A Sensing Role of the Glutamine Synthetase in the Nitrogen Regulation Network in Fusarium fujikuroi

In the plant pathogenic ascomycete Fusarium fujikuroi the synthesis of several economically important secondary metabolites (SM) depends on the nitrogen status of the cells. Of these SMs, gibberellin and bikaverin synthesis is subject to nitrogen catabolite repression (NCR) and is therefore only executed under nitrogen starvation conditions. How the signal of available nitrogen quantity and quality is sensed and transmitted to transcription factors is largely unknown. Earlier work revealed an essential regulatory role of the glutamine synthetase (GS) in the nitrogen regulation network and secondary metabolism as its deletion resulted in total loss of SM gene expression. Here we present extensive gene regulation studies of the wild type, the Δgln1 mutant and complementation strains of the gln1 deletion mutant expressing heterologous GS-encoding genes of prokaryotic and eukaryotic origin or 14 different F. fujikuroi gln1 copies with site-directed mutations. All strains were grown under different nitrogen conditions and characterized regarding growth, expression of NCR-responsive genes and biosynthesis of SM. We provide evidence for distinct roles of the GS in sensing and transducing the signals to NCR-responsive genes. Three site directed mutations partially restored secondary metabolism and GS-dependent gene expression, but not glutamine formation, demonstrating for the first time that the catalytic and regulatory roles of GS can be separated. The distinct mutant phenotypes show that the GS (1) participates in NH4 +-sensing and transducing the signal towards NCR-responsive transcription factors and their subsequent target genes; (2) affects carbon catabolism and (3) activates the expression of a distinct set of non-NCR GS-dependent genes. These novel insights into the regulatory role of the GS provide fascinating perspectives for elucidating regulatory roles of GS proteins of different organism in general.


Introduction
The glutamine synthetase (GS) is the only enzyme in living organisms that synthesizes glutamine as the principal nitrogen source for protein and nucleic acid biosynthesis. There are only two routes of ammonium assimilation in living cells: (1) NADPdependent glutamate synthesis, which is catalyzed by anabolically active glutamate dehydrogenases, and (2) assimilation of ammonium by the GS. This ATP-dependent reaction involves the initial formation of γ-glutamyl phosphate from glutamate, which reacts with ammonia forming glutamine and inorganic phosphate [1,2].
There are three distinct GS enzyme families that can easily be distinguished by length: GSI with 360, GSII with 450 and GSIII with 730 amino acids on average, respectively [3]. All of them form multimeric proteins containing double-ringed quaternary structures composed of identical units: GSI-and GSIII-type enzymes contain 12 identical subunits, whereas GSII-enzymes contain 10 identical subunits consisting of 2 pentameric rings [3][4][5][6]. Recent molecular studies and genome projects have shown that the genes of GSI, GSII and GSIII families are broadly distributed among prokaryotes and eukaryotes suggesting that the GS family members arose prior to the divergence of prokaryotes and eukaryotes [7][8][9][10][11][12][13]. Thus, GSI enzymes, which were previously thought to be limited to bacteria, have recently been identified in mammals and plants [13]. In addition to the GSI family members, representatives of the GSII family, which were thought to be exclusive to eukaryotes, have been found in all screened Streptomyces, Agrobacterium, Rhizobium and Frankia spp. strains [14]. However, in filamentous fungi only GSII family proteins were identified so far, and in most cases only one GS-encoding gene is present in the fungal genomes. Interestingly, in the filamentous fungus Neurospora crassa two non-identical subunits of the GSII family, GSα and GSβ, have been identified and in vitro translation indicated that different mRNAs code for GSα and GSβ subunits [15][16][17]. The presence of two GSencoding genes was later confirmed by genome sequencing [18].
Since glutamine is a key metabolite in nitrogen metabolism both the intracellular glutamine pool as well as the activity of GS are tightly regulated. Based on experimental data in various fungi it became widely accepted that glutamine is a (if not "the") key effector of nitrogen catabolite repression (NCR), a regulatory circuit that ensures the preferential utilization of reduced nitrogen sources such as ammonium and glutamine over more complex and energy-demanding ones, e.g. nitrate, purines and proteins [19][20][21][22][23][24][25][26]. However, beside glutamine also ammonium, glutamate and nitrate might be sensed by specific sensors and generate signals for nitrogen metabolite repression [24,25,27,28].
Seminal work in Saccharomyces cerevisiae established a model where the target of rapamycin (Tor) complex kinase 1 (TorC1) senses intracellular glutamine provided by the GS, thereby transmitting the signal of glutamine availability to the GATA-type transcription factors Gln3 and Gat1. These GATA factors activate transcription of NCR-sensitive genes only under nitrogen-limiting conditions or in the presence of a nonpreferred nitrogen source [29]. Recently, this linear model of signal transduction has been challenged. The addition of rapamycin (inhibiting TorC1) had a different effect on Gln3 and Gat1 (the homologs of AreA and AreB in Fusarium fujikuroi) localization, their DNA-binding capabilities and NCR-sensitive gene expression, than a treatment with L-methionine-Ssulfoximine (MSX; inhibiting GS enzymatic activity). These data suggest the existence of two parallel nitrogen-responsive signalling pathways in yeast, one dependent on TorC1, and the other one dependent on the GS [30]. However, the question of how the GS (MSX)-responsive signal is wired to the level of NCR-sensitive genes in S. cerevisiae and filamentous fungi remains unresolved.
In prokaryotes, a direct role of GS in nitrogen regulation is well established. In B. subtilis, GSI activity is feedback-inhibited by glutamine [31], and this inhibited form of the GS modulates the DNA-binding capabilities of the two transcription factors TnrA and GlnR by direct protein-protein interaction [31,32]. Both transcription factors regulate gene expression in response to changes in nitrogen availability. In Escherichia coli, GSI is subject to cumulative feedback inhibition by glutamine and end products of glutamine metabolism, such as ADP, AMP and other nucleotides by competing with the substrate glutamate for the active site [33].
Although a direct interaction of GS enzymes with potential transcription factors was never demonstrated in any eukaryotic organism, it has been proposed that the GS might be the key regulator in the nitrogen regulation network also in higher organisms. In N. crassa, two allelic gln1 gene mutants, gln1a and gln1b, have been generated in the 1980's and were shown to have derepressed levels of nitrate and nitrite reductase activity in the presence of ammonium, glutamate, and/or glutamine, suggesting that the GS may potentially be involved in sensing the nitrogen status and signaling to downstream effectors [21,22]. However, the nature of these mutations in the two GS-encoding genes was never elucidated.
The rice pathogenic fungus F. fujikuroi is well known for its production of a family of plant hormones, the gibberellic acids (GAs) [34]. Beside GAs, the fungus produces a set of other economically important SMs, such as the red pigments bikaverin [35] and fusarubins [36][37][38], as well as mycotoxins such as fusarins, fusaric acid, beauvericin and fumonisins [38][39][40][41][42]. The biosynthesis of GAs and bikaverin are repressed by high nitrogen concentrations by different molecular mechanisms. GAs were the first SMs for which a strict dependency on the major nitrogen regulators, the GATA transcription factor AreA and AreB has been shown [43,44,49] (unpublished data). In contrast, bikaverin is subject to a noncanonical AreA-independent nitrogen regulation [35]. Beside AreA and AreB, other regulators, such as the AreA-binding protein Nmr [43][44][45] and the bZIP transcription factor MeaB [46] are also involved in regulation of SM production in F. fujikuroi.
Previously we deleted the GS-encoding gene, gln1 in order to increase GA production by lowering the intracellular level of glutamine. However, in contrast to our expectation, deletion of gln1 or addition of MSX did not result in upregulation, but in repression of GA and also of bikaverin biosynthetic genes, suggesting a regulatory function of the GS in F. fujikuroi [47]. Beside SM biosynthetic genes, several other target genes of the GS were identified by differential cDNA screening between the wild type and the gln1 mutant, e.g. those encoding the isocitrate lyase, the uricase, translation initiation and elongation factors, and two small proteins with unknown function, DDR48 and CipC [47]. Our first speculation was that the cross pathway control regulator Cpc1, which is known to sense imbalances in amino acid pools, is involved in loss of SMs in response to glutamine limitation. However, this possibility was ruled out: targeted deletion of cpc1 did not alter the down-regulation of GA and bikaverin biosynthetic genes when the GS was inactivated by MSX [48].
The aim of this work is to deepen our knowledge about the role the GS plays in the nitrogen regulation network in F. fujikuroi. Our working hypothesis was that the GS exerts a regulatory function independently of its glutamine-providing enzymatic activity. To test this hypothesis, we complemented the F. fujikuroi Δgln1 mutant 1) with different fungal (wild-type and mutant GS-encoding genes of N. crassa) and prokaryotic (GSI and GSII of Streptomyces coelicolor) GS-encoding genes, and 2) with gln1 gene copies of F. fujikuroi carrying point mutations in all 14 highly conserved sequence domains. The complemented strains generated by both approaches were monitored for their dependency on glutamine (glutamine auxotrophy/prototrophy) and their ability to express NCRsensitive secondary and primary metabolism genes. Collectively, our data support the existence of a GS-dependent nitrogen-responsive pathway for SM biosynthetic genes that is independent of the glutamine-providing enzymatic function.
F. fujikuroi strains were precultivated for 72 h in 300 ml Erlenmeyer flasks with 100 ml Darken medium [50] with 18 mM glutamine, on a rotary shaker at 28°C on a rotary shaker at 200 rpm for 3 days. 500 μl of this culture was used as inoculum for cultivations in ICI media (Imperial Chemical Industries Ltd., UK) [51] with 6 mM (10 % ICI medium) or 60 mM (100% ICI medium) glutamine. For DNA isolation and protoplasting, F. fujikuroi strains were incubated in 100 ml modified ICI medium (Imperial Chemical Industries Ltd., UK) [51] containing 10 g/l fructose and additional 18 mM glutamine at 28°C on a rotary shaker at 200 rpm for 3 days or 18 h, respectively. For RNA isolation, the mycelia were harvested after addition of indicated nitrogen sources and time points (see text). For analysis of SMs, the the strains were grown in ICI medium with 6 mM or 60 mM glutamine.

Bacterial strains and plasmids
E. coli strain TOP10 (Invitrogen, Groningen, The Netherlands) was used for plasmid propagation. Glutamine synthetase gene copies from F. fujikuroi, N. crassa and S. coelicolor were cloned by PCR using primers with ApaI and SalI restriction sites (Table S1) and cloned adjacent to the F. fujikuroi gln1 promoter in pUCH-N-gln1 prom [48,52] carrying the nourseothricin resistance cassette. Plasmids pLHAp and pLHIIp, carrying the S. coelicolor gln1 and glnII genes, respectively, were kindly provided by W. Wohlleben (University Tübingen, Germany).

Nucleic acid isolation and sequence analysis
Lyophilized mycelium was ground into a fine powder and dispersed (in the case of DNA for use in PCR) in extraction buffer as described by Cenis [53]. DNA for Southern hybridization experiments was prepared following the protocol of Doyle and Doyle [54]. Plasmid DNA was extracted using the Genomed plasmid extraction kit (Genomed, Germany). Total F. fujikuroi RNA was isolated using the RNAgents total RNA isolation kit (Promega, Mannheim, Germany). Samples of 20 µg of total RNA were transferred to Hybond-N + membranes after electrophoresis on a 1% (w/v) agarose gel containing 1% (v/v) formaldehyde, according to Sambrook et al. [55]. Northern blot hybridizations were accomplished by the method of Church and Gilbert [56]. For Southern analysis, genomic DNA was digested with appropriate restriction enzymes, fractionated in 1 % (w/v) agarose gels, and transferred to nylon membranes. DNA probes were randomly labelled using P 32 oligonucleotides and hybridizations were carried out overnight at 65°C. PCR reactions contained 25 ng DNA, 5 pmol of each primer, 200 nM desoxynucleotide triphosphates, and 1 unit BioTherm DNA polymerase (GeneCraft GmbH, Lüdinghausen, Germany). The reactions were started with 4 min at 94 °C, followed by 35 cycles of 1 min at 94 °C, 1 min at 56°C to 65°C, 1 min at 70°C, and a final 10 min at 70°C. DNA and protein sequence alignments and phylograms were done with DNA STAR (Madison, WI, USA). Sequence homology searches were performed using the NCBI database server. Protein homology was based on BlastX searches [57].

PCR
All primers used for PCR were obtained from Biolegio (Netherlands) (

Site directed mutagenesis and generation of knock-out mutants
Site-directed mutagenesis was carried out as described by the manufacturer using the QuikChange® II Site-Directed Mutagenesis Kit (Agilent Technologies). To generate a template vector for site directed mutagenesis, the full length gln1 cDNA fragment from F. fujikuroi was amplified from cDNA by using the RT-primers gln1-F-SalI and gln1-R-ApaI (Table  S1), and the PCR fragment was cloned into vector pCR ® 2.1-TOPO ® (Invitrogen) resulting in vector pgln1-cDNA. Primers for introducing specific point mutations into the wild-type gln1 cDNA sequence are listed in Table S1. The mutated gene copies were then cut out with ApaI and SalI and cloned into the destination vector pUCH-N-gln1 prom [44] and subsequently transformed into mutant ∆gln1 [47]. For gene replacements, the plasmids pΔniaD and pΔniaD were assembled using yeast recombinational cloning as essentially described for N. crassa deletion vectors [58] and recently established for F. fujikuroi vectors [59]. The 5' and 3' flanks of niaD and niiA were amplified using primer pairs "gene"-5'-F1/-R1 and "gene"-3'-F1/-R1, respectively. Plasmid DNA from S. cerevisiae was extracted using the GeneJET TM Plasmid Miniprep Kit (Fermentas GmbH, St. Leon-Rot, Germany) with slight modifications: cells were resuspended in 300 µL Resuspension solution plus 100 µl glass beads, lyzed by addition of 600 µl Lysis Solution and neutralized with 450 µl Neutralization Solution.

N. crassa transformations and crosses
Assembly of the gene replacement cassette and transformation of N. crassa were performed according the Gene Knock-out Protocol of the Neurospora Functional Genomics Project (http://www.dartmouth.edu/ neurosporagenome/). Crosses were performed as described in [61].

Analysis of secondary metabolites
For analysis of gibberellin formation, the wild type strain, the Δgln1 mutant and all transformants expressing heterologous or mutated gene copies were cultivated in ICI medium with 6 mM glutamine for 5 days. GAs were analyzed by HPLC according to Barendse and van de Werken [62] using a Merck HPLC system with a UV detector and a Lichrospher 100 RP18 column (5μm; 250x4; Merck).
Bikaverin was analyzed in the same cultures as GAs by HPLC [35].

Plate assays
The growth of the transformants complemented with one of the GS-encoding genes from different microorganisms or mutated copies of F. fujikuroi gln1 was compared with that of the F. fujikuroi wild-type and the recipient strain, the Δgln1 mutant, on complete medium (CM) [63] with additional 18 mM of glutamine as nitrogen source and minimal Czapek-Dox medium containing NaNO 3 as nitrogen source. Plates were incubated at 28° C for three days in the dark. Solidified ICI complemented with 1 mM glutamine and either 9 mM glutamine, 4.5 mM ammonium tartrate, 9 mM sodium nitrate or 9 mM glutamate was used for growth assays of the wild type, the Δgln1 and ΔniaD mutants.
To compare growth and development of the N. crassa wildtype with the single and double Δgln mutants, strains were grown on Vogel's minimal medium [64] without glutamine or supplemented with 27 mM glutamine. Slant tube cultures were incubated at 30°C, race tubes at room temperature.

Western blot analyses
Total protein extraction was performed as described in [65]. 50 µg of the protein extract were used per lane and separated by discontinuous SDS-polyarcylamide gel electrophoresis. The 5% loading gel was used at a pH value of 6.8 while the 10% separation gel had a pH value of 8.8. The resulting gel was electro-blotted (semi-dry) to a nitrocellulose transfer membrane. For detection of the F. fujikuroi GS, polyclonal anti-TbGS antibodies (1:5000 dilution) were used [23]. To detect the primary antibodies HRP (Horse radish peroxidase) conjugated anti-rabbit secondary antibodies were used in a 1:10000 dilution, followed by visualization of the occurring chemoluminescence.

Determination of intracellular free amino acid concentrations
Lyophilized mycelium of the appropriate cultivations was ground into a fine powder and extracted according to the protocol in [66]. The re-suspended amino acids were derivatized with o-phthalaldehyde and analyzed by RP-HPLC according to a published procedure [67]. The cell debris remaining after the four extractions was dried to constant weight at 80°C and the dry weight was determined.

Regulation of gln1 expression and GS protein levels by AreA, MeaB and Nmr
To intensify our understanding about regulation of GS activity, we first compared the gln1 transcript and GS protein levels in the wild type with those in three regulatory mutants harboring targeted deletions of the highly conserved AreA-, Nmr1, and MeaB-encoding genes [43][44][45][46]68,69], respectively.
We studied gln1 transcript and GS protein levels under both nitrogen-limiting (6 mM glutamine) and nitrogen-sufficient (60 mM glutamine) conditions. In the wild type as well as ∆nmr1 and ∆meaB mutants the gln1 transcript levels were high under both conditions, with slightly stronger expression under nitrogen starvation conditions. In contrast, expression of gln1 is slightly reduced in the ∆areA mutant independently of the glutamine concentration (Figure 1 A).
Western blot analysis correlated well with the results of transcript studies regarding the role of AreA in regulating GS activity. The protein level of the GS was significantly reduced in the ∆areA mutant compared to the other strains. It is noteworthy that the highest GS levels were found in the ∆nmr mutant, probably due to higher activity of AreA ( Figure 1B). In summary, the gln1 transcript and GS protein levels are positively regulated by AreA, while Nmr and MeaB exert weak antagonistic functions.

The GS affects expression of nitrogen regulatory and secondary metabolite genes
Since we observed a strong down-regulation of GA biosynthetic genes in the gln1 deletion mutant in earlier studies [47], we wanted to investigate if this down-regulation is due to altered expression of nitrogen regulatory genes, such as areA, Regulatory Role of the Glutamine Synthetase PLOS ONE | www.plosone.org areB, and nmr, in the ∆gln1 mutant. Alteration of areA expression levels could be an explanation for the significantly decreased expression of AreA-dependent GA biosynthetic genes (e.g. cps/ks encoding the ent-copalyl/ent-kaurene synthase), but not for down-regulation of AreA-independent bikaverin biosynthetic genes [35,44,49].
The wild type and the ∆gln1 strains were grown in submerged cultures with low levels of glutamine (6 mM). After 72 hours of incubation, when glutamine is exhausted, we added either no nitrogen (nitrogen starvation) or one of the three nitrogen sources: glutamine, ammonium nitrate (each 60 mM) or sodium nitrate (120 mM) and incubated the mycelia for two more hours in order to study the immediate response to nitrogen addition (short term effect). In the wild type, areA, areB and nmr genes are expressed under nitrogen starvation conditions. Additionally, areA is well expressed in the presence of nitrate, probably due to its role in activating of nitrate utilizing genes as shown in Aspergillus [70][71][72]. In contrast, glutamine and ammonium represses the transcription of all three genes. Surprisingly, areA, areB and nmr are significantly upregulated in the ∆gln1 mutant compared to the wild type under all conditions tested (Figure 2A). These results indicate that the ∆gln1 mutant is unable to sense the exogenously added nitrogen sources under these experimental conditions resulting in enhanced expression of these otherwise repressed genes. However, despite the enhanced expression of areA and areB in the ∆gln1 mutant, expression of the AreA-and AreB-dependent GA biosynthetic gene cps/ks was abolished, indicating that either AreA and/or AreB are nonfunctional, or that the GS itself is important for cps/ks expression. Since nmr, an AreA target gene in F. fujikuroi [44], was highly expressed in the ∆gln1 mutant non-functionality of AreA seems unlikely ( Figure 2A). To substantiate this hypothesis, we grew both the wild type and ∆gln1 mutant on medium containing the nitrate analogon The wild-type (WT), ΔareA, ΔmeaB and Δnmr strains were grown for 3 days in ICI submerse culture with either 6 mM or 60 mM glutamine as nitrogen source. Total RNA was used for northern analysis using the genomic gln1 fragment as probe. 18S rRNA was visualized as a loading control. B. Total protein extract was used for western analysis using the polyclonal anti-TbGS antibodies and HRP-conjugated anti-rabbit secondary antibodies. Ponceau staining as loading control. doi: 10.1371/journal.pone.0080740.g001 KClO 3 . The gln1 mutant showed a much more restricted growth on KClO 3 compared to the wild type, indicating enhanced activity of AreA and the AreA-dependent nitrate reductase, resulting in an elevated conversion of KClO 3 to the toxic KClO 2 by nitrate reductase activity ( Figure 2B).
To gain a deeper insight into the role of GS in nitrogen sensing and signaling, we compared the effects of different nitrogen sources on the expression of NCR sensitive genes such as the ammonium permease-encoding genes (mepB and mepC) [73], SM biosynthetic genes (cps/ks and bik2), and the regulatory genes areA, areB, and meaB under long term (three days) incubation conditions. The wild type and the gln1 mutant were grown for three days in synthetic medium with either 6 mM glutamine (nitrogen-limiting conditions) or in the same medium with additional glutamine, glutamate, ammonium tartrate, or sodium nitrate to an equal final nitrogen concentration of 120 mM (nitrogen surplus conditions). Similar to our previous observations for short term response of nitrogen addition (two hours), areA and areB transcript levels were up-regulated in the ∆gln1 mutant after 3 days incubation, but only under nitrogen-limiting conditions (6 mM glutamine) or if one of the two GS substrates, glutamate or NH 4 + , was present in addition to 6 mM glutamine ( Figure 2C). Also meaB, encoding a nitrogen-responsive bZIP trancription factor [46] is upregulated in the gln1 mutant compared the wild type. Previously, we have shown that meaB gives rise to two distinct mRNA transcripts, meaB S (expressed under nitrogen-limiting conditions in an AreA-dpendent manner) and meaB L (expressed under nitrogen sufficient conditions) [46]. In the wild-type, meaB S is detectable only under nitrogen-limiting conditions, whereas meaB S appears in the ∆gln1 mutant also under high ammonium or glutamate concentrations ( Figure  2C). These data suggest that ammonium and glutamate appear not to be sensed in the gln1 mutant resulting in derepression of areA and subsequent appearance of the meaB S transcript despite nitrogen sufficiency.
Several AreA-dependent genes such as mepB and mepC [73], resembled the areA transcription profile in the ∆gln1 mutant: they are repressed with glutamine and nitrate, but deregulated with ammonium and glutamate ( Figure 2C). In contrast, expression of genes involved in secondary metabolism such as the AreA-dependent cps/ks and AreAindependent bik2 genes was lost in the ∆gln1 mutant even under favoring nitrogen starvation conditions compared to the wild type ( Figure 2C).
Taken together, these data indicate that there is a clear regulatory difference between NCR-sensitive genes involved in primary metabolism (e.g. mepB and mepC) and regulation (e.g. areA, areB, meaB), and those involved in secondary metabolism (e.g. GA and bikaverin genes). Genes involved in primary metabolism and regulation are repressed by 1) glutamine and nitrate independently of the GS, and 2) by NH 4 + and glutamate in a GS-dependent manner, i.e. they are derepressed in the ∆gln1 mutant under conditions where these nitrogen sources are present in access. In contrast, SM biosynthetic genes are not expressed in the ∆gln1 mutant under all conditions tested underlining their strict dependency on the presence of a functional GS ( Figure 2C).

Do heterologous GSI-and GSII-encoding genes complement enzymatic and regulatory defects in the F. fujikuroi Δgln1 mutant?
Comparative genomic analyses between the recenty sequenced genome of F. fujikuroi and all other so far sequenced fungal genomes showed that several SMs produced by F. fujikuroi, e.g. GAs, are very specific for this species [42]. Therefore, we wanted to explore the possibility if heterologous GS-encoding genes can restore both the enzymatic activity (glutamine formation) and regulation of SM production. Therefore, we complemented the F. fujikuroi Δgln1 mutant with the two GSII-encoding genes from N. crassa and GSI-and GSII-encoding genes from S. coelicolor. Our hypothesis was that the heterologous GS-encoding genes might overcome glutamine auxotrophy, but probably not the defective Fusarium-specific secondary metabolism.
For S. coelicolor it was shown that both glnII (GSII-type) and gln1 (GSI-type) genes are fully functional independently of each other [14,74]. However, it is not yet known why N. crassa harbors two highly similar GS copies, encoded by the two distinct genes NCU04856.5 (Ncgln1) and NCU06724.5 (Ncgln2), and if both genes are essential to form an active GS multimer as suggested [75][76][77]. Growth tests on minimal medium showed that both genes encode fully functional GS proteins that can complement each other (detailed experimental informations in supplementary material, Figure  S1). However, it was not possible to generate homokaryotic ΔΔNcgln1/Ncgln2 double knock-out mutants, and even the heterokayotic mutant strains revealed growth defects on minimal medium ( Figure S1).
The S. coelicolor and N. crassa GS-encoding genes were individually fused with the F. fujikuroi gln1 promoter and transformed into the F. fujikuroi Δgln1 strain, respectively. Beside the wild type genes Ncgln1 and Ncgln2 we also examined the functionality of the Ncgln2 gene copies from the GS-defective N. crassa mutants gln-1a and gln-1b, which were previously shown to have a 20-to 30-fold lower GS-activity [75,76] and to be deregulated in the presence of ammonium [20,21]. Since the nature of the mutations in these two strains (gln-1a and gln-1b) has never been studied, we sequenced Ncgln1 and Ncgln2 genes in both mutants and identified several point mutations in both GS-encoding genes, but mainly in the Ncgln2 gene of both strains (Table S2).
Both N. crassa wild type genes, Ncgln1 and Ncgln2, were able to restore wild-type-like growth on minimal medium without glutamine, whereas none of the Ncgln2 mutant alleles (Nc2-1a and Nc2-1b) did ( Figure 3A), thereby substantiating our previous conclusions that both N. crassa wild type genes encode fully functional enzymes ( Figure S1). Surprisingly, transformants carrying either the GSI-or the GSII-type GSencoding S. coelicolor gene were also able to grow on minimal medium without glutamine, indicating a high level of functional conservation between GS from prokaryotic and eukaryotic organisms ( Figure 3A). Furthermore, Ncgln1, Ncgln2, Scgln1 and ScglnII genes were able to restore wild-type-like growth in submers culture with 6 mM glutamine (Table 1). By contrast, the defective Nc2-1a and Nc2-1b gene copies accumulated significantly less biomass similar to the Δgln1 mutant. Transcription of NCR-sensitive andinsensitive genes in response to different nitrogen sources in the wild type and the Δgln1 strain. A. The wild type (WT) and the Δgln1 strain were grown for 3 days in ICI submerse cultures with 6 mM glutamine as nitrogen source. After 3 days glutamine (gln), NH 4 NO 3 or NaNO 3 (NO 3 -) were added to a total concentration of 60 mM in case of gln and NH 4 NO 3 , and 120 mM in case of NaNO 3 . An equal volume of H 2 O was added as control. The mycelia were harvested 2 hours after nitrogen addition (short term exposure), and total RNA was used for northern analysis. 18S rRNA was visualized as a loading control. B. The wild-type (WT) and the Δgln1 mutant were grown for five days on CM agar with 100 mM glutamine and with or without 50 mM KClO 3 . If AreA is active the nitrate reductase-encoding gene niaD is expressed, and the nitrate reductase reduces KClO 3 to the toxic KClO 2 . In the wild type, high glutamine levels repress the expression of areA and niaD, while the growth of the Δgln1 mutant is restricted due to an active AreA and subsequent expression of niaD leading to accumulation of toxic KClO 2 . C. The wild-type (WT) and the Δgln1 strains were grown for 3 days (long term exposure) in ICI submerse cultures with 6 mM glutamine (control; only water was added) or 6 mM glutamine and additionally 54 mM glutamine (gln), 54 mM ammonium tartrate (NH 4 + ), 108 mM NaNO 3 (NO 3 -) or 108 mM glutamate (Glu) as nitrogen source. Total RNA (15 µg) was used for northern analysis and hybridized with probes as indicated. 18S rRNA was visulaized as a loading control. Abbreviations: mepB and mepC, ammonium transporter genes. To show if restoration of enzymatic GS activity and wild typelike growth by heterologous GS-encoding genes correlates with restoration of secondary metabolism, and if missing enzymatic activity of the Nc2-1a and Nc2-1b gene alleles correlates with the lack of secondary metabolism, GA and bikaverin production were analyzed in all strains after five days growth under inducing conditions (6 mM glutamine). The data obtained show a strong correlation between glutamine prototrophy and the ability to produce these SMs ( Table 1). The two N. crassa wildtype genes Ncgln1 and Ncgln2 and the two Streptomyces genes, Scgln1 and ScglnII, which could restore wild-type growth of the Δgln1 mutant on minimal medium ( Figure 3A), were also able to restore bikaverin and GA gene expression and concomitant production of the chemical compounds ( Figure 3B and C; Table 1).
All transformants carrying an active GS enzyme revealed wild-type-like expression levels for canonical AreA target genes such as the amino acid permease-encoding genes aap1 and aap8 ( Figure 3C). By contrast, aap1 and aap8 are not expressed in the Δgln1 mutant and in transformants containing non-functional Ncgln2 gene copies (Ncgln2-1a; Ncgln2-1b) from N. crassa mutant strains gln-1a and gln-1b ( Figure 3C). Furthermore, we studied the correlation between GS enzyme activity and expression of the previously identified GS target genes ddr48, cipC and cpc [47]. Similar to the amino acid permease-encoding genes, expression of ddr48 was only restored in transformants carrying one of the wild type N. crassa or Streptomyces GS-encoding genes. An unexpected expression pattern was observed for cipC, a nitrogenindependent GS target gene [47]. The cipC gene family is unique to the fungal kingdom and is reported to be expressed during the adaption to pathogenic growth in Aspergillus fumigatus and Ustilago maydis [78,79]. Unlike all other tested genes, expression of cipC could not be restored by the two functional GS-encoding genes from S. coelicolor or Ncgln1 from N. crassa. Only Ncgln2 and, surprisingly, its mutated, nonfunctional copies Ncgln2-1a and Ncgln2-1b were able to restore cipC expression. Similarly, the lower wild-type-like expression of the GS target gene cpc encoding the crosspathway control transcription factor FfCpc [47,48] was also restored by the non-functional mutant copy of Ncgln2-1b and partially by Ncgln2-1a ( Figure 3C).
In summary, GS-encoding wild type genes (GSI and GSIItype) of N. crassa and S. coelicolor restored wild-type-like Total RNA was isolated from mycelia grown for 5 days in ICI submerse cultures with 6 mM glutamine and used for northern blot analysis. 18S rRNA was visualized as a loading control. Abbreviations: aap1 and aap8, amino acid transporter genes; cipC, ddr48 and the cross-pathway control gene cpc were found to be GS-target genes [47]; mepB, ammonium transporter gene. growth on minimal solidified medium (prototrophy), wild typelike biomass formation in submers culture with 6 mM glutamine, and the wild-type-like expression pattern of NCR sensitive genes involved in primary (aap1, aap8) and secondary metabolism (GAs, bikaverin). A separation of enzymatic activity and regulatory function of the GS was obtained only regarding expression of the two GS-dependent genes cipC and cpc, which were also expressed in strains complemented with non-functional Ncgln2 gene copies. The partial restoration of cipC and cpc by non-functional GS proteins is a first indication for a potential regulatory role of the GS protein itself.

Site-directed mutagenesis of the F. fujikuroi gln1 gene
So far, restoration of glutamine biosynthetic activity by heterologous GSI and GSII-encoding genes correlated with restoration of SM production probably due to restoration of wild type-like growth in submers culture under nitrogen-limiting conditions. Next we wanted to investigate if mutations in certain conserved amino residues would result in deregulation of NCRsensitive genes (e.g. GA and bikaverin genes) without alteration of GS activity, or in the opposite phenotype, i.e. loss of GS activity but unchanged ability to produce SMs. To identify those residues, we performed an amino acid alignment between fungal, bacterial, plant and human GSI and GSII isozymes. The resulting phylogram shows a clear separation of the three classes of GS enzymes, and within the eukaryote clade, between fungi and animals on one hand, and photosynthetic eukaryotes on the other hand ( Figure S2). In contrast to bacteria like Streptomyces spp. and Rhizobium spp. that posses GS proteins of various types, ascomycetous fungi harbor only GSII type proteins ( Figure S2). Despite the separation into the three classes of GS, the amino acid alignment revealed a high degree of conservation between fungal, bacterial, plant and human GSI and GSII isozymes in 14 domains, probably responsible for substrate and ATP binding, enzymatic activity or the interaction among subunits ( Figure 4) [4,5,80,81].
To test our hypothesis, we generated mutants with specific amino acid substitutions in all 14 highly conserved domains ( Figure 4; Table 2). Transformants carrying mutated gene copies under the control of the native gln1 promoter were analyzed for their growth ability on solid minimal medium without glutamine and for biomass and SM formation in submers cultures under optimal GA-and bikaverin producing conditions (6 mM glutamine). In addition, the mutants were analyzed for the expression of NCR-sensitive and known NCRinsensitive GS target genes, such as cipC and cpc [47].
Of the 14 mutants generated, 5 (R311A, E297A, E200A, L291A, and L76A) were able to grow in a wild-type-like manner on agar plates without glutamine, indicating full enzymatic functionality, whereas 9 are glutamine auxotrophs showing no growth at all ( Figure 5A). The 5 enzymatically functional mutants were able to restore bik and GA gene expression ( Figure 5B) and concomitant bikaverin and GA formation ( Figure 5B; Table 2) underlining the above mentioned correlation between GS activity and secondary metabolism.
Surprisingly, three of the nine glutamine auxotrophic mutants (D60A/S62A, G246A/G248A, S72A/D73A) highly expressed bikaverin biosynthetic genes (e.g. bik3) and exhibited wild-typelike bikaverin production ( Figure 5B), while the signal for the GA biosynthetic gene cps/ks was almost non-detectable ( Figure 5C). However, low concentrations of GA 3 were measured by HPLC analysis in the three mutants in contrast to the Δgln1 mutant that does not produce any GAs (Table 2). Beside the restoration of secondary metabolism, these three mutants (D60A/S62A, G246A/G248A and S72A/D73A) also showed wild-type-like expression of the other NCR-sensitive genes (aap1, app8, areA, areB and mepB) and the GS target gene ddr48, despite the loss of glutamine-forming activity ( Figure 5C).
In contrast, the expression pattern for the cross pathway gene cpc revealed a clear correlation with the ability to synthesize glutamine. Strains carrying non-functional gln1 alleles (including D60A/S62A, G246A/G248A and S72A/D73A) displayed high, Δgln1-like cpc expression ( Figure 5C). As previously observed, the GS target gene cipC shows a specific expression pattern. Its expression seems to depend on a specific trait of either the gln1 gene or its deduced protein independently of enzymatic functionality. Among the mutants with point-mutated gln1 gene copies, only mutant R311A showed wild-type-like cipC expression ( Figure 5C). The partial restoration of some wild-type phenotypes in the three regulatory mutants despite their lost catalytic activity is an indication for a complicated regulatory network centering on GS as one of the major players.
To ensure that the mutated gln1 gene copies are transcribed and translated into GS proteins in the different transformants, northern and western blot analyses were performed (Figure 5  Figure 5E). To our surprise, no GS protein signal could be detected for the mutant D60A/S62A ( Figure 5D and E). However, since this mutant displayed a wild-type-like phenotype regarding gene expression and secondary metabolism, the GS protein level in this mutant is probably below the detection sensitivity but sufficient for its regulatory function.
In summary, the gln1 site-directed mutagenesis approach clearly showed that it is possible to separate enzymatic activity and regulatory functions of the GS. Three of the fourteen GS mutants with specific point mutations have lost the ability to synthesize glutamine but were able to produce SMs and showed wild-type-like expression pattern for NCR-sensitive genes. These data strongly indicate that the F. fujikuroi GS is an important player in the nitrogen regulatory network. These regulatory functions are not dependent on its enzymatic activity and can now be pinpointed to specific amino acid residues.

The three catalytic mutants have intermediate phenotypes between the Δgln1 mutant and the wild type
To further characterize the three mutants with restored secondary metabolism (D60A/S62A, G246A/G248A, S72A/ D73A), we compared their intracellular glutamine and glutamate concentrations to those of the wild type and the Δgln1 mutant. As controls we also analyzed the glutamine and glutamate pools in one mutant with a totally inactive GS variant (Y159) and one mutant with wild-type-like GS activity (R311). All strains were grown for three days in media with 6 mM and 60 mM glutamine, respectively. The most striking difference was the dramatically elevated glutamate level in the Δgln1 mutant and the completely inactive mutant Y159 in both media, probably due to the block of the dynamic shuttling between glutamate and glutamine by the glutamine synthetase/ glutamate synthase (GS/GOGAT) cycle [82].Surprisingly, glutamate levels are significantly lower in the three catalytic mutants with restored secondary metabolism, displaying intermediate levels between the wild type and the Δgln1 mutant ( Figure 6A). Glutamine levels are generally low in the wild type and all analyzed mutant strains after three days growth under nitrogen-limiting (6 mM glutamine) conditions, and significantly elevated in all strains in high nitrogen conditions. Interestingly, the three catalytic mutants (D60A/S62A, G246A/G248A and S72A/D73A) revealed again intermediate glutamine levels between the Δgln1 and Y59 mutants on the one hand, and the wild type and R311 mutant with an active GS, on the other hand ( Figure 6A).
To explore what consequences different glutamate levels have on glutamate catabolism, we compared the expression of genes involved in the formation or degradation of glutamate between the wild type, the Δgln1 mutant and the three deregulated GS mutants (D60A/S62A, G246A/G248A and S72A/D73A) after three days growth in media with 6 mM glutamine ( Figure 6B). The NADPH-dependent glutamate dehydrogenase encoding gene gdhA (glutamate producing) was only detectable in the three deregulated GS mutants (D60A/S62A, G246A/G248A and S72A/D73A), but neither in the Δgln1 mutant nor in the wild type. In contrast, the glutamate synthase (GOGAT) encoding gene, gltA, was weakly expressed in all tested strains compared to the wild type, probably due to glutamine starvation. However, the NAD +dependent glutamate dehydrogenase (GdhB) encoding gene, gdhB, and the glutamate-5-kinase encoding gene, pro4, which is involved in proline metabolism, were up-regulated in the Δgln1 mutant and much lower expressed in the wild type and the three point mutants ( Figure 6B). The up-regulation of gdhB and pro4 in the Δgln1 mutant is not surprising. Both encoded enzymes use glutamate as substrate, and the high expression of these two genes might be a response to the accumulation of glutamate in the Δgln1 strain. These data indicate that the three regulatory mutants show an intermediate phenotype between the wild type and the Δgln1 mutant regarding expression of genes involved in glutamate metabolism.
One of the most obvious phenotypes of the Δgln1 mutant is the significantly reduced growth in media with high amounts (440 mM) of glucose and 6 mM glutamine compared to the wild  Figure 4) grown for 3 days in ICI medium with either 6 mM (GA-and bikaverin-inducing (red coloration) conditions) or 60 mM glutamine as nitrogen source. C. Total RNA was isolated from mycelia grown for 5 days in ICI medium with 6 mM glutamine and used for northern blot analysis. 18S rRNA was visualized as a loading control. Abbreviations: see legend of Figure 3.D. Total protein extract of the mycelia from B was used for western analysis using the polyclonal anti-TbGS antibodies and HRP-conjugated anti-rabbit secondary antibodies. Ponceau staining as loading control. E. Total RNA (Northern blot analysis) and protein (Western blot analysis) was isolated from mycelia of the wild type, the Δgln1 mutant and the transformants complemented with the three pointmutated copies of gln1 which restore secondary metabolism but not glutamine formation. The strains were grown for 3 days in ICI medium with 6 mM glutamine. doi: 10.1371/journal.pone.0080740.g005 Figure 6. Impact of the F. fujikuroi GS on amino acid and carbon metabolism. A. Amino acid measurement of the wild-type, the Δgln1 mutant and three complemented strains carrying point-mutated gln1 gene copies. The strains were grown for 3 days in ICI medium with 6 mM glutamine as nitrogen source. After 3 days the mycelia were harvested, lyophilized and used for amino acid analysis. B. Transcription of nitrogen and carbon metabolism genes. The wild type, the Δgln1 and the three deregulated mutants were grown for 3 days in ICI medium with 6 mM glutamine as N-source. The nitrogen-starved mycelia were harvested and the total RNA was used for northern analysis. 18S rRNA was visualized as a loading control. Abbreviations: gdhA, NADH-dependent glutamate dehydrogenase; gdhB, NAD + -dependent glutamate dehydrogenase; gltA, glutamine oxoglutarate aminotransferase (GOGAT); pro4; glutamate-5-kinase; gpd1, glyceraldehyde-3-phosphate dehydrogenase gene; pki1, pyruvate kinase. C. The wild type and D60/S62 mutant strains were grown for 3 days in ICI medium containing 6 mM glutamine. Mycelia were harvested 30 minutes after addition of H 2 O, 60 mM glutamine (gln), 60 mM ammonium tartrate (NH 4 + ) or potassium glutamate (glu) and hybridized with the indicated probes. 18S rRNA was visualized as a loading control.  (Table 2). These data indicate that the mutant is not able to utilize glucose as efficiently as the wild type, as indicated by the reduced dry weight formation (only 15% dry weight compared to the wild type). Comparable mutants such as the ΔΔgdhA/gltA double and the Δaar1 single deletion mutants with strict glutamate and lysine auxotrophy, respectively, show neither growth defects on media with the complementing amino acid substrates nor impairments in secondary metabolism [48,83]. Therefore, we suggested additional consequences of the Δgln1 deletion on primary metabolism, e.g. on the ability to efficiently utilize glucose and generate ATP. To prove this hypothesis, we compared the growth of the wild type, the Δgln1 and the mutants with pointmutated gln1 gene copies in the standard minimal medium with 6 mM glutamine and 440 mM glucose. All prototrophic mutants grow similarly as the wild type while the glutamine auxotrophic strains accumulate significantly less biomass similar to the Δgln1 mutant. Surprisingly, the three deregulated mutants (D60A/S62A, G246A/G248A and S72A/D73A) produced five to six-fold more dry weigth than the Δgln1 mutant and the other auxotrophic strains, thus demonstrating an intermediate growth between the wild type and the Δgln1 mutant (Table 2).
To show if the inability to efficiently utilize glucose might be the reason for loss of secondary metabolism in the glutamine auxotrophic mutants, we compared the growth of the wild type, the Δgln1 and the three specific mutants in nitrogen-limited media (6 mM glutamine) with or without glucose (glutamine serves as sole nitrogen and carbon source). While all strains grew poorly without glucose, the auxotrophic strains accumulated even more biomass than the wild type under these conditions (Table S3). Neither the wild type nor the Δgln1 or the three site-directed mutants were able to promote GA nor bikaverin production without glucose indicating that secondary metabolism depends on optimal energy balance and carbon source availibility.
In summary, the GS seems to play an important role also in carbon/energy metabolism. Most of the site-mutated glutamine auxotrophic mutants are affected in their ability to efficiently utilize glucose, to grow in a wild-type-like manner and to produce SMs. However, only those three auxotrophic mutants which show almost wild-type-like growth in the standard medium with glucose and 6 mM glutamine were also able to produce SMs suggesting that the GS is essential for an effective and balanced carbon/energy metabolism.

The GS most likely senses NH 4 + availibility
The mutant D60A/S62A contains a mutation in a conserved domain probably involved in NH 4 + binding. To examine if this mutation leads to altered expression pattern of NCR sensitive genes after addition of ammonium to nitrogen-starving mycelia, we grew this mutant and the wild type under glutamine limiting conditions for three days before adding either glutamine, NH 4 + , glutamate or no nitrogen source (starvation). Gene expression of NCR sensitive primary (mepB) and SM biosynthetic (cps/ks; bik2) genes was monitored 30 minutes after addition of the respective nitrogen sources. In the wild type, glutamine, NH 4 + and to a lesser extent glutamate repressed expression of the AreA-dependent (cps/ks, mepB) and -independent (bik2) genes ( Figure 6C). In the D60A/S62A mutant carrying a mutation in the putative NH 4 + binding site, addition of glutamine and glutamate but not NH 4 + caused a decrease in bik2 expression compared to the water control ( Figure 6C). These data indicate that NH 4 + does not act as repressing nitrogen source in this mutant and that the mutations in the putative NH 4 + binding site D60/S62 cause loss of NH 4 + sensing and subsequent loss of wiring the repressing signal towards bikaverin gene expression.
Summarizing, the data provided in this study strongly support our hypothesis that the GS plays an important role in the nitrogen regulation network. The GS is involved in nitrogen sensing, regulation of NCR-sensitive genes as well as primary, secondary and carbon/energy metabolism. Using a site directed mutagenesis approach we were able to demonstrate that the enzymatic function of the GS can partially be separated from the additional regulatory and sensory functions. The inability of the Δgln1 mutant to produce SMs is probably due to its failure to utilize glucose for sufficient ATP formation. Furthermore, the putative NH 4 + binding site D60/S62 in the GS seems to be responsible for NH 4 + sensing independently of the enzymatic functionality.

Discussion
In this work, we performed detailed expression analyses and SM profiling of the wild type, the Δgln1 mutant and transformants carrying heterologous or specifically mutated GS-encoding alleles in order to show that the GS is involved in nitrogen sensing, regulation of NCR-sensitive primary and secondary metabolism genes and carbon/energy metabolism. Based on this new information, we are summarizing all aspects of our study in a model for the GS-dependent regulation network 1) in the wild type under nitrogen-limiting and sufficient conditions ( Figure 7A, B), and 2) in the Δgln1 and D60A/S62A mutants ( Figure 7C, D).

The GS is involved in nitrogen sensing and regulation of areA expression
Previously we have shown that GA and bikaverin gene expression is repressed by nitrogen in an AreA-dependent and independent manner, respectively, and that AreA also regulates the expression of gln1 [44,47] (Figure 7A, B). Here we demonstrate for the first time that deletion of areA also resulted in significant down-regulation of GS protein levels. In addition, we provide evidence that the expression levels of the regulatory genes areA, areB and nmr depend on the GS: the genes are strongly upregulated in the gln1 deletion mutant compared to the wild type despite the addition of sufficient amounts of different nitrogen sources (Figure 2A; Figure 7B, C). These data suggest that the mutant is not able to sense these high nitrogen concentrations as fast as the wild type. The up-regulation of areA, areB and nmr is probably an indirect effect of the even more reduced intracellular glutamine pool in the deletion mutant which is sensed by a yet unknown intracellular glutamine sensor ( Figure 7C, shown for areA). A. Growth of the wild type under nitrogen-limiting conditions. GS transcript and protein levels are under positive control of AreA, similar to the gibberellic acid (GA) cluster genes. Bikaverin genes are also highly expressed in an AreA-independent manner probably due to the low glutamine pool in the cell which is sensed by a putative intracellular sensor. B. Growth of the wild type under nitrogen-sufficient conditions. Excess of glutamine is detected by the postulated GS-independent sensor protein, exerting inhibition of AreA (no GA gene expression) and bikaverin cluster gene expression. Ammonium and glutamate are sensed by the GS itself. This signal is transduced by one or multiple putative GS interactor proteins that enhance carbon metabolism and inhibit expression nitrogen metabolite repression-sensitive gene expression. The inducing or repressing effect of these interactors on SM gene expression depends on the nitrogen status of the cell. Beside the direct regulatory impact of the GS by binding interactor(s), its ATP consuming activity depletes the intracellular ATP pool, which in turn prohibits the ATP-mediated inhibition of the carbon metabolism. An efficient carbon metabolism is required for the expression of SM cluster genes. To maintain the ATP-consuming GS activity the produced glutamine is reverted to glutamate by the activity of the glutamine oxoglutarate aminotransferase (GOGAT). C. Situation in the Δgln1 mutant. No sensing of ammonium and glutamate, and no interaction with putative interactors are possible. The missing ATP drain probably results in negative feedback regulation of carbon metabolism leading to significantly reduced growth and downregulation of secondary metabolism despite the elevated areA expression. The low glutamine concentration can not overcome the the disturbed glucose consumption. D. In the D60/S62 mutant, ammonium and glutamate are probably not sensed. The presence of a GS protein (though mutated and unable to produce glutamine) allows the binding to yet unknown interactors thereby partially restoring secondary metabolism, carbon metabolism and growth in low glutamine/high glucose conditions. Abbreviations: Mepammonium permease; AAP-amino acid permease.
(not included in scheme) seem to be sensed by a GSindependent mechanism, resulting in down-regulation of the regulatory genes ( Figure 2B; Figure 7C). These data are in accordance with findings in N. crassa or F. fujikuroi, where inhibition of GS by MSX or mutations in the GS-encoding genes resulted in relieved gene repression by NH 4 + and glutamate but not by glutamine [73,84,85].

The GS is involved in regulation of primary and secondary metabolism
The most unexpected phenotypes of the Δgln1 mutant were the loss of SMs (GA and bikaverin), significantly reduced growth with low glutamine and altered expression of specific sets of genes upon varying nitrogen conditions compared to the wild type. Based on the high conservation of GS proteins, we expressed several GS-encoding genes of prokaryotic (Streptomyces) and eukaryotic (N. crassa) origin in the F. fujikuroi mutant. We expected a restoration of glutamine prototrophy as shown for the A. nidulans ΔglnA mutant complemented with the glnA gene of the cyanobacterium Anabaena sp. [86]. However, we did not expect restoration of fungus-specific secondary metabolism by any heterologous GS. To our surprise, both N. crassa genes, Ncgln1 and Ncgln2, restored growth and secondary metabolism indicating that each of the N. crassa GS proteins alone can assemble to fully functional GS multimers. These data fit to the prototrophic phenotypes of single Ncgln1 and Ncgln2 deletion mutants, respectively ( Figure S1), and are in accordance with prototrophic single deletion mutants in Streptomyces spp. [87]. Even more surprising was the full restoration of both glutamine prototrophy and the ability to produce GAs and bikaverin by the GSI-type gln1 and the GSII-type glnII of S. coelicolor, although the genes share only 22.2 % and 48.3 % of amino acid identity with the GS of F. fujikuroi, respectively. Since it is very unlikely that bacterial GS proteins can control the activity of specific transcription factors in the background of an eukaryotic cell, we suggest that the restoration of SM production is due to the restored formation of glutamine and subsequent wild-type-like growth of these mutants.
A specific indicator for full GS functionality is the expression of the recently identified GS-dependent gene cipC [47]. Its expression can not be restored by feeding the mutant with glutamine but depends on the presence of a highly similar GS protein such as NcGln2. CipC expression is not restored in mutants carrying prototrophic gln1 and glnII genes from S. coelicolor or Ncgln1 from N. crassa. These data indicate that the F. fujikuroi GS fulfills several regulatory functions probably by interacting with yet unknown and highly specific regulatory proteins. Strikingly, beside the Ncgln2 wild-type gene, the defective Ncgln2 gene products from both the gln-1a and gln-1b mutants also showed expression of cipC supporting our assumption that specific GS-interacting proteins could exist ( Figure 7).
As the heterologous expression of N. crassa and S. coelicolor GS-encoding wild type genes resulted in concomitant restoration of almost all wild-type phenotypes, we performed a site directed mutagenesis approach by specifically mutating all 14 highly conserved amino acid motifs in order to separate the different functions of the GS. Similar studies of GS have been reported for Bacillus subtilis [88], but not for any eukaryotic GS protein so far.
Three of the fourteen generated mutants (D60A/S62A, G246A/G248A and S72A/D73A) expressing specific pointmutated F. fujikuroi gln1 gene alleles are strictly glutamine auxotrophic but retained some but not all regulatory effects of wild-type GS. Thus, all three mutants are able to produce wildtype levels of bikaverin and low amounts of GAs. It is likely that the expression of bikaverin biosynthetic genes depends mainly on a functional carbon and energy (ATP) metabolism (as discussed below) and the absence of AreA-independent nitrogen repression, while the GA gene expression needs active AreA and AreB regulators in addition ( Figure 7A). This differing regulatory function of the GS on the expression of the two distinct SMs is most likely due to the different mechanisms of nitrogen repression for GAs and bikaverin and the involvement of different yet unknown regulatory protein(s). These regulatory proteins might interact with the GS under distinct circumstances, allowing an induction of the carbon metabolism and simultaneous repression of areA (Figure 7).
In contrast to the wild type, one of the generated GS mutants, D60A/S62A, having point mutations in a putative NH 4 + -binding site, revealed derepressed bik2 levels when exposed to ammonium. This result argues for the importance of these residues to sense and transduce this specific repressing signal via a putative interaction partner ( Figure 7B).
In addition to partial restoration of secondary metabolism, the three mutants show a wild-type-like expression pattern for a set of NCR sensitive genes, but Δgln1 mutant-like behavior for another set of genes, such as the cross pathway regulatory gene cpc and genes involved in glutamate metabolism.

The GS is tightly linked to energy metabolism
One of the most surprising phenotypes of the ∆gln1 mutant is the drastically impaired growth under standard GA-and bikaverin-inducing conditions (6 mM glutamine) compared to the wild type. While the wild type accumulates about 1.4 g/l dry weight, the mutant produces only about 0.2 g/l dry weight under identical nitrogen conditions (Table 2). Northern blot analyses and amino acid concentration measurements suggest that the GS has a strong impact on anabolic amino acid pathways in F. fujikuroi. The ∆gln1 mutant seems to be unable to maintain metabolic homeostasis due to the interrupted GS-GOGAT cycle, the connecting step between carbon and nitrogen metabolism. As a consequence, the isocitrate lyaseencoding gene (ICL), the key enzyme of glyoxylate cycle [47], and several other genes involved in glutamate metabolism, are miss-regulated ( Figure 6B). However, the altered amino acid (glutamate, glutamine) concentrations in the Δgln1 mutant compared to the wild type seem unlikely to cause the severe growth restrictions, since the deletion mutant displayed enhanced growth when glutamine was supplied as sole carbon and nitrogen source ( Table 2). From these data we conclude that the Δgln1 mutant is defective in glucose consumption and subsequently in providing sufficient ATP to allow wild-type-like growth. Our assumption is in accordance with earlier findings in N. crassa where inhibition of GS-activity resulted in strong inhibition of carbon catabolism and growth [17,89]. Most likely, the reason for this dependency of carbon catabolism on the GS activity is the high ATP consumption by the functional GS, creating an ATP drain, which in turn activates the catabolism of carbon sources [17,89]. This suggestion is supported by the finding that a N. crassa gdhA deletion mutant, inhibited in GS activity, was able to grow with glutamine as sole carbon-and nitrogen source in contrast to the wild type [90]. The growth defect of the wild type under these conditions was attributed to the highly active ATP-consuming glutamine/glutamate cycling (GS-GOGAT) [91] contributing to an ATP-drain that would drive the ATP-producing glycolysis in the presence of glucose [90,92]. Similar observations have been reported in S. cervisiae [93]. The data obtained in this work make a comparable mechanism in F. fujikuroi highly conceivable. A disrupted ATPconsuming glutamate/glutamine cycling mechanism could explain the low growth rate of the ∆gln1 mutant with low glutamine concentrations when glucose is present. Glutamate accumulates due to the missing conversion back to glutamine by the GS, and the unused ATP would accumulate and repress efficient glycolysis ( Figure 7C). Evidence that an inactivated carbon/energy metabolism in the Δgln1 mutant is responsible for the loss of SMs comes from the fact that the wild type did not produce bikaverin or GAs when glucose was absent and glutamine served as the only nitrogen and carbon source. In accordance with this hypothesis, the three deregulated mutants showed a partial restoration of the growth defect. Although they are strictly glutamine auxotrophic, they produce significantly more biomass than the ∆gln1 mutant indicating that the enzymatically inactive GS enzymes with mutations in D60/S62, G246/G248 and S72/D73 residues are partially able to regulate the energy metabolism ( Figure 7D).
In summary, detailed comparisons of growth rates, gene expression analyses and SM production between the wild type, the Δgln1 mutant and transformants carrying heterologous or specifically mutated GS-encoding alleles allowed us to postulate that the GS not only provides glutamine for cell viability, but also functions as a regulatory hub by controlling gene expression. We showed that the GS most likely senses the abundance of its substrates, NH 4 + and glutamate, and transmits this information to downstream targets by an as yet unidentified mechanism that involves AreA (GA biosynthesis) and other yet unknown (bikaverin biosynthesis) regulators. Our data indicate that the GS is tightly linked to carbon/energy metabolism, thereby connecting primary to secondary metabolism ( Figure 7). Furthermore, by generating specific site-directed GS mutants we were able to separate the enzymatic function from the role as a major regulator of many metabolic processes. Three point mutations led to partial restoration of secondary metabolism, wild-type-like growth behavior, intracellular amino acid concentrations, and gene expression patterns despite their inability to produce glutamine. Furthermore, the putative NH 4 + binding site D60/S62 in the GS seems to be responsible for NH 4 + sensing independently of enzymatic functionality. These data support our assumption of a regulatory role of the mere GS protein disclosing the possibility for putative direct interaction partners. We therefore suggest that the F. fujikuroi GS, similar to bacterial GS, belongs to the group of bifunctional "trigger enzymes" that recognize their substrates and often undergo structural alterations upon this interaction. These enzymes are active in metabolism and in the same time are able to regulate gene expression, probably by their interference with transcriptional regulators [94]. Unraveling the effects caused by an altered metabolic status (carbon and energy) in the Δgln1 mutant and/or by altered signal transduction of putative GS interaction partners will be a fascinating future task.