lower amount of KapK_S-tag captured compared to reduced Met169 (compare lanes "M" and "M ox " in Fig 5A). These in vitro binding assays confirm that direct interaction between KapK and the NES only occurs when the NES consensus is intact (i.e. hydrophobic residue of a specific length at position 169) and indicate that in fact reduced methionine in the NES results in a stronger interaction with KapK than oxidized methionine. This results pose the question of why, in conditions

Met ox 169 indirectly determines NES-KapK interaction by affecting NES accessibility
The above paradoxical finding could be accounted for if Met ox 169 -although weakening the KapK interaction,-had a regulatory function affecting protein structure and folding and thereby influencing either directly via domain exposure, or indirectly via protein-protein interactions, the accessibility of the NES to exportin. There are 12 cysteines in NirA, besides those involved in chelating the Zn ++ -ion in the bi-nuclear cluster. Querying the DiANNA disulphide prediction server (http://clavius.bc.edu/~clotelab/DiANNA/) with the NirA sequence omitting the cysteines in the DBD (query sequence residues 71 to 892) predicts four disulfide bonds whereas the same query submitted to DINOSOLVE (http://hpcr.cs.odu.edu/dinosolve/) predicts five disulfide bonds but also in this prediction, the cysteines potentially involved in disulfide formation partially overlap with the residues predicted by the DiANNA program. We tested this possibility by comparing the interaction of KapK with NirA in co-immunoprecipitation (coIP) experiments using extracts prepared from cells grown under NI (NirA export) or IND (no NirA export) conditions. To test if the NirA-KapK interaction is sensitive to disulphide bridge-mediated protein folding we performed the coIP experiments with two biochemically different setups, i.e. conditions, in which the natural folding of the protein is disrupted by addition of dithiothreitol (DTT) known to reduce disulfide bridges within or between proteins and under native conditions omitting DTT thus not changing any potential disulfide bonds within NirA.
To this aim we performed the coIP analysis in extracts of a strain producing both FLAG- In extracts prepared from the strain carrying the M169A mutation (alcA p -FnirA M169A ) FLAG-NirA never co-precipitated with KapK. These pull-down assays confirmed the in vitro interaction data (Fig 5A) and demonstrated that in vivo the constitutive nuclear positioning of NirA M169A -GFP is due to the fact that the NES in NirA M169A is not recognized by KapK. When co-IP was carried out in extracts prepared under disulfide-reducing conditions (addition of 5mM DTT) which should unfold the native protein structure, FLAG-NirA permanently interacted with KapK-S-tag, regardless of whether the coIP was carried out with extracts from NI or IND cells (Fig 5B, left panels labelled +DTT). Thus, under IND conditions the NES must be masked by intramolecular interactions with are sensitive to DTT. Under NI conditions the NES seems to be permanently accessible to KapK. Interestingly, when the native protein structure was unfolded, the interaction between NirA and KapK is equally strong regardless of whether the NirA NES carries Metox169 (NI conditions) or Met169 (IND conditions). Thus the NES activity in the NirA protein depends on its global folding, rather than on the oxidation state of Met169. Under native protein folding conditions the NES of NirA (containing Metox169) is exposed only when the extracts have been prepared from NI cells but the NES containing reduced Met169 is not exposed or at least not accessible for KapK interaction when the extracts have been prepared from nitrate-grown (IND) cells.

A C-terminal portion of NirA influences KapK-NES interaction and NirA subcellular localization
In the co-IP experiments we detected besides the expected Flag-NirA 105 kDa band a *70 kDa band. It is reasonable to suppose that this band results from a C-terminal truncation of N-terminal-FLAG-tagged-NirA. Whether this truncation occurred intracellularly or artefactually during extraction, this result shows that a C-terminally truncated NirA can bind to KapK irrespectively whether the extracts have been prepared from NI or IND cultures or whether the extracts have been treated with DTT or not. This faster migrating~70 kDa band is most likely not an unspecific signal because it is not appearing in the extracts from the NirA M169A strain. In other words, the deletion of the C-terminus has the same effect as unfolding the full-length protein by DTT. In both cases the NirA NES is able to interact with KapK in extracts prepared from nitrateinduced cells.
To further investigate the role of the NirA C-terminus in the NES accessibility to KapK we constructed a gpdA p -NirA-GFP variant lacking 278 amino acids of the C-terminus (encompassing the previously defined 193 amino acids of the NirA activation domain, AD) and expressed this construct in A. nidulans (strain termed NirA ADΔ -GFP expressing amino acids 1-614 of NirA). If the C-terminal AD be functionally involved in masking the NES under IND conditions, a C-terminally truncated form should always expose the NES and never accumulate in the nucleus. Against our expectations, NirA ADΔ -GFP showed nuclear GFP staining constitutively, even when no nitrate was present in the medium (Fig 6A). Thus, the 278 amino acid domain of NirA is involved in two apparent functions. It is necessary for expression of genes regulated by NirA, as it includes the activation domain, but also includes a sequence or structure determinant that is necessary for NES exposure under NI conditions. While the proteolysed form always interacts with KapK (Fig 5), the in vivo behaviour of NirA ADΔ -GFP implies that it does not. This could be explained if the truncation resulting from proteolysis removed residues beyond the 278 C-terminus and this additional stretch was affected directly or indirectly the NES/KapK interaction.

A central region of NirA is necessary to mask the NES
We thus wanted to characterize this region and deleted in the nirA gene the sequence coding for the central portion between the NES and the activation domain of NirA. This "minimal" version of NirA displayed a deletion from aa 230 to 737, thus fusing the N-terminal quarter of the protein (putative nuclear localization sequences, the DNA binding domain, and a stretch containing the NES) to the 193 amino acids of the AD thus removing all putative internal regulatory sequences (to be called nitrate regulatory domain, NiRD, and the strain lacking this region consequently NirA NiRDΔ ). This construct, driven by the strong gpdA promoter and Cterminally fused to GFP was tested for subcellular distribution. Remarkably, the strict nuclear accumulation after nitrate induction is lost in this NirA variant and it is evenly distributed between nuclei and cytoplasm under both non-induced and induced conditions (Fig 6B). This demonstrates that not the AD but a central portion of NirA is required to mask the NES and most likely, the proteolytically processed form always interacting with KapK in the coIP experiments lacks an essential part of this regulatory domain of NirA. Growth tests performed with this strain showed that this "minimal NirA" is still sufficiently active to support growth on nitrate and nitrite as nitrogen sources, although overexpression of the construct may obscure some subtle differences in growth (Fig 6C).
A forward-genetic screen pin-points a short stretch within the NiRD with putative regulatory functions To narrow down the roughly 500 amino acid large region with regulatory functions and/or to identify extragenic factors mediating the NES-inhibitory effect of the NiRD we performed a screen for revertants of a conditional, cryo-sensitive mutations in nirA. Cryo-sensitive mutations are supposed to pin-point regions of protein-protein interaction, as they result in an increased energy requirement of activation for the assembly of specific structures [32]. We have selected (see Materials and Methods) a cryo-sensitive mutant unable to grow on nitrate at 25°C but able to grow at 37°C. We mapped the mutation to the nirA locus and sequencing of this allele identified a missense mutation resulting in a R347S substitution (see S2 Table for  nucleotide changes of the individual mutations). R347 is strictly conserved among NirA homologues (see B-link results). We have selected revertants of this mutation, which all turned out to be intragenic suppressors. We identified ten different mutations which led to a suppression of the conditional cs phenotype (Table 1) and 7 out of these mapped in a very narrow region immediately upstream of the R347S cs mutation between amino acid 302 to 315 and, additionally, at positions 342 and 344. Bracketing the 302-315 region, basic residues were replaced by acidic ones (K302E, K315E) but within these boundaries the amino acid substitutions maintained or increased hydrophobicity (I305L, L306F, N308I, E310A, V312F). The V312F not only suppressed the original R347S mutation, but it resulted in constitutive gene expression of niaD and niiA.

Discussion
To the best of our knowledge, this is the first report which demonstrates a regulatory role of Met oxidation on the function of a transcription factor. We have demonstrated a conditionspecific oxidation of a conserved methionine in the NES of NirA which correlates with the relocation of the factor from the nucleus to the cytosol. In vivo data shows that M169 oxidation necessitates an active flavoprotein (FmoB) even though the cognate enzyme expressed in E.coli (rFmoB), which can catalyse cysteine, DTT and β-mercaptoethanol oxidation (Fig 3), failed to oxidise the Met of a synthetic NES peptide. Thus, FmoB could mediate Met169 oxidation in the NES directly or indirectly. The former would require NirA sequences outside the NES to be necessary for this FmoB specific activity. Such direct methionine oxidation has been shown only for MICALs (for Molecule Interacting with CasL), an animal protein involved in actin oxidation and de-polimerisation [33,34].
There are multiple possible indirect pathways via which FmoB could act to mediate NES oxidation. FMOs oxidize a wide range of compounds in microbes, plants and animals [33,34]. The S. cerevisiae FMO1p enzyme makes a major contribution to the pools of oxidized thiols and thus influences the cellular redox environment [35]. It could be proposed that FmoB in A. nidulans alters intracellular redox balance, which subsequently results in Met169 oxidation of the cytosolic and/or nuclear fraction of NirA. However, FmoB is not isofunctional to FMO1p. The substrate specificity of FMO1p is broader than FmoB and it is localised specifically in the endoplasmic reticulum [36]. Our experiments have demonstrated that, in the absence of substrate, FmoB is an NADPH oxidase releasing highly reactive hydrogen peroxide (H 2 O 2 ), similar to the activity of human FMO isoforms [37], which by itself would alter the oxidoreduction state of the cell.
Structural studies of the human CRM1-snurportin1 places the LR-NES in a hydrophobic groove thus providing a structural basis for the hydrophobic NES consensus with intervening electronegative residues [38]. We have demonstrated the requirement of hydrophobic residues such as leucines L171/173 as well as of glycine G167 for NirA-NES function [11]. M169 can be replaced by isoleucine, but not by a small, less hydrophobic amino acid (alanine). The NirA NES contains oxidized Met169 under conditions in which it interacts efficiently with the KapK exportin. This is counter-intuitive as Met oxidation renders the NES more hydrophilic. This apparent contradiction could be resolved if the relevant parameter for the interaction of the NES and KapK was the exposure and availability of the NES, whether Met169 be oxidised or not. Immunoprecipitation (IP) and deletion analysis of NirA internal regions support this hypothesis. NirA in induced samples containing reduced Met169 can only be precipitated with KapK when the protein extract is treated with the strong thiol-reducing agent DTT (Fig 6B, left panels) known to reduce disulphide bonds [36]. This reducing treatment is presumably unfolding NirA and this may consequently expose the NES and allow KapK interaction in induced samples. Secondly, a partially degraded, C-terminally truncated form of NirA (called NirA trunc , a peptide roughly spanning residues 1-550, see Fig 6B and graphical representation in Fig 7A) always co-precipitated with KapK irrespectively of unfolding by DTT treatment which implies that the C-terminal~350 residues contain a region which blocks NES accessibility.
We thus hypothesised that Met169 oxidation acts indirectly by influencing NirA conformation. A NirA construct lacking a shorter part of the C-terminus (NirA ADΔ -GFP expressing a NirA peptide spanning residues 1-614, see Figs 6A and 7A) showed the opposite effect and (in vivo) never interacted with KapK indicating a permanent block of the NES. This latter result delimits the region which affects the accessibility of the NES.
We propose that a region between NirA residues of around 550 to 614 is involved in sequestering the NES and that the region from residues 615-982, comprising the activation domain (AD), negates the sequestering function of region 550-614. A version of NirA composed of only the N-terminal 230 residues (NLS, DBD and the NES) fused directly to the very C- terminal 155 residues (covering the minimal AD, residues 737-892) has lost the ability to accumulate in the nucleus under any condition. The region of NirA which is missing in this construct (231-732) also overlaps with the region that was found to be required to block the NES in the in vitro co-IP experiments (~550-892).
Taken together, these data led to the model shown in Fig 7B. In the absence of inducer, FmoB (directly or indirectly, indicated by a dotted line) would oxidize Met169 in the NES, this step leading to an open NirA conformation exposing the NES and allowing KapK interaction and nuclear export. In the presence of inducer, Met ox 169 would be rapidly reduced and this step allows the region between the NES and the AD to block the NES. Whether the inducer acts on the blocking domain itself or on the AD itself remains to be determined. As many modification-of-function mutations as well as intragenic suppressor mutations map to the central region we consider it likely that this portion of NirA carries nitrate-sensitive regulatory amino acids. We therefore termed it NiRD for "Nitrate-Regulatory Domain".
We have so far not been able to detect direct protein-protein interactions by yeast twohybrid assays and co-IP using different portions of NirA corresponding to the previously defined baits I, II, III and IV [39]. This failure may be due to the absence of Met ox 169 in the protein expressed during the yeast double hybrid screen. It is known from calmodulin, (CaM) that Met ox formation leads to drastic conformational changes and uncoupling of interacting domains [40,41].
Additionally Met ox at position 169 might influence the differential formation of disulphide bridges. Although no methionine oxidation is known to be involved in the regulation of Yap1, the S. cerevisiae regulator of the oxidative stress response, nuclear export of the protein is regulated by a specific disulphide switch which buries or exposes the NES of this activator [42][43][44].
It is remarkable that, so far, all analysed mutations in the NirA LR-NES sequence (L 165 AGKM 169 SNLVL 174 ) also influence the function of the protein as a transcription factor. In our previous study we have seen that replacements of leucines 172 and 174 by alanines (L172A/L174A) completely abolish the transcriptional activation function and in this study we determined that isoleucine at position 169 (M169I) partially inactivates the protein (Figs 4D and  S4B). In contrast, valine at position 167 (G167V, the NirA c 1 mutant) or alanine at position 169 (M169A, a mutation created during this work) transform the normally nitrate-dependent transcription factor into a permanently active transcriptional regulator. The suppressor mutations of R347S, all mapped in a relatively short stretch of 50 amino acids between residues 300-350 and included changes to more hydrophobic or electronegative residues (see Table 2). These changes suggested that this small region may form part of the nitrate-regulatory domain (NiRD) which, according to our present model (Fig 7) influences the function of the activation domain.

A. nidulans strains, growth conditions and genetic techniques
A. nidulans strains were grown in Aspergillus glucose minimal medium (GMM) and manipulated as previously described [11,17,21]. S1 Table lists all fungal strains used in this study. Some strains are not discussed in the text but were used for sexual crosses to obtain the strains of interest. S2 Table indicates the nucleotide changes in the mutant strains. S3 Table shows all oligonucleotide primers used in this work. Construction of plasmids for gene expression in A. nidulans and gene knock out methods are described in S1 Text. The strain displaying cold-sensitive (cs) growth on nitrate was selected as being unable to grow at 25°C (restrictive temperature) but grows like wild type at 37°C (permissive temperature). Revertants of the cs strain were obtained after UV mutagenesis as being able to grow on nitrite and nitrate as sole Nsource at 25°C. Strains were backcrossed to a wild type strain (pabaA1) to test for intragenic or extragenic reversion events. All revertants showed intragenic suppressor mutations (100% of progeny showed growth on nitrate at the restrictive temperature) and mutations were identified by sequencing of the coding region of nirA.

E. coli expression, purification and activity determination of recombinant FmoB
cDNA for cloning was prepared as described in S1 Text. FmoB activity was analyzed either by monitoring the NADPH consumption at 340 nm when incubated with different substrates (methionine, cysteine, glutathione, methimazole, N-octylamine), or following product formation, methionine sulfoxide, by HPLC analysis.

Transcriptional analysis in A. nidulans
Analysis of nitrate reductase (niaD) gene expression in different mutant backgrounds was carried out as described previously [21]. Growth conditions were as follows: strains were grown in AMM containing 0.2% fructose (to de-repress the alcA promoter in case of FLAG-nirA strains) and 3 mM arginine as a sole nitrogen source for 16 h at 37°C. Mycelia were harvested, washed with AMM and aliquots of the culture were incubated for 30 minutes in nitrogen free (-N) AMM. Subsequently, individual aliquots received either 3 mM arginine (NI) or 10 mM sodium nitrate (IND) and, in case of n-octylamine treatment, additional 2 mM n-octylamine (INDo) (Sigma Aldrich). These cultures were then further incubated on an orbital shaker at 37°C for 20 minutes.

Fluorescence microscopy
Strains were grown for 16 h at 25°C on cover slips in glucose minimal medium (GMM) and 3 mM arginine as a sole nitrogen source. Compounds were added to the media in the following concentrations: 10 mM n-octylamine, 10 ng/ml leptomycin B (LMB). Germlings were observed by an Olympus Fluoview FV1000 confocal microscope (DAPI: excitation at 405 nm wavelength, emission between 426-479 nm wavelengths; GFP: excitation at 395 nm wavelength, emission at 509 nm wavelength). Microscopy was performed as described [17]. Image manipulation was performed by Image J Fiji software.  Preparation of cell-free extracts from A. nidulans Strains expressing the FLAG-tagged nirA gene from the alcA promoter were grown on AMM containing 0.2% fructose as carbon source when NirA should be expressed at wild-type levels (TCA extracts). For overexpression of NirA cultures were treated with 50 mM butanone (EMK) for 3 h. For nitrate-induced conditions, 10 mM NaNO 3 was added and the cells incubated for further 5 minutes. Cells were harvested and cell-free extracts prepared as previously described [45]. For the preparation of whole cell extracts under denaturing conditions (TCA extracts) cells were washed in 20% trichloroacetic acid (TCA), harvested and frozen in liquid nitrogen. Pulverized mycelia was resuspended in 12.5% TCA followed by 3 washing steps with pure acetone. Samples were pelleted, air dried and resuspended in 1x Laemmli Buffer. 15 μg of extracts were loaded per lane. Intensity of signals were calculated with ImageQuant software (Molecular Dynamics) against actin (ActA) (Ab 691001, MP Biomedicals) as loading control. Western blot analysis GFP-and FLAG-tagged of nirA strains, were carried out by preparing total cell extracts according to the methods described by [46]. Grained mycelia were resuspended in protein extraction buffer (20 mM Tris-HCl, pH8, 150 mM NaCl, 0,01% Triton X-100) containing protease inhibitor cocktail (P8340, Sigma). 30 μg per lane of total protein extract was loaded on a 10% acrylamide/bis-acrylamide gel in the presence of β-mercaptoethanol and SDS as denaturing agents. GFP-NirA membranes were incubated with anti-GFP primary Ab (Roche, 11814460001) and with Anti-Mouse IgG (H+L) secondary Ab (Jackson, 115-001-003). FLAG-NirA membranes were incubated with anti-FLAG-tag primary Ab (F1804; Sigma Aldrich) and detected with goat anti-mouse HRP-conjugated secondary Ab (Jackson, 115-035-008).
Preparation of DNA affinity columns and purification of NirA protein 1 ml Streptavidin Sepharose High Performance (GE Healthcare) slurry was prepared following the manufacturer´s protocol. 30 μg of a biotinylated synthetic 104 bp DNA fragment containing NirA binding site 2 (NirA BS2) [47] in blocking buffer were added and incubated on a rotator over night at 4°C. For NirA protein purification cell extracts of relevant strains were prepared as described above setting the DTT concentration to 0.5 mM in the protein extraction buffers and adding a supplemental centrifugation step at 100,000g for 1 h at 4°C prior to loading the extracts onto the affinity column. Non-specific bound proteins were removed by washing the column with 5 ml washing buffer (25 mM HEPES pH 7.5, 100 mM NaCl, 0.1 mM EDTA, 50 μM ZnCl 2 , 10% glycerol, 0.5 mM DTT). Subsequently NirA BS2 interacting proteins were eluted with 3 ml elution buffer (25 mM HEPES pH 7.5, 50 mM KCl, 5 mM MgCl 2 , 0.1 mM EDTA, 0.5 mM DTT, 10% glycerol, 1 M NaCl, 0.5 mM DTT).

Mass spectrometric analysis
In-gel protein reduction, alkylation and trypsination were performed as described previously [48]. NanoHPLC-MSMS protein sequencing was performed on In vitro peptide interaction studies Different forms of synthetic biotinylated peptides containing 18 amino acids (DQFESE-LAGKMSNLVLDG) of the NirA nuclear export sequence (M169 underlined) were synthesized (Biomatik, USA). 1 μmol of the biotinylated peptides were bound to Streptavidin Sepharose Resin (SSR, GA Healthcare) which was equilibrated before by washing it 3 times with binding buffer (20 mM NaH 2 PO 4 , pH 7.4, 100 mM NaCl). After 2 h of incubation with the biotinylated peptides, the supernatant was removed after centrifugation and the SSR washed three times with binding buffer and then equilibrated with protein extraction buffer. The SRR-bound peptides were then incubated with 1 ml of the cell-free extract of the strain Stag-KapK for 2 h on ice. After this incubation period the SRR-peptide-protein complexes were washed twice each with buffer A (150 mM NaCl, 10 mM Tris-HCl, pH 8), buffer B (250 mM NaCl, 10 mM Tris-HCl, pH 8) and buffer C (500 mM NaCl, 10 mM Tris-HCl, pH 8). Samples were denatured in 1x Laemmli buffer, separated on 1.5 mm 10% SDS-PAGE gels and transferred to nitrocellulose membranes for Western blotting. To probe for the presence of S-tag-KapK the membranes were incubated with anti-S-tag primary antibody (Abcam ab19321) and detected with HRPconjugated rabbit anti-goat secondary Ab (SIGMA A5420).

Co-immunoprecipitation (Co-IP)
5 μl of the "capture" anti-S-tag antibody (Ab) was first bound to 40 μl protein G agarose beads (PGAB, New England Biolabs) in TSA buffer (10 mM Tris-HCl, pH 8.0; 150 mM NaCl) in a final volume of 800 μl. The mixture was incubated on a spinning wheel at 4°C for 2 h. During this time, the cell-free extracts were pre-cleared to avoid unspecific binding to PGAB. Cell-free extracts of relevant strains were prepared as described above with the modification that the extraction buffer either contained or did not contain 5 mM DTT, depending on thiol-reducing (+DTT) or thiol-non reducing (-DTT) conditions, respectively. For pre-clearing, 1 ml of the resulting protein extracts (5 mg/ml) from strains expressing FLAG-tagged NirA and S-tagged KapK were incubated in the same volume of dilution buffer (100 mM Tris-HCl, pH 8; 5 mM NaCl; 0.1% Triton X-100) containing 50 μl of Protein G Agarose Beads (PGAB, New England Biolabs) for 2 h on ice. The pre-cleared lysates and the linked S-tag-antibody-PGAB complexes were combined after centrifugation and incubated 2 h on ice. After incubation the samples were treated as described above for the immunoprecipitation assays. To detect FLAG-NirA the membranes were incubated with anti-FLAG-tag primary Ab (F1804; Sigma Aldrich) and detected with goat anti-mouse HRP-conjugated secondary Ab (Jackson, 115-035-008). Stagged KapK was detected as described above. Heavy chain IgG from S-tag primary Ab, used in the Western analysis as loading control of the co-IP samples, was visualized on the same blot during incubation with HRP-conjugated rabbit anti-goat secondary Ab (SIGMA A5420). Comparison of niaD mRNA levels between the control strain (alcA p -FnirA) and the msr double deletion mutant (alcA p -FnirA msrABΔΔ). Numbers below the niaD hybridization panel are relative values compared to the niaD IND expression in the control strain (alcA p -FnirA) that was arbitrarily set to 100. (C). NirA is fully functional in thioredoxin mutants. Transcription of the NirA target gene niiA was tested in strains carrying a deletion in the single A. nidulans thioredoxin gene (trxAΔ, accession number ANID_00170) under noninduced and induced conditions and compared to wild type (WT), nirA c 1 and a strain in which the trxA deletion has been complemented by a functional ectopic copy (27). Growth conditions were as in B, except that induction by nitrate was proceeding for 2 hours instead of 20 minutes. nirA mRNA levels were compared between strains expressing the gene from its native promoter or from the alcA promoter under carbon de-repressing conditions (0,2% fructose as sole carbon source) and in the presence of 3 mM arginine as sole nitrogen source (except lane IND NO 3 which contained 10 mM nitrate in addition to arginine). Northern blots show that nirA mRNA can barely be detected in Northern hybridizations when expressed from its own, very weak promoter (WT, nirA c 1) but is clearly seen in strains expressing nirA from the derepressed alcA promoter. The mRNA level of the wild type nirA gene was tested in the wild type background (lane alcA p -FnirA) and in the fmoB or msrAmsrB deletion backgrounds (lanes alcA p -FnirA fmoBΔ and alcA p -FnirA msrABΔΔ), respectively. NirA variants carrying replacements of the conserved methionine 169 by alanine (two independent strains were tested in lanes alcA p -FnirA M169A#1 and alcA p -FnirA M169A#2 ) or isoleucine (lane alcA p -FnirA M169I ) were also examined. In all cases in which nirA was expressed from the alcA promoter, transcripts were clearly detectable and signals similarly strong. Northern hybridization with a probe from the actin (actA) gene served as loading control. . Western blot analysis shows that the NirA-GFP protein is detected at an apparent molecular size of approximately 130 kDa by the anti-GFP monoclonal antibody (Roche). The fusion protein is highly abundant in all strains expressing the construct from the strong constitutive gpdA promoter independently from the presence of nitrate (NI or IND conditions) or the allele of kapK present in the strain (kapK + or kapK1). Also, the deletion of fmoA or fmoB did not greatly influence fusion protein abundance (strains gpdA p -nirA-gfp fmoAΔ or fmoBΔ), however, msrA and msrB deletions seemed to have a slightly negative effect on protein amounts (lane gpdA p -nirA-gfp msrABΔΔ). Clearly, expression of the fusion construct from the hybrid ERE-nirA p promoter (EREp under estrogenic conditions, see Materials and Methods) reduced the amount of NirA-GFP under both nitrate induced and non-induced conditions. In comparison to the respective wild type NirA-GFP control, neither NirA-GFP mutants carrying M169A (lane gpdA p -nirA M169I -gfp kapK1) nor M169I exchanges (lane ERE p -nirA M169I -gfp) showed lower protein levels. The nirA637 strain which was used as transformation recipient is used here as negative control for the Western analysis In some protein preparations partial degradation of NirA has occurred which becomes evident as higher mobility band below the main NirA signal in the GFP-Western shown in panel A. (B). Western blot analysis after SDS-PAGE of the FLAG-NirA fusion constructs. The FLAG-NirA protein is detected at an apparent molecular size of approximately 105 kDa by the anti-FLAG monoclonal antibody (SIGMA). The fusion protein is expressed from the de-repressed alcA promoter (0,2% fructose) in the absence of nitrate (3 mM arginine) and its abundance is similar in all tested strains. Although the same total protein amounts are loaded in panels A and B for SDS PAGE, the lower signal intensity in panel B may not correlated to reduced FLAG-NirA levels because the anti-FLAG antibody may have a lower affinity compared to the GFP Ab. However, the relative strength of the gpdA promoter is known to be higher than the de-repressed alcA promoter and hence the stronger signal in the GFP Western (panel A) may reflect also higher protein levels.  S2 Table. Nucleotide changes in the mutant strains. Table shows changes in nirA nucleotide sequences in strains isolated as suppressors of the cold-sensitive nirAstrains. Numbers in the row "position" refer to amino acid positions within the altered proteins, mutations describe the amino acid changes caused by the nucleotide change given in the same lane and number of strains describes how many independent mutants were isolated carrying the described nucleotide change.

Supporting Information
(DOCX) S3 Table. Oligonucleotides used in this study. Table shows names and sequences of oligonucleotides used for cloning and biochemical experiments throughout this study.
(DOCX) S1 Text. This section contains supporting experimental procedures and references.
(DOCX) S1 Video. Time course of induction of NirA-GFP nuclear localisation. Strain gpdA p -nirA-GFP was grown directly on microscope coverslips for 6 h on a non-inducing nitrogen source (3 mM arginine) and subsequently exposed to 10 mM nitrate. Pictures were taken in the laser scanning microscope (scale bar = 10 μm) immediately before addition of nitrate (0 sec) and 20, 40, 60, 80 and 100 seconds after induction.
(AVI) different genetic backgrounds. We are grateful to Harald Berger for composing the video of rapid nuclear accumulation shown in S1 Video, to Thomas Dalik for the HPLC determination of methionine sulfoxide formation and to Yazmid Reyes-Dominguez, Sotiris Amillis and Lukas Mach for stimulating discussions and expert technical advice. We would like to thank the anonymous reviewers for their constructive criticism and numerous suggestions for how to improve this manuscript.