Interaction of Yna1 and Yna2 Is Required for Nuclear Accumulation and Transcriptional Activation of the Nitrate Assimilation Pathway in the Yeast Hansenula polymorpha

A few yeasts, including Hansenula polymorpha are able to assimilate nitrate and use it as nitrogen source. The genes necessary for nitrate assimilation are organised in this organism as a cluster comprising those encoding nitrate reductase (YNR1), nitrite reductase (YNI1), a high affinity transporter (YNT1), as well as the two pathway specific Zn(II)2Cys2 transcriptional activators (YNA1, YNA2). Yna1p and Yna2p mediate induction of the system and here we show that their functions are interdependent. Yna1p activates YNA2 as well as its own (YNA1) transcription thus forming a nitrate-dependent autoactivation loop. Using a split-YFP approach we demonstrate here that Yna1p and Yna2p form a heterodimer independently of the inducer and despite both Yna1p and Yna2p can occupy the target promoter as mono- or homodimer individually, these proteins are transcriptionally incompetent. Subsequently, the transcription factors target genes containing a conserved DNA motif (termed nitrate-UAS) determined in this work by in vitro and in vivo protein-DNA interaction studies. These events lead to a rearrangement of the chromatin landscape on the target promoters and are associated with the onset of transcription of these target genes. In contrast to other fungi and plants, in which nuclear accumulation of the pathway-specific transcription factors only occur in the presence of nitrate, Yna1p and Yna2p are constitutively nuclear in H. polymorpha. Yna2p is needed for this nuclear accumulation and Yna1p is incapable of strictly positioning in the nucleus without Yna2p. In vivo DNA footprinting and ChIP analyses revealed that the permanently nuclear Yna1p/Yna2p heterodimer only binds to the nitrate-UAS when the inducer is present. The nitrate-dependent up-regulation of one partner protein in the heterodimeric complex is functionally similar to the nitrate-dependent activation of nuclear accumulation in other systems.


Introduction
The ascomycete yeast Hansenula polymorpha (syn. Pichia angusta) is able to use nitrate as sole nitrogen source [1]. The genes involved in nitrate assimilation are inducible by the substrate and encode nitrate reductase (YNR1), nitrite reductase (YNI1) and a high affinity nitrate transporter (YNT1). These co-regulated genes reside in a cluster together with the two pathway-specific regulatory genes YNA1 and YNA2 [2] (Fig 1A).
Nitrate transport has a key role for the entrance of the inducer in the cell where the transporter acts as a modulator that triggers nitrate gene expression. Unlike Arabidopsis thaliana, where the nitrate transporter CHL1 plays the role as nitrate sensor independently to that of nitrate transporter, H. polymorpha Ynt1 and A. nidulans NrtA/NrtB do not show this sensor activity [1,3,4]. As in A. nidulans and N. crassa, also in H. polymorpha the expression of the transporter and enzymes involved in nitrate assimilation is subject to both nitrate induction and nitrogen metabolite repression, which blocks the nitrate pathway when the products of the reduction process such as ammonium or glutamine, are available in the growth medium [2,[5][6][7][8][9][10][11][12][13]. Depending on the nitrogen source, Ynt1 is subject to post-translational modifications by ubiquitination and phosphorylation. Recently, Martìn and co-workers (2011) identified and described the NPR1 gene, encoding the Npr1 (Nitrogen Permease Reactivator 1) kinase involved in Ynt1 phosphorylation and regulation in response to the nitrogen and carbon source availability [14,15].
The YNA1 and YNA2 regulatory genes, also located in the nitrate utilization cluster, encode two binuclear Zn(II) 2 Cys 6 trans-activators involved in transcriptional activation of genes necessary for nitrate utilization [5,6]. These factors belong to the same fungal-specific protein family as the Aspergillus nidulans NirA [16] and Neurospora crassa NIT4 [17] proteins even if they are not their orthologues [18]. The latter reference also discusses the variation in clustering in different fungi and the transcription factors associated with the nitrate assimilation cluster in different yeasts able to assimilate nitrate. Yna1p shows a uniquely deviant Zinc cluster motif with seven amino acids, instead of six, between the second and the third cysteine [19,20]. C-terminal of the DNA binding domain Yna2p contains a region of periodically spaced leucines which may form a leucine-zipper domain involved in protein-protein interactions. Interestingly, NirA contains a similar motif and there it was shown to be essential for in vitro binding to cognate DNA sequences [21].
In eukaryotes, the chromatin structure plays an essential role in the modulation of transcriptional regulation [22][23][24][25][26], where chromatin remodelling enables interaction between the transcriptional machinery and the target DNA sequences [24,[27][28][29]. Among the Zn(II) 2 Cys 6 transcription factors, Gal4p of S. cerevisiae directly promotes chromatin remodelling associated with expression of their target genes [29][30][31][32]. Chromatin transitions in nitrate assimilation pathways were also studied in N. crassa [33] and A. nidulans [34][35][36][37]. Interestingly, in N. crassa the formation of hypersensitive chromatin regions at the nit-3 gene promoter (coding for nitrate reductase) is mediated by both regulators NIT4 (nitrate-specific transcription factor) and NIT2 (mediating nitrogen metabolite de-repression) independently of the induction process [31]. In A. nidulans, while both nitrate induction and AreA are required to restructure the chromatin landscape in the bidirectional niiA-niaD promoter [37], NirA (the NIT4 ortholog) is only required for sliding nucleosome -1 and + 1 around the central binding sites for both NirA and AreA. Therefore, the majority of chromatin reorganisation depends on AreA and can take place independently from transcriptional activation of the cognate genes. Some years ago we have shown that one function of AreA is to recruit histone acetylation activities to the bidirectional niiA-niaD promoters under nitrogen metabolite de-repressing conditions, i.e. in the absence of repressive nitrogen sources such as glutamine or ammonium [11,12,37]. The H. polymorpha nitrate utilization cluster (GenBank #AJ223294) and its regulation. A. The complete sequence (nt 1-13684) consists of five nitrate assimilation genes (grey shaded) flanked by a predicted glutathione S-transferase (ORF-1) on the one side and by a predicted RAD3 homolog (ORF-3) on the other. Numbers (in bp) above the coding regions correspond to the size of the ORFs and numbers between the genes indicate the length of the intergenic regions. YNT1, nitrate transporter; YNI1, nitrite reductase; YNA1, Zn(II) 2 Cys 6 transcriptional activator; YNR1, nitrate reductase; YNA2, Zn(II) 2 Cys 6 transcriptional activator. B. Northern blots showing the effect of different nitrogen sources on the expression of the nitrate assimilation genes in the H. polymorpha wild type strain (L1). The actin gene mRNA signal was used as loading control. Nitrogen sources were 50mM alone or in combinations as indicated above the autoradiographs. N, nitrate; A, ammonium; Q, glutamine; E, glutamate; P, proline. Mean densitometric values (± standard error) of separate experiments, normalised to the loading control signal, are reported (expressed as percentage of maximum induction value i.e., growth on nitrate). nd, not-detectable. C. Northern blot analysis investigating the role of YNA1 and YNA2 in the expression of the nitrate cluster genes. Mutant strains are derived from the wild type (L1) and carry either a marker insertion at the YNA1 locus (L1-yna1 I6632 ) or maker insertion and internal deletion at the YNA2 locus (L1-yna2 Δ12274-12595 ). The actin gene signal was used as loading control.
In this study we identified differential roles for Yna1 and Yna2 in the transcriptional activation events. In particular, we gained insights into (i) their subcellular localization, (ii) in vitro and in vivo binding to putative target sites (iii) nucleosomal positioning dynamics at the YNT1 promoter in response to different physiological conditions.

Yeast strains
All strains used in this work are derivatives of H. polymorpha homothallic haploid NCYC 495. The NCYC495 derivative strain L1 (leu1-1) [38] was used as recipient strain to obtain the YNA1 and YNA2 null mutants, respectively denoted L1-yna1 16632 and L1-yna2 Δ12274-12595 . L1 was used as wild type strain throughout this work. Mutant strains L1-yna1 16632 (leu1-1, yna1 16632 ::LEU2) and L1-yna2 Δ12274-12595 (leu1-1, yna2 Δ12274-12595 ::LEU2) were constructed by standard methods. L1-yna1 16632 was obtained by target integration of a 4095 bp-DNA fragment containing the Candida albicans LEU2 gene flanked by YNA1. The 4095 bp DNA fragment (LD-I cassette; S2 Fig) was amplified by PCR from the pYNA1LEU-I plasmid using primers CCYNA1F and CCYNA1R (S1 Table). Transformants containing the yna1 16632 ::LEU2 mutation were selected as Leu + Nitstrains, and the presence of a single homologous integration was checked by PCR and Southern blots. L1-yna2 Δ12274-12595 was constructed by substitution of the 322 bp ClaI-ClaI fragment of the YNA2 coding region, with the Candida albicans LEU2 gene. The DNA fragment containing the Candida albicans LEU2 gene flanked by YNA2 (D2 cassette; S3 Fig) used to obtain this mutant by target integration was amplified by PCR from the pYNA2LEU plasmid using primers YNA2Ampf and YNA2AmpR (S1 Table). Transformants, containing the yna2 Δ12274-12595 ::LEU2 mutation, were selected as Leu + Nitstrains and the presence of a single homologous integration was checked by PCR and Southern blots.

Growth conditions
All H. polymorpha strains were grown at 37°C in rich medium (YPD) containing 2% glucose, 1% yeast extract and 2% peptone, or in a minimal medium (MM) containing 2% glucose, 0.2% yeast nitrogen base lacking of both amino acids and ammonium sulphate (Difco, Detroit, MI, USA), plus 0.5% ammonium sulphate (corresponding to 75mM NH 4 + ) or 50 mM of proline, glutamine or glutamate. When necessary, 1.8% agar or 0.006% leucine were added. The standard growth conditions for northern blots and MNase digestion were as follows: cells were grown in 250 ml flasks with 50 ml medium in an orbital incubator at 37°C and 200 rpm. Preculturing in YPD for 15 hours, the cells were washed with sterile water and inoculated (10 6 cells/ml) on minimal medium containing 50 mM proline as nitrogen source. After 16 hours culture the cells were again washed with sterile water and then transferred to a MM containing one of the following nitrogen sources: 50 mM NaNO 3 , 50 mM NH 4 Cl, 50 mM L-glutamine, 50 mM Na-glutamate, 50 mM L-proline, 50 mM NH 4 NO 3 , 50 mM NaNO 3 + 50 mM L-glutamine, 50 mM NaNO 3 + 50 mM Na-glutamate, 50 mM NaNO 3 + 50 mM L-proline. After the optimal repressing conditions were established, four standard conditions were used: 50 mM L-proline (not-induced-de-repressed condition, NI); 50 mM NaNO 3 (induced-de-repressed conditions, I); 50mM L-glutamine (not-induced-repressed condition, R); 50 mM L-glutamine + 50 mM NaNO 3 (not-induced-repressed conditions, IR). When not otherwise stated, cultures were further incubated for 2 hours and then harvested.
The growth conditions for in vivo footprinting of H. polymorpha were as follows: cells were grown in 500 ml flasks with 100 ml of MM containing 50 mM proline as nitrogen source at 37°C and 180 rpm. After pre-culturing for 15 hours, cells were washed with sterile water and inoculated (106 cells/ml) in MM containing one of the following nitrogen sources: 50 mM Lproline (not-induced-de-repressed condition, NI) or 50 mM NaNO 3 (induced-de-repressed condition, I). Cultures were further incubated for 2 hours. Guanines and adenines methylation of genomic DNA in whole and living cells was performed by adding to each flask 10 ml of a solution containing 9,925 ml of 20 mM MES buffer pH 6.2 and 75 μl of pure dimethylsulphate (DMS, Sigma-Aldrich). Cells were incubated for 2 minutes at 37°C and 180 rpm. Methylation was stopped by adding TLE-β stop buffer (Tris-Cl pH 8.0, 10 mM; EDTA pH 8.0, 1 mM; LiCl 100 mM; β-mercaptoethanol 2%). Cells were then harvested for genomic DNA extraction.
Growth conditions for H. polymorpha strains L1-GFP-1 and L1-GFP-M2 harbouring the chimeric constructs YNA1(p)::YNA1::GFP and MOX(p)::YNA2::GFP respectively, were as follows: cells were pre-cultivated in 100 ml flasks with 20 ml of MM supplemented by 50 mM proline as nitrogen source, at 37°C and 180 rpm. Cells were washed with sterile water and resuspended in 20 ml of MM w/o nitrogen source. 3 ml of each sample was aliquotted in a 15 ml-tube containing MM plus 50 mM of nitrogen source related to the physiological condition under investigation. Cells were then incubated at 37°C, 180 rpm, at different times (from 20 minutes to 2 hours) and observed by confocal microscopy.
The H. polymorpha strain AL-YFP18, harbouring the chimeric cassettes YNA1(p)::YNA1:: YFP B ::LEU2::YNA1(t) and YNA2(p)::YNA2::YFP C ::PUR7::YNA2(t), was pre-grown in 100 ml flasks containing 20 ml of MM supplemented by 50 mM Glutamine as sole nitrogen source. Cells were washed with sterile water and resuspended in 20 ml of MM w/o nitrogen source. 3 ml of each sample was aliquotted in 15 ml-tube containing MM plus 50 mM of nitrogen source related to the physiological condition under investigation. Cells were incubated at 37°C, 180 rpm for 2 hours and observed by confocal microscopy.

Plasmids
Plasmid pA1 contains YNA1 amplified using primers YNA1ef and YNA1dr and inserted in EcoRI of EBOX37p6. pA2 contains the YNA2 gene, amplified using primer YNA2df and YNA2dr cloned in BamHI of EBOX42p1. EBOX37p6 and the EBOX42p1 are two H. polymorpha replicative plasmids [PMR1 (GeneBank # AJ011906), H. polymorpha LEU2, E. coli ori; KanR (EBOX37p6) or AmpR (EBOX42p1)]. The pYNA1LEU-I plasmid, containing the LD-I cassette (S2 Fig), derives from pTZ18 (Pharmacia) and was constructed as follows: a 2352 bp DNA fragment containing the YNA1 locus and its regulatory regions was amplified using the YNA1df and YNA1dr primers, that contain the SalI and the EcoRI restriction sites, respectively, at their 5' ends. Genomic DNA was used as template. The amplified fragment was cloned into pTZ18 previously cut with SalI and EcoRI. Then, a 2122 bp DNA fragment containing the Candida albicans LEU2 was amplified from the pMK155 plasmid (a gift from R. D. Cannot; University of Otago, New Zealand) using primers leu2cf and leu2cr and cloned into the EcoRV restriction site of this intermediate construct (EcoRV is in the YNA1 gene), the latter having been previously linearized with EcoRV and then dephosphorylated. The LEU2 containing PCR fragment was cut with SmaI and then ligated with the plasmid. The construction was confirmed by restriction analyses and sequencing (automatic sequencing; MWG biotech, Ebersberg, Germany). The pYNA2LEU plasmid, containing the D2 cassette (S3 Fig), derives from pTZ18 (Pharmacia) and was constructed as follows: a 2777 bp DNA fragment containing YNA2 was amplified using the primers YNA2df and YNA2er, which contain BamHI at the 5' ends. Genomic DNA was used as template. The amplified fragment was cloned into BamHI linearized pTZ18. Successively, this intermediate construct was cut with ClaI; the ClaI sticky ends were filled-in using the Klenow DNA polymerase I fragment (Promega, Madison, WI, USA), dephosphorylated and finally ligated with a SmaI/ SmaI-end 2122 bp DNA fragment containing the Candida albicans LEU2 gene obtained as described above. The construction was confirmed by restriction analyses and sequencing.
For the expression of Yna1p  and Yna2p  truncated forms in E. coli, YNA1  and YNA2  were cloned by Gateway Cloning Technology, following manufacturer instructions. Yna1p  and Yna2p  contain putative nuclear localisation signals, activation domains and Zn(II) 2 Cys 6 binding domains. In addition, Yna1p (1-185) harbours a putative nuclear export signal (NES), which is not predicted for Yna2p. YNA1  and YNA2  were amplified from plasmids by using primers GATYNA1_fw, GATYNA1tr_rv and GATYNA2_fw, GATYNA2tr_rv respectively. pDEST17 destination vector was used as final expression vector in E. coli BL21DE3 strain. Gene expression is driven by E. coli T7 promoter and mediated by T7 RNA polymerase.
Plasmid pH1 (Lab. E. Berardi, auto-replicative vector Kanr, ScLEU2, MOX(p)::eGFP-Tamo derived from pHIPX4, [39] was used for YNA1(p)::YNA1::GFP and MOX(p)::YNA2::GFP fusions. A SmaI/NotI YNA1-fragment of 2219 bp and a BamHI/SmaI YNA2-fragment of 2206 bp were amplified by PCR using genomic DNA as template. The chimeric fragment YNA2:: GFP was assembled under the control of methanol-inducible promoter (MOX) from H. polymorpha methanol-oxidase gene because of the weak fluorescent signal detectable under control of YNA2 native promoter. Primers YNA1gfp_fw, YNA1gfp_rv, BamYNA2gfp_fw and YNA2dr were used. YNA1 and YNA2 PCR products were respectively cloned into SmaI/NotI and BamHI/NotI restriction sites in the linearized plasmid pH1. Constructions were confirmed by restriction analysis, PCR amplification and sequencing (Agowa DNA sequencing, GmbH). All primers used are indicated in S1 Table. Bimolecular fluorescence complementation Bimolecular fluorescence complementation is a technology used to prove the interaction of two proteins belonging to the same macromolecular complex. In this work, Yna1p and Yna2p have been fused to unfolded complementary fragments of yellow fluorescent protein (YFP) and expressed in the H. polymorpha control strain. Firstly, a cassette for C-terminal genomic integration for both YNA1 and YNA2 was constructed by fusion PCR. Each cassette harbors the ORF C-terminal and the terminator sequences of the target gene as flanking regions, the YFP unfolded fragment (YFP B for YNA1 and YFP C for YNA2) and the selection marker. As selection markers, CaLEU2 and PUR7 were used. YNA1 and YNA2 cassettes were expressed in L1 (leu1.1, mat a) and in A11 (ade11.1, mat α) strains, respectively. L1 and A11 were crossed and sporulated on ME plates as described [40]. Spores were analyzed by PCR and phenotype selection. The expression of each construct integrated in the YNA locus is under the control of the native promoter. The resulting strain AL-YFP18 carrying YNA1(p)::YNA1::YFP B ::CaLEU2:: YNA1(t) and YNA2(p)::YNA2::YFP C ::PUR7::YNA2(t) genomic modifications was pre-cultivated for 16-18h in MM containing glutamine as sole nitrogen source. After pre-cultivation, cells were harvested, washed three times and cultivated in minimal medium containing nitrate (induction, I) or proline (not-induction-de-repression, NI) or glutamine (repression) as nitrogen source. After 2h, cells were observed by confocal microscopy.

Molecular techniques
Standard recombinant DNA techniques were carried out according to Sambrook et al. [41]. Transformation of H. polymorpha was performed by the lithium acetate method described by Berardi and Thomas [42]. Restriction enzymes and biochemicals were obtained from Promega (Madison, WI, USA), Fermentas (Burlington, Ontario, Canada) and Sigma-Aldrich (St. Louis, Missouri, USA) and used as detailed by the manufacturer.

Northern blots
Total RNAs were isolated from cells as described by Schmitt et al. [43]. Equal amounts (10 μg) of total RNA were loaded per lane and separated on glyoxal agarose gel. Blots were hybridised with the YNT1, YNI1, YNR1, YNA1 or YNA2 probes labelled with 32α dCTP using a Random Priming Kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA). The actin probe (HpACT) [44] was used to normalize the amounts of RNA loaded on a blot. All probes used to detect the different transcripts were amplified from genomic DNA by PCR using the following primers: YNT1, YNT1SF and YNT1SR; YNI1, YN2SF and YN1SR; YNR1, YNRSF and YNR1SR; YNA1, YNA1SF and YNA1SR; YNA2, YNA2ST and YNA2SF; HpACT, ACTesa (forward) and ACT1HPF (reverse) (S1 Table). Northern blots were quantified using a PhosphorImager with ImageQuant software (Molecular Dynamics, Sunnyvale, CA, USA). Northern blots presented are typical examples from at least two experiments. When quantitative measures are shown these are averages and standard errors from 5 independent determinations.

Low resolution micrococcal nuclease mapping
Micrococcal nuclease (MNase) sensitivity analyses were carried out by the indirect end labelling technique, using the method described for filamentous fungi by Gonzalez and Scazzocchio [36]. This method was adapted to yeast H. polymorpha as indicated below. The cells were grown as indicated above and harvested by centrifugation. The remaining water was eliminated by a rapid filtering through a sterile membrane filters (1,2 μm; Millipore, Bedford, MA, USA). Successively the cell-cream was immediately frozen in liquid nitrogen and then reduced to powder by grinding under liquid nitrogen in a mortar. The powder was suspended in MNase digestion buffer to a final concentration of 100 mg of cell powder/ml (nystatin was added to a final concentration of 50 μg/ml). MNase digestion was started by the addition of different amounts of the enzyme to 200 μl of this suspension, and then incubated for 10 min at 37°C. The reaction was stopped by adding 1 vol. of 40 mM EDTA, 2% SDS. The mixture was extracted with phenol-chloroform, then treated with 10 mg/ml RNase A (10 min.-37°C), precipitated from 2 vol of ethanol and the pellet suspended in 50 μl of TE pH 7.4.
In preliminary experiments the nucleosomal repeat length was estimated at 145 bp, consistent with that previously reported by Costanzo and coworkers [45] for H. polymorpha. The MNase concentrations varied in a range 3.75-7.5 U/g of cell powder. Naked DNA samples were digested with MNase (from 0.003 to 0.05 U/μg of DNA) in 200 μl of digestion buffer (7 min, 30°C) and then processed as described for the other samples above.
For low resolution mapping of nucleosomes by indirect end labelling the purified DNAs were subject to a secondary digestion by EcoRV. The specific DNA sequences were detected by hybridisation with the YNI1 probe that is the 348 bp EcoRI-EcoRV fragment from the I-1 PCR product. The probe is labelled with 32α dCTP using a Random Priming Kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA). The I-1 PCR product was obtained using the primer forward croYNIF and croYNIR (S1 Table). Blotting and hybridization were as in Wu (1980) [46]. For each strain and growth condition experiments were performed at least in duplicate. In order to determine the size of each DNA band and to make an accurate comparison of samples run on different gels, a Pharmacia 100 bp ladder and a Smart Ladder (0.2 Kbp to 10 Kbp; Eurogentec, Seraing, Belgium) were included in every gel. In all cases, gels were scanned and the size of the bands estimated using the Molecular Analyst/Macintosh Software (Bio-Rad, Hercules, CA, USA), with the ladders as standards.  and HIS-Yna2p (1-180) purification HIS-Yna truncated proteins expressed in E. coli were purified by nickel affinity chromatography (NTA-Ni ProBond resin), as indicated by manufacturer. Protein elution from ProBond resin was done by using an imidazole gradient from 10 to 200 mM. Protein concentration was determined by colorimetric Bradford assay (λ = 595 nm) and spectrophotometric evaluation (λ = 280 nm), using BSA standard curve as reference.

In vivo footprinting
Growth conditions for H. polymorpha were indicated above. In vivo footprinting method for Aspergillus nidulans [47] was adapted to H. polymorpha. After extraction of methylated DNA, a 10-fold diluted piperidine (Roth, Germany) was added to each DNA aliquot. Samples were incubated for 3 minutes at 90°C and cooled on ice. In vitro methylation was performed by dissolving DNA in 200 μl of MES buffer and 1 μl of DMS and incubating the mixture at 90°C for 90 s. After ethanol precipitations, DNA was resuspended in TE at final concentration of 0.7 mg/ml. A sample DNA aliquot was loaded on agarose gel to confirm the existence of a smear containing fragments in a range of 0-1000 bp. Methylated and piperidine treated DNA samples were analysed by LMPCR [47], using Vent polymerase (NEB). [ 32 P]ATP-labelled amplification products were separated on a sequencing gel which was dried and exposed to intensifying screens. To investigate the H. polymorpha YNT1 promoter, Primer1_sense, Primer2_sense, Pri-mer3_sense were used (S1 Table).

Chromatin Immunoprecipitation (ChIP)
H. polymorpha cells were grown as indicated above and treated with formaldehyde at a final concentration of 1% for 20 min at 37°C and shaking at 180 rpm, and subsequently, glycine was added to a final concentration of 2,5 M and incubated for additional 10 min. Cells were harvested and washed with ice-cold PBS and subsequently treated for 30 min at 30°C with a buffer containing Zymolyase to obtain spheroplasts. Cells were harvested by centrifugation and resuspended in lysis buffer. Each sample was sonicated at 20-22 amplitude for 5 times, 5 min each, with intervals of 1 min. Cell extracts were immunoprecipitated overnight using magnetic beads (Dynabeads Protein G and Dynabeads Protein A, Novex, Life Technologies) coated with Ab290 Abcam anti-GFP antibody. In the next step, each mixture was washed and the Yna:: GFP-DNA complexes eluted using TES buffer at 65°C. Samples were treated with Proteinase K for 1,5 h and DNA isolated by phenol-chloroform extraction and 2-propanol precipitation. Each pellet was eluted in 30 μl of Tris-EDTA solution. Samples were analysed by quantitative PCR, using different concentrations of genomic DNA ranging from 10 ng to 10 pg as serial dilutions to generate a standard curve. The sequence of the primers YT_Ch_fw and YT_Ch_rv used to amplify a region of 114 bp containing the Yna1/Yna2 binding site are indicated in S1 Table. The PCR cycle used is as follows: 95°C, 3 min; 50x(95°C, 30 sec; 60°C, 20 sec; 72°C, 30 sec); 95°C, 15 sec. Relative amounts of precipitated DNA are expressed as ratio between the amount of DNA obtained after ChIP relative to the amount of DNA contained in the respective sample before precipitation (input DNA).

Confocal microscopy
Growth conditions for H. polymorpha cells were as indicated above. 5 μl of each culture were spotted on a cover slide. For Yna1p::GFP and Yna2p::GFP subcellular fluorescence detection, a Leica TCS SP2 confocal microscope was used (GFP excitation 488 nm, Ar-Kr laser; GFP emission measured between 500 and 555 nm). For YFP (excitation 514 nm, emission 527 nm) and DAPI (excitation 350 nm, emission 470 nm) detection an Olympus FV1000 microscope was used.

Regulation of the nitrate utilisation cluster
Similarly to other fungi, transcriptional activation of nitrate assimilation genes in H. polymorpha requires the presence of nitrate and, at the same time, the absence or low levels of reduced nitrogen sources [5,6,[48][49][50]. To set the appropriate experimental conditions for subsequent molecular analyses, we first incubated cells under a variety of nitrogen conditions. We also investigated whether different reduced nitrogen compounds afford different levels of nitrogen metabolite repression when present at equimolar amounts in the medium simultaneously with the inducer. Fig 1B shows that glutamine is the most efficient repressing source, followed by glutamate and ammonium. All the nitrate cluster genes, with the exception of the constitutively transcribed YNA2 gene are strongly repressed by glutamine, whereas addition of ammonium does not result in complete repression. The least repressive effect is exerted by proline as the steady state mRNA levels on proline plus nitrate are from 40% to 85% of those obtained upon incubation on nitrate alone. This indicates that nitrogen catabolite repression (NCR) depends on the metabolic rate by which a given metabolite is converted to the possibly direct effector for NCR, glutamine. Induction of the structural nitrate utilisation genes (YNT1, YNR1, YNI1) has a considerable lag-time of around 40 minutes whereas the regulatory gene YNA1 is already transcribed after 20 minutes of transfer to nitrate (S1 Fig). YNA2 levels show an expression peak 20 minutes after the transfer to nitrate (S1 Fig). Our subsequent work was carried out (unless otherwise indicated) under four physiological conditions, (i) not-induced-de-repressed, NI (proline); (ii) induced-de-repressed, I (nitrate); (iii) induced-repressed, IR (nitrate + glutamine); (iv) not-induced-repressed, R (glutamine).

YNA1 and YNA2 form a positive feedback loop
As expected from previous work, both YNA1 and YNA2 mutants (L1-yna1 I6632 and L1-yna2 Δ12274-12595 ) were unable to utilise nitrate or nitrite as nitrogen sources due to the lack of expression of assimilatory and transporter genes (Fig 1B and 1C). Each mutation was complemented by a replicative plasmid carrying respectively the YNA1 or the YNA2 gene (see Materials and Methods and S2 and S3 Figs). As shown in Fig 1C, a null mutation in either YNA1 or YNA2 results in non-detectable expression of YNR1, YNI1 and YNT1 under all growth conditions tested. The only exception was YNT1, which shows very low, albeit detectable expression on nitrate in the YNA2 mutant. In their initial characterization of YNA1 and YNA2 Ávila and co-workers [6,51] found that YNA1 expression is independent from YNA2. Yet in our slightly different genetic background (NCYC495 -leu ) this is only true for nitrate as sole nitrogen source, when other nitrogen sources were employed (e.g. proline plus nitrate, ammonium plus nitrate), the absence of YNA2 led to a defect in YNA1 expression. This clearly indicates that Yna2p is required for full expression of its partner activator in this genetic background.

Yna1p nuclear localization is Yna2p-dependent
To study whether the transcription factors are regulated by nuclear-cytoplasmic shuttling, the YNA1 and YNA2 genes were fused to GFP. YNA1-GFP expression was driven by its native promoter, whereas YNA2-GFP was under the control of the methanol-inducible promoter (MOX (p) from methanol oxidase gene). Heterologous YNA2-GFP expression by the MOX promoter was chosen because Yna2p-GFP should also be studied in YNA1Δ backgrounds and, as determined above, the YNA2 gene is almost not expressed in a strain lacking Yna1p (L1-yna1 I6632 , Fig 1C). Both constructs contained in an autonomous, self-replicating plasmid were transformed into the wild type, YNA1Δ and YNA2Δ backgrounds and strains were grown under all relevant physiological conditions. To detect Yna2p::GFP, the fusion gene was always expressed to high levels using methanol as inducer. Confocal microscopy of the transformants containing the GFP fusion constructs in the wild type background (labelled L1) revealed that both Yna1p:: GFP and Yna2p::GFP are located predominantly in the nucleus under all physiological conditions tested, namely non-inducing (NI, proline), inducing (I, nitrate), inducing-repressing (IR, nitrate and glutamine) or repressing (R, glutamine) conditions. The fact that Yna1p::GFP is detected at all on glutamine is remarkable as YNA1 mRNA is barely detectable under these repressing conditions (Fig 1B, "Q" conditions). This indicates that the fusion protein must be very stable. In the YNA1Δ strain (L1-yna1 I6632 ), Yna2p::GFP also locates constitutively and predominantly to the nucleus (Fig 2E), indicating that Yna1p does not play a significant role to bring Yna2p to the nucleus or to keep it there.
This demonstrates that the predicted putative NLS in Yna2p is functional under all physiological conditions, either in the presence or the absence of the co-activator Yna1p. This is in contrast to the dynamics of Yna1p subcellular distribution: in the YNA2Δ strain (L1-yna2 Δ12274-12595 ), Yna1p::GFP has lost its capacity to accumulate in the nucleus and uniformly distributes throughout the cell under all tested physiological conditions (Fig 2C and S4 Fig). This suggests that one function of Yna2p is to either facilitate import and/or to prevent export of Yna1p. This finding may provide an explanation for the YNA2Δ phenotype as in this strain the amount of accumulated Yna1p in the nucleus and thus transcriptional activity seems to be reduced.
Yna1p and Yna2p physically interact on non-inducing and inducing, but not on repressing nitrogen sources Results presented above suggest a direct interaction between Yna1p and Yna2p and such a complex may act as hetero-dimer to bind cis-acting elements to activate their target genes. To directly test this hypothesis we constructed split-YFP fusions of Yna1p::YFP Nterm and Yna2p:: YFP Cterm . The proteins were fused to partial fragments of YFP and in this case both fusion genes were transcribed from their native promoters (see Materials and Methods for details) and tested for their interaction. The appearance of yellow fluorescence at both non-induced and induced conditions indicates that the two proteins physically interact in vivo (Fig 3) and that this interaction is independent from the presence of the inducer.
This corroborates our previous results that nuclear accumulation of Yna1p depends on the presence of Yna2p which putatively regulates the function of Yna1p subcellular localization signals. In this context, Yna2p may either mask the Yna1p NES or be involved in exposing the NLS of Yna1p. A clear signal was only detected under inducing (I) and non-inducing conditions (NI) but was absent under repressing conditions (R, glutamine). This result could be due to the small amount of Yna1p::YFP, that is expected to be present in cells grown on glutamine although the Yna1p::GFP signal was readily detected on this N-source (compare S4 Fig). Alternatively, glutamine as N-source may disrupt the Yna1p-Yna2p interaction. The fact that Yna1p::GFP is readily detected might also indicate that the GFP signal requires less protein than detection of the YFP signal. Taken together, our studies on the subcellular localization and the split-YFP experiments suggest that physical interaction between Yna1p-Yna2p occurs in vivo and that this heterodimeric complex triggers transcriptional up-regulation of one of the regulatory genes (YNA1) as well as of the structural genes necessary for nitrate assimilation. Yna1p and Yna2p bind cis-acting sites in the YNT1 promoter An in silico analysis of the structural and regulatory gene promoters (Fig 4A) revealed the presence of a conserved, asymmetric motif 5 0 CGGAGA 3 0 flanked by 3 0 AT-rich sequences in  Alignment of the promoter regions of genes belonging to the nitrate cluster detects a conserved motif (5 0 CGGAGA) indicated by six consecutive asterisks. This motif could represent a putative upstream activating sequence (UAS) bound by Yna1p and Ynap2. Probes used in the band shift assays are indicated as YNT1 probe 1 and YNT1 probe 2; the two different derivatives of YNT1 probe 2 containing the indicated changes in conserved nucleotides are shown as m1 and m2. B. In vitro interaction experiments using electric mobility shift assays (EMSA). FP*, labelled probe (double-stranded oligonucleotide as indicated in A) was incubated without recombinant proteins; P*, labelled probe incubated with the indicated recombinant protein but no unlabelled, specific competitor DNA has been added.; +1P, same conditions as in P* but an equimolar amount of the unlabelled double-stranded oligonucleotide was present in the incubation mixture as specific competitor for complex formation; +2P, same conditions as in +1P but double molar amount of the unlabelled double-stranded oligonucleotide was present in the incubation mixture as specific competitor.   doesn't interact with the YNT1 probe 2. Exchanging 3 nucleotides in the conserved 5 0 -CGGAGA-3 0 strech of YNT1 probe 2 disrupts the Yna1p (1-185) binding capability (m2). Likewise, exchange in an upstream conserved residue potentially forming a partial inverted repeat with the CGGAGA motif also disrupts complex formation (m1). This indicates that both motifs are involved in binding Yna1p. C. In vivo footprinting methylation protection profile of the YNT1 gene promoter. Cells from the wild type strain (wt) as well as from the yna mutant strains (y1, corresponding to L1-yna1 I6632 , and y2, corresponding to L1-yna2 Δ12274-12595 ) have been incubated at the standard physiological conditions (I, nitrate; NI, proline) and subjected to ligation-mediated methylation protection footprinting as detailed in Materials and Methods. The sequence pattern of a control reaction containing in vitro methylated DNA is shown in lane V and the corresponding sequence derived from the YNT1 promoter is aligned to the left of the autoradiograph. The sequence corresponding to the banding pattern in the autoradiograph lies between the diagonal dotted lines. In some cases guanines are not properly resolved in the gel but appear as compressed regions. Adenines can become hypersensitive and eventually appear as strong as guanines, mainly in the in vivo lanes. For a better orientation, vertical dotted lines next to the V lane (in vitro control reaction) and next to the sequence correspond to each other. Protected guanines are marked with a triangle and hypersensitive adenines are marked with circles. Asterisks on the displayed vertical sequence correspond to the asterisks marking the conserved residues in the YNT1 promoter region used for designing YNT1 probes 1 and 2. Note that the sequence displayed next to the autoradiograph is complementary to the sequence in panel A. Below the figure, the scheme shows the location of protected regions (white rectangles) in the YNT1 promoter. The conserved 5 0 -CGGAGA-3 0 sequence is indicated by the grey rectangle and asterisks. promoters of the inducible genes YNT1, YNR1 and YNI1, as well as in the promoter of YNA1, but not in the constitutively transcribed YNA2 gene.
Thus, we suspected that these sequences may mediate nitrate induction by providing recognition sites for the Yna1p and Yna2p proteins. On the basis of these nucleotide alignments, we first performed band shift experiments with recombinant forms of Yna1 and Yna2. We synthesized double-stranded, partially overlapping oligonucleotides representing the region -180 to -235 of the YNT1 promoter as probes for the binding assays (Fig 4A). Both probes are similar in length (~40 nucleotides) and contain the conserved 5 0 CGGAGA 3 0 motif, while probe 1 extends further towards the translation start point (-180 to -224), and probe 2 (-200 to -235) extends more distally into the promoter region. The truncated forms Yna1p  and Yna2p  containing the predicted Zn(II) 2 Cys 6 DNA-binding domain, a putative dimerization domain and the predicted NLS (Nuclear Localization Signal; S5 Fig) were expressed as HIS-fusion proteins in E. coli and purified (S6 Fig). The purified proteins were subsequently used in the band shift experiments together with the radiolabelled probes. Both fusion proteins HIS-Yna1p and HIS-Yna2p bound to probe 1 containing the conserved motif and extending towards the transcription start, but t HIS-Yna2 did not bind probe 2 which lacks this latter region (Fig 4B). The formation of the DNA-protein complexes with both probes is specific for HIS-Yna1p under our conditions because an equimolar or two-fold excess of non-labelled probe significantly diminished (probe 1) or even abrogated (probe 2) the formation of the complex. This indicates that the HIS-Yna1p: probe 2 complexes are less stable than complexes formed with probe 1 containing a longer AT-rich stretch in the sequence towards the YNT1 ORF. For HIS-Yna2p the scenario was even clearer, as this protein does not form a detectable complex with probe 2 lacking the 3 0 AT-rich sequences. These data indicate that in vitro both proteins are able to independently form stable and specific complexes with DNA when both the conserved 5 0 CGGAGA3 0 motif as well as the 3 0 AT-rich sequences are available.

In vivo footprinting reveals an extended region protected by Yna1p and Yna2p
We then determined which sequences in the YNT1 promoter are targeted in vivo and under which conditions putative recognition sequences interact with the activators. To this end, we performed in vivo methylation protection footprinting experiments to single out the purines protected by a DNA-binding protein under a given physiological condition [47]. Again, we targeted the YNT1 promoter. The pattern obtained after ligation-mediated PCR is shown in Fig  4C along with the display of the relevant sequence of this promoter. The autoradiograph shows that in the wild type, under fully inducing (I) conditions, an extended protection region becomes evident that overlaps with the region covered by probe 1 and probe 2 in the band shift assays. The protected region contains the strictly conserved CGGAGA motif roughly in its centre and hypersensitive adenines (circled dots) appear in the flanking regions of this motif. This protection is not evident in the sample analysed from non-inducing (NI) conditions, indicating that the nuclear Yna1p-Yna2p heterodimer does not bind DNA in the absence of the inducer. In YNA1Δ and YNA2Δ strains incubated on nitrate (inducing conditions), the protected region is not present and this suggests that in vivo both proteins in the heterodimer are required to form this protection pattern. Due to this large protection window, we were not able to determine exactly which guanines and adenines are participating in the complex-DNA interaction. Besides the protected bases, we also noted the appearance of at least two hypersensitive adenines (indicated by circled dots) in the autoradiograph which were not present in the in vitro methylation control (indicated by V) in a region corresponding to the central part of probe 1 and the end of probe 2. It should be remarked that these hypersensitive bands always specifically appeared in both mutant strains when they were shifted to inducing conditions (nitrate). Generally, the appearance of such residues displaying hypersensitivity towards the methylating agent dimethylsulphate (DMS) is a good indication of local DNA structure distortions generated by binding events. The same principle would apply when Yna1p is binding alone. Note that YNA1 is only expressed to a detectable level in a YNA2Δ strain under nitrate induction (Fig 1C, N condition; Fig 2C) and this fact would make it possible for the hypersensitive adenine to appear in the YNA2Δ mutant. Taken together, our in vivo footprint data suggest that a functional heterodimer of Yna1p-Yna2p binds to the region corresponding to probe 1 and probe 2 and that this binding depends on the presence of both activator proteins.
Yna1p and Yna2p most efficiently bind the YNT1 promoter as heterodimers but both are able to bind also as homodimers  GFP-tagged Yna1p expressed from its own nitrate-inducible promoter in the wild type (Yna1::GFP) and YNA2Δ (Yna1::GFP-YNA2Δ) background was analysed by ChIP in cells grown under nitrate-inducing (I) or glutamine-repressing (R) conditions. B. Experimental set-up for the GFP-tagged Yna2p in the wild ytpe (Yna2 OE ::GFP) and YNA1Δ background (Yna2 OE ::GFP-YNA1Δ) was identical to the one described for Yna1p with the exception that the Yna2p-GFP fusion protein was expressed from the strong MOX promoter in order to be able to analyse Yna2p occupancy also in a YNA1Δ background in which YNA2 is normally not expressed. All values shown in this figure represent the amount of YNT1 promoter DNA relative to the amount of input DNA in the same sample. Experiments were done in two biological repetitions and in each experiment ChIP analysis was duplicated. Error bars thus represent the standard deviations of four measurements in each condition and strain. inducing conditions (I), but can barely be detected under repressing conditions (Fig 5A). In agreement with in vivo footprinting ChIP results also show that high Yna1::GFP binding only occurs when Yna2p is present presumably allowing heterodimer formation. However, the fact that still around 30% of Yna1::GFP can be detected in the YNA2Δ background indicates that Yna1p can also bind to DNA as monomer or homodimer under induced conditions. A similar picture evolved when GFP-tagged Yna2p (indicated as Yna2 OE ::GFP) was tested in the ChIP assay. Also in this case, significantly more Yna2 protein is found associated with the YNT1 promoter region under induced conditions compared to repressed conditions. But a striking feature of Yna2p binding was observed in the strain lacking Yna1p: in this case a roughly 10-fold higher amount of Yna2::GFP is precipitated compared to the amount detected in the wild type background. Strikingly, this hyper accumulation of Yna2::GFP is dependent on the presence of nitrate and does not occur on the repressing nitrogen source (glutamine). These data suggest that also Yna2p -similar to Yna1p -can bind as monomer or homodimer and needs the inducer for this activity. Obviously, in the absence of the heterodimer forming Yna1p, Yna2:: GFP is present at higher amounts indicating that Yna1p balances the heterodimer formation with Yna2p under inducing conditions. The fact that Yna2 expression is driven by the strong MOX promoter cannot solely account for the hyper-accumulation of the GFP fusion protein as the amount of Yna2::GFP present at the YNT1 promoter in the wild type background (putative heterodimer) is only around twice the amount of Yna1::GFP in the same heterodimeric complex (wild type background). In contrast, when Yna1p::GFP is lacking, the amount of Yna2p:: GFP present as homodimer or monomer is roughly 10-fold higher than when Yna2p::GFP occupies the promoter in the heterodimeric Yna1p/Yna2p form.
Chromatin remodelling in the YNT1 promoter depends on both Yna1p and Yna2p As Yna1p and Yna2p directly bind to target promoters and initiate transcription, we determined if chromatin remodelling is also mediated by these two factors. To this end, we performed indirect end-labelling chromatin analysis following MNase digestion of intact H. polymorpha cells grown under different physiological conditions. As shown in Fig 6, there are drastic differences in the nucleosomal organisation in the promoter and the 5 0 ORF region of the YNT1 gene between different physiological conditions. When the gene is silent under non-inducing (NI) and inducing-repressing (IR) conditions, three positioned nucleosomes are detectable in the promoter region (-3, -2, -1 upstream of the ORF and six positioned nucleosomes become evident inside the YNT1 ORF (+1 to +6). Each of these nucleosomes is flanked by MNase sensitive sites located at positions +85, -81, -275, -449, -655. A very strong MNase hypersensitive site (indicated by a bold arrow in Fig 6A) becomes visible in chromatin isolated under conditions in which YNT1 is not transcribed. In addition, there is a larger than expected distance (360 bp) between the MNase cutting sites corresponding to the end of nucleosomes -1 (position -275) and +1 (position + 85). Taken together, both these findings indicate that a nucleosome-free region may be extant between these nucleosomes potentially providing a docking platform for wide-domain and/or pathway-specific transcription factors. Upon nitrate induction a striking nucleosome rearrangement becomes detectable in which all nucleosomes in the transcribed region (+1-+6) as well two promoter nucleosomes (-1 and -2) lose their position. Chromatin restructuring in transcribed regions is known to be a consequence of transcription and thus it is not surprising that, when the gene is induced, nucleosomes starting from the transcriptional start point (nucleosome -1) and downstream lose their positioning. In many cases nucleosomes upstream of the transcriptional start point are remodelled by specific transcription factors and this process is independent from transcription [23]. To test this in H. polymorpha, we performed MNase assays in YNA1Δ and YNA2Δ strains and probed again the YNT1 region for nucleosomal positioning. As seen in Fig  6 (panels B and C) nucleosomal rearrangements do not occur in either mutant. This is visible along the YNT1 ORF and consistent with lack of gene transcription in both strains under inducing conditions. But also the upstream nucleosomes -1 and -2 do not rearrange in the mutants: this indicates that both regulators are involved in remodelling the chromatin in the promoter of this target gene. We also tested for chromatin structure changes for another gene belonging to the pathway. To this end, we targeted the nitrite reductase gene YNI1 and analysed the MNase pattern obtained from cultures grown under conditions where the gene is silent (non-inducing, repressing, inducing-repressing) or activated (inducing conditions). This analysis is presented in S7 Fig and shows an identical picture to the YNT1 pattern. Nucleosomes are strictly positioned under non-expressing conditions and lose this positioning upon Nitrate Assimilation Hansenula polymorpha gene induction. Remarkably, the nucleosome-free-region seems to be more extended in this gene than in YNT1. Nevertheless, a nucleosome-free region still extends here between nucleosomes -1 and + 1. As in YNT1, in YNI1 all nucleosomal rearrangements are strictly dependent on the presence of both transcriptions factors (S7, S8 and S9 Figs).

Discussion
Previous work has shown that in H. polymorpha the nitrate assimilation genes are regulated at the level of their cognate mRNA steady states, and most likely at the transcriptional level [5,6]. We have confirmed, refined and extended these findings, establishing that the appearance of the YNT1, YNI1 and YNR1 mRNAs follows that of the transcriptional activator genes YNA1 and YNA2. The strong constitutive expression of YNA2 suggests that this transcription factor is poised to initiate the regulatory process when nitrate becomes available. It is necessary for basic YNA1 transcription which, through the nitrate-dependent positive autoregulation loop, subsequently further amplifies the induction signal leading to the formation and nuclear accumulation of a functional Yna1p-Yna2p heterodimer. While agreeing with the regulatory role proposed for Yna1p and Yna2p in previous studies [5,6], our experiments with the L1-yna1 I6632 and L1-yna2 Δ12274-12595 mutants disclose new features regarding the auto-regulatory role of Yna1p and Yna2p. In particular, while in Ávila (2002) no effect of the YNA2 mutation on YNA1 expression was reported, our experiments clearly reveal that, in the absence of Yna2p, YNA1 expression is almost completely abolished under all conditions with the exception that it is weakly expressed on nitrate as sole nitrogen source. Considering that the YNA1 gene promoter carries a potential binding site for the activators which may operate as heterodimer, this observation might be consistent with the proposed auto-regulatory function of the Yna1p-Yna2p heterodimer for the expression of the YNA1 gene. Interestingly, YNA2 is constitutively and strongly expressed independently of nitrate (although with a transient induction peak after 20 minutes on nitrate, see S1 Fig) and Yna1p appears to be required for YNA2 expression. This suggests that Yna1p carries out both nitrate-dependent and nitrate-independent functions at the YNA2 promoter. In other words, the functions of the two activators are inter-dependent and, specifically, a positive feedback loop exists in which Yna2p -constitutively present in the cellfacilitates YNA1 gene expression. In turn, the latter acts as co-activator for YNA2 transcription. Upon nitrate availability, this feedback loop would be enhanced and subsequently increase the level of the inducible Yna1p and of downstream targets of the activators, i.e. the structural genes necessary for assimilation of nitrate. We now know that the induction process of the considered nitrate-inducible gene responds only to a functional heterodimer of Yna1p and Yna2p. It is, however, surprising that also the expression of YNA2 is dependent on YNA1 because the YNA2 gene is constitutively transcribed and it does not carry the CGGAGA box present in nitrateresponsive genes (YNA1, YNI1, YNT1, YNR1). This indicates that YNA2 gene expression is not regulated by the Yna1-Yna2 heterodimer. Instead, Yna1p could bind as homo-dimer to a different sequence or may indirectly influence the YNA2 expression, a situation that has been detected in our ChIP analysis of Yna1::GFP in the YNA2 mutant (Fig 5).
The ChIP assay is able to quantify protein:DNA interactions and we could confirm our microscopic analysis of Yna1p::GFP and Yna2p::GFP subcellular localizations but also found that Yna1p can bind sequences in YNT1 promoter in the absence of Yna2p. Plausibly, the binding must occur either as monomer or as homodimer, possibly interacting with different sequences than the heterodimer and providing an explanation for the finding that Yna1p is needed for YNA2 expression although no classical "CGGAGA" motif is present in this promoter. The fact that much less Yna1::GFP is found at the YNT1 promoter in the absence of Ynt2p is most likely due to a reduction in both nuclear accumulation and DNA-binding affinity of the monomer or homodimer. At the moment we can only speculate why less nuclear Yna1p is present in the absence of Yna2p but it is tempting to speculate that the permanently nuclear Yna2p is responsible for masking the NES of Yna1p thus preventing its interaction with the nuclear export machinery. However, this interpretation is only partially in agreement with the split-YFP interaction studies. They showed that the heterodimer between Yna1p and Yna2p does not form in the presence of glutamine (repressing conditions) but our microscopic studies clearly showed nuclear positioning of Yna1p::GFP under all conditions, including glutamine. This would mean that some other mechanism keeps Yna1p::GFP in the nucleus. Alternatively, Yna1p::YFP levels expressed from its own promoter might be insufficient for detection by fluorescence microscopy. This in fact seems to be the most likely explanation because YNA1 is expressed to a very low level on glutamine (compare Fig 1B) and the reconstituted split-YFP is known to be less fluorescent than GFP.
On the other hand it is remarkable that in the absence of Yna1p around 10-fold more Yna2p are detected at the YNT1 promoter compared to the wild type where Yna1p is present in the heterodimer. This surprising result suggests that the Yna2p monomers or homodimers strongly associate with DNA but we do not know if this binding is sequence-specific for the CGGAGA-box or not. In any case Yna1p seems to have an essential role in balancing the heterodimer to equimolar levels and prevent the hyper-accumulation of transcriptionally incompetent Yna2p mono-or homodimers at their common targets in nitrate-responsive genes.
To our surprise, this heterodimer is constitutively located to the nucleus and does not need the inducer to trigger nuclear accumulation. This is in contrast to the best studied fungal model system A. nidulans in which NirA accumulates in the nucleus only when nitrate is present inside the cell [52]. Similarly, also NLP7, a nitrate-responsive transcription factor in Arabidopsis thaliana, accumulates in the nucleus after nitrate induction [53]. It should be also noted that, the Yna1p up-regulation by nitrate and the formation of high amounts of a functional Yna1p-Yna2p heterodimer just under inducing conditions can also be seen as an indirect "nitrate-dependent nuclear accumulation". This mechanism is functionally equivalent, but mechanistically different, to the situation extant in A. nidulans and A. thaliana. Many yeast transcription factors of the Zn(II) 2 Cys 6 type (e.g. Gal4p, Put3p and Leu2p of S. cerevisiae) are also constitutively nuclear and bound to cognate UAS (Upstream Activation Sequences), but this does not automatically lead to target gene activation. Only upon dissociation of a repressing cofactor (e.g. GAL80p from GAL4p) the activation domain is exposed and target genes are expressed [19]. Only a few studies looking at this particular aspect of transcription factor regulation have been carried out in other fungi. As for A. nidulans, the proline activator PrnA has also a permanent nuclear localisation, but needs the inducer to bind to its DNA target sites in vivo [54]. Nevertheless, importantly, forced nuclear accumulation (blocking export) is not sufficient for NirA activity and this indicates a two-step process, where the nitrate not only blocks nuclear export of NirA, but also converts nuclear NirA from an inactive to an active configuration [55]. In H. polymorpha the Yna1p-Yna2p heterodimer shows constitutive nuclear localization. This implies that the presence of nitrate elicits heterodimer DNA-binding and/or a conformational activation of the DNA-bound proteins.
Yna2p is crucial for the nuclear import or retention of Yna1p. It is possible that the NLS of Yna1p is weakly active in the absence of Yna2p. Alternatively or additionally, Yna2p may mask the Yna1p-NES ("AKLTTLLNEMDDLRLD" [55]), thus preventing its export from the nucleus (see S5 Fig for the predicted NLS of these proteins).
We identified a common motif in the promoters of the Yna1p-Yna2p target genes. This nucleotide motif does not conform to the typical CCG-n x -CCG Zn(II) 2 Cys 6 protein target site [20,56] but harbours the sequence 5 0 -CGGAGAATA-3 0 (Fig 4A), similar to the structure of A. nidulans NirA binding site (5 0 -CTCCGHGG; H = A,C,T) [12]. In conclusion, we assume that the 9-nucleotide sequence represents the H. polymorpha UAS for the nitrate gene expression and the transcriptional activators Yna1p and Yna2p bind this site preferentially as a heterodimer.
We have also shown in this study that Yna1p and Yna2p physically interact and that this interaction is crucial for bringing the heterodimer to the nucleus or for keeping it there. Our in vivo footprinting results from the YNT1 promoter region suggest, however, that this form of the transcription factor complex does not bind DNA in the absence of nitrate but needs the inducer to specifically interact with their recognition sequences in target genes. Our chromatin-immunoprecipitation studies confirmed these results by demonstrating quantitatively that in vivo both Yna1p and Yna2p only bind to the YNT1 promoter when the inducer is present. Thus, although both proteins are present in the nucleus under all conditions, they still need the inducer to bind to their target promoters. At the moment it is not clear which biochemical changes occur upon nitrate induction that transform the inactive transcription factors to an active heterodimeric complex. A mechanism how a fungal nitrate regulator can respond to the presence of the inducer was recently discovered in A. nidulans [55]. In this paper we reported that reversal of methionine sulfoxidation in the nuclear export signal of NirA is associated with the transformation process. If a similar mechanism exists for H. polymorpha, is unknown.
This work also elucidates new aspects in the regulation of the H. polymorpha nitrate metabolism. Our data show the existence of hierarchies of repressing compounds, glutamine being the strongest repressing metabolite, and proline being a neutral, or nearly-neutral, nitrogen source. It has been proposed that the existence of such hierarchies is generally related to the conversion rate at which a given nutritional nitrogen compound is transformed into the effector molecules which respond to and signal the "nitrogen state" of the cell [57].
Taken the results of this study together we propose that the "CGGAGA" DNA motif present in the nitrate-responsive gene promoters serve as upstream-activating sequence (UAS) that mediate nitrate induction by serving as binding site for a functional Yna1p/Yna2p heterodimer. Formation of the heterodimer itself does not need the inducer, but stable binding of the heterodimeric activator complex to target DNA and transcriptional competence requires nitrate either as effector molecule or as signalling compound. However, the activator complex seems to be sensitive to glutamine repression because neither interaction nor transcriptional activity is detected under these conditions. UAS are conserved in promoters of all known genes involved in nitrate assimilation, including in the promoter of YNA1. This latter sequence arrangement results in a nitrate-responsive positive auto-regulation loop at the level of transcription factor gene expression and a balanced stoichiometry in the Yna1p/Yna2p heterodimer in which Yna2p is expressed to high levels constitutively and Yna1p is induced to high levels when the inducer and thus the potential substrate for the assimilation process becomes available. Although both proteins can bind individually, the balanced stoichiometry seems to be important for optimal function of the activator complex. To ensure optimal binding, binding sites of the regulators in the investigated target gene promoters appear to lie preferentially in a nucleosome-free region that presumably allows efficient DNA recognition without nucleosomal barriers. Ultimately these binding events lead to a drastic chromatin rearrangement, followed by the transcriptional activation process. The latter is likely mediated through the recruitment of remodelling complexes, presumably containing histone acetyl-transferases and/ or ATP-dependent remodelling factors along with the pre-initiation complex.  Fig. Yna1 null mutant. A. Schematic representation of the YNA1 locus and adjacent regions and of the LD-I cassette used to generate the yna1 I6632 ::LEU2 allele (EV, EcoRV; S, SacI; EI, EcoRI). Probe-Y1, used for Southern analyses is also indicated (panel C). B. Growth tests show that L1-yna1 I6632 is unable to used nitrate as sole nitrogen source for growth. L1+p37 (Leu + ) is the wt strain with a LEU2 containing vector (EBOX37p6); yna1Δ is the L1-yna1 I6632 mutant; yna1Δ+pA1 is the same mutant with a YNA1 containing plasmid; yna1Δ−pA1 is a plasmid-less derivative of the previous strain. C. Southern blot showing the yna1 I6632 null allele. Genomic DNA, digested with EcoRV, was hybridized with probe-Y1 (panel A). Lane 1, wt (L1); lane 2, null mutant (L1-yna1 I6632 ) (TIF) S3 Fig. Yna2 null mutant. A. Schematic representation of the YNA2 locus and adjacent regions and of the strategy used to generate the yna2 Δ12274-12595 :: LEU2 allele (P, PstI; EV, EcoRV; C, ClaI). Probe-Y2, used for Southern analyses is also indicated (panel C). B. Growth tests showi that L1-yna2 Δ12274-12595 is unable to use nitrate as sole nitrogen source for growth. L1+p42 (Leu+) is the wild type with a LEU2 containing vector (EBOX42p1); yna2Δ is the L1-yna2 Δ12274-12595 mutant; yna2Δ+pA2 is the same mutant with a YNA2 containing plasmid; yna2Δ−pA2 is a plasmid-less derivative of the previous strain.