Carbon Source-dependent Assembly of the Snf1p Kinase Complex in

The Snf1p/AMP-activated kinases are involved in transcriptional, metabolic, and developmental regulation in response to stress. In Saccharomyces cerevisiae , Snf1p (Cat1p) is one of the key regulators of carbohy-drate metabolism, and cat1 ( snf1 ) mutants fail to grow with non-fermentable carbon sources. In Candida albicans , Snf1p is an essential protein and cells depend on a functional Snf1 kinase even with glucose as carbon source. We investigated the Ca Snf1p complex after tandem affinity purification and mass spectrometric analysis and show that the complex composition changes with the carbon source provided. Three subunits were identified, one of which was named Ca Snf4p because of its homology to the Sc Snf4 protein and the respective Ca SNF4 gene could complement a S. cerevisiae snf4 mutant. The other two proteins revealed similarities to the S. cerevisiae kinase (cid:1) subunits Sc Gal83p, Sc Sip2p, and Sc Sip1p. Both genes complemented the scaffold function in a S. cerevisiae gal83,sip1,sip2 triple deletion mutant and were named according to their scaffold a of Kis2p activator Kis2

Candida albicans is a widely distributed commensal fungus that is carried as a part of the human microbial flora. However, it is also an opportunistic pathogen that can cause serious infections, particularly in immunocompromised individuals (1). In C. albicans, Snf1p (CaSnf1p, NCBI accession number 46437276) seems to be essential for growth; moreover, CaSnf1p has the ability to complement the snf1 mutant of Saccharomyces cerevisiae (2,3). In Candida tropicalis, a close relationship of CtSnf1p (86% identity to CaSnf1p) mRNA levels and cell growth was demonstrated (4) that may relate CtSnf1p to more fundamental cellular processes, but to date no direct interaction partners of the essential CaSnf1p have been identified.
In the model yeast S. cerevisiae, the ScSnf1p kinase is a homologue to the highly conserved AMP-activated serine/threonine kinases that are found in plants, Drosophila, Caenorhabditis elegans, mammals, and fungi (for review see Ref. 5)). These kinases seem to be essential components of cascades that function as metabolic sensors in eukaryotic cells and are activated under conditions of nutrient stress. Additionally, they were reported to be involved in pathogenesis and the treatment of several human diseases, including type 2 diabetes, obesity, heart disease, and cancer (6).
The importance of ScSNF1 (ScCAT1) was first identified genetically after the isolation of cat1 (snf1) mutants. This gene was identified as a key element for the regulation of glucose repression because of the failure of cat1 mutants to grow with non-fermentable carbon sources and to derepress glucose-repressible genes (7). The mutation of the regulatory subunit cat3 had a similar phenotype (8) and was later also identified as snf4 (9). 1 ScSnf1p assists cells to adapt to glucose-limited conditions, regulating the transcription of metabolic genes and the activity of metabolic enzymes (11,12), meiosis and sporulation (13,14), pseudohyphal and invasive growth (15)(16)(17), and life span and aging (18,19). The ScSnf1p kinase activity correlates with increased concentrations of AMP, but the enzyme does not exhibit allosteric regulation by AMP in vitro (20). ScSnf1p down-regulates ATP-consuming enzymes such as acetyl-CoA carboxylase (21) and phosphorylates transcription factors like ScMig1p, causing its translocation from the nucleus to the cytoplasm and derepressed gene expression (22).
In S. cerevisiae, the ScSnf1p kinase complex contains a catalytic ␣ subunit encoded by ScSNF1 (23) comprising 633 amino acids with a catalytic N-terminal serine/threonine kinase domain. The first half of the regulatory domain in the remaining part down-regulates the kinase activity under nutrient-rich conditions (24), whereas the second half is needed for the interaction with the ␤ subunits (25). The ␥ subunit encoded by ScSNF4 plays a key regulatory role here (26,27). It associates with the ␤ subunit by a consecutive interaction, and it binds to the autoinhibitory domain of ScSnf1p (as previously mentioned), releasing its catalytic domain. Under glucose-limited conditions, phosphorylation of Thr 210 in the ScSnf1p C-terminal loop promotes kinase activity (28). The upstream activating Snf1p kinases in S. cerevisiae were recently identified as ScPak1p, ScTos3p, and ScElm1p (29). The ␤ subunits encoded either by ScSip1p, ScSip2p, or ScGal83p mediate heterotrimer * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. formation between ScSnf1p and ScSnf4p, and hence three distinct forms exist in vivo (30). The ␤ subunits are also responsible for the interaction with downstream targets and regulate the subcellular localization of the kinase (31,32). The C terminus contains an internal kinase association (KIS) 2 domain associated with the ␣ subunit and a conserved 80-residue ASC (association with the Snf1p kinase complex) domain that is associated with the ␥ subunit (25).
The transmission of a wide range of biological signals depends upon direct physical interaction of specific cellular proteins (33).
To study these interactions we used the tandem affinity purification (TAP) method (34) to purify the Snf1p kinase complex from the human pathogen C. albicans. A strain expressing only the TAP-tagged fusion protein was chosen, which allowed the test for functionality to be close to physiological concentrations by forming a complex with endogenous components. Because the deletion of CaSnf1p is lethal, it was interesting whether the TAP tag would interfere with the physiological function. After deletion of the second copy, the cells showed no phenotypical change on either glucose or ethanol. The combination of two different affinity tags greatly reduced the binding of nonspecific proteins to the complex. We could identify the ScSnf4 homologue protein from C. albicans and two scaffold subunits, which we named CaKis1p and CaKis2p. The protein interactions within the Snf1p kinase complex were also confirmed using yeast two-hybrid analysis. In addition, a transcriptional activator domain in CaSip2p could be identified.

MATERIALS AND METHODS
Unless otherwise noted, all chemicals were of analytical grade. Plasmid Constructs and Genetic Manipulation-Established protocols for molecular biology were followed (35). Transformation of yeast strains utilized the lithium acetate (36) procedure. The CaURA3 open reading frame including 421 nucleotides of upstream 5Ј-sequence and 121 nucleotides of downstream 3Ј-sequence was amplified by PCR using primers CaURA3-ApaI/CaURA3-PstI and SC5314 genomic DNA as a template. The 1361-bp PCR product was digested with ApaI/PstI and subsequently ligated into pBS1479 (34) opened with the same enzymes. In the resulting plasmid, pPK335, the ScTRP1 gene was replaced by the CaURA3 gene.
All genetic modifications of the C. albicans genome were performed via homologous recombination using PCR products with 60-bp short flanking homologies to the CaSNF1 gene (37). The CaURA3-TAP tag cassette was amplified by PCR using primers (supplemental Table II) homologous to both the CaURA3-TAP cassette and CaSNF1 (CaSNF1-TAP-S1/CaSNF1-TAP-S2) and pPK335 as a template. After transformation of C. albicans RM1000 using the CaURA3-TAP tag cassette, correct integration in uridine prototrophic colonies was verified by diagnostic PCR. In the resulting strain, CPK140 -6, the second CaSNF1 allele was deleted using CaHIS1 as a selection marker. The CaHIS1 gene deletion cassette was amplified using primers homologous to both the CaHIS1 gene and CaSNF1 (CaSNF1-S3/CaSNF1-S4) using pCS-CaHIS1 (pRS426ϩCaHIS1) 3 as a template. The resulting histidine prototrophic strain CPK141-28 was verified by diagnostic PCR.
Southern Blot-The integration of the TAP cassette and the deletion of the coding region of CaSNF1 were verified by Southern blot analysis on genomic DNA. Genomic DNA was prepared as described (38).
Tandem Affinity Purification-TAP-tagged Snf1p kinase was purified from C. albicans according to Ref. 39. Cells were grown to A 600 ϭ 10 absorption units in yeast extract medium containing either 4% (w/v) glucose or 3% (w/v) ethanol and collected by centrifugation, washed with water, frozen in liquid nitrogen, and stored at Ϫ80°C. Cells were lysed in Buffer A (10 mM HEPES-KOH, pH 7.9 (4°C), 10 mM KCl, 1.5 mM MgCl 2 , 0.5 mM dithiothreitol, 2 mM benzamidin, 1 M leupeptin, 2 M pepstatin, 2.6 M aprotinin) (three ϫ cell volume) using glass beads (one cell volume) at 4°C. Prior to centrifugation, KCl was adjusted to 0.2 M and centrifuged for 90 min at 100,000 ϫ g. The cytosolic phase was dialyzed against Buffer D (20 mM HEPES-KOH, pH 7.9 (4°C), 50 mM KCl, 0.2 mM EDTA, pH 8, 0.5 mM dithiothreitol, 20% glycerol, 0.5 mM phenylmethylsulfonyl fluoride, 2 mM benzamidin) (3 h, 4°C), frozen in liquid nitrogen, and stored at Ϫ80°C. Binding and elution steps were performed in Poly-Prep columns (Bio-Rad). 200 l of IgG-bead suspension (Sigma) was washed with 10 ml of IPP150 buffer (10 mM Tris-HCl, pH 8 (4°C), 150 mM NaCl, 1 mM imidazole, 2 mM CaCl 2 , 0.1% Nonidet (NP-40)), elution buffer contains also 2 mM EGTA, and the composition of the extract was adjusted to 10 mM Tris-HCl (pH 8), 100 mM NaCl, and 0.1%(v/v) Nonidet P-40. Extract and beads were incubated (2 h, 4°C), drained by gravity flow, and washed with 30 ml of IPP150 buffer and 10 ml of tobacco etch virus (TEV) cleavage buffer. Recovery of complexes was achieved after incubation with 100 units of TEV protease (Invitrogen) in 1 ml of TEV cleavage buffer of the beads at room temperature for 2 h and elution by gravity flow. 200 l of TEV cleavage buffer was used to purge the dead volume of the column. 3.6 ml of IPP150 calmodulin binding buffer and 3.6 l of 1 M CaCl 2 were added to 1.2 ml of eluate. 200 l of calmodulin binding beads were washed with 10 ml of IPP150 calmodulin binding buffer. Eluate and beads were incubated (1 h, 4°C), drained by gravity flow, and washed with 30 ml of IPP150 calmodulin binding buffer. Recovery of complexes was achieved by elution with 1 ml of IPP150 elution buffer. For all further analyses, proteins were precipitated with an equal volume of 30%(w/v) trichloroacetic acid and resuspended in 20 l of SDS loading buffer in preparation for gel electrophoresis.
Western Blotting-Western blotting was performed as described (40). The peroxidase-anti-peroxidase antibody was directed against the protein A domain (Sigma).
All gels were entirely sliced and subjected to in-gel digestion protocols as described (43,44), which were adapted for use on a Microlab® Star digestion robot (Hamilton, Bonaduz, Switzerland). After 12 h the supernatant was removed, and the remaining peptides were extracted three times with 50% (v/v) ACN/5% FA. All fractions were pooled and dried in a vacuum centrifuge prior to analysis. For MALDI mass spectrometric analysis the samples were solved in 5 l of 50% ACN/1% (v/v) trifluoroacetic acid (Fluka, Buchs, Switzerland). 0.5 l of the sample was mixed with 0.5 l of matrix (2 mg/ml ␣-cyano-4-hydroxycinnamic acid (Bruker, Bremen, Germany) in 50% ACN/0.5% (v/v) trifluoroacetic acid) directly on a stainless steel MALDI target (Applied Biosystems (ABI), Darmstadt, Germany) and dried under ambient conditions. MALDI-TOF MS-Delayed extraction TM (DE) MALDI time-of-flight (TOF) mass spectra were recorded on a Voyager-DE STR instrument (ABI) using a nitrogen laser ( ϭ 336 nm, repetition rate ϭ 20 Hz) for desorption and ionization with an acquisition mass range from 600 to 5000 m/z and the low mass gate set to 550 m/z. The total acceleration voltage was 20 kV with 68.5% grid voltage on the first grid, 0.02% guide wire voltage, 150 ns delay, and a mirror voltage ratio of 1.12. Spectra were externally calibrated with a Sequazyme TM peptide mass standards kit (ABI). Between 1000 and 2000 laser shots were accumulated for each mass spectrum. All spectra were smoothed, noise-filtered, and deisotoped using the Data Explorer (Version 4.3; ABI). Deisotoped peaks were labeled by the software, and the 100 most intense peaks were used for data base searching. Autolytic tryptic peptides or peptides resulting from the identified protein were used for internal calibration (45).
Protein Data Base Queries-Proteins were identified using Spectrum Mill (Version 3.0; Agilent Technologies, Waldbronn, Germany) installed on a local server using the current NCBI data base. For peptide mass fingerprint (PMF) data, the 100 most intense peaks were submitted to Spectrum Mill using a search tolerance of 25 ppm. For nano-LC MS/MS data the mass tolerance for the precursor ion was 1.5 Da and the fragment ion mass tolerance was 0.8 Da.
iTRAQ TM Labeling-The final eluate of each sample after TAP was precipitated, washed with ice-cold acetone, and resuspended in 40 l of 0.5 M triethylammonium bicarbonate, 0.1% (w/v) SDS. The proteins were reduced with 5 mM Tris(2-carboxyethyl)phosphine and alkylated with 10 mM methyl methane-thiosulfonate followed by overnight digestion using trypsin. The dried sample was resuspended in 100 l of labeling buffer (0.25 M triethylammonium bicarbonate, 75% (v/v) ethanol). The corresponding label was added and incubated (1 h at room temperature). Residual reagents were quenched by adding 300 l of water and incubated another 30 min. Prior to strong cation exchange 2 The abbreviations used are: KIS, kinase association; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight; ASC, association with the Snf1p kinase complex; TAP, tandem affinity purification; TEV, tobacco etch virus; ACN, acetonitrile; FA, formic acid; PMF, peptide mass fingerprint; NCBI, National Center for Biotechnology Information; MS, mass spectrometry; LC, liquid chromatography. 3 Peter Koetter, personal communication.
chromatography the sample was dried.
Strong Cation Exchange Chromatography-The combined mixture was separated by strong cation exchange chromatography on an HP 1100 HPLC system (Hewlett-Packard, Waldbronn, Germany) using a PolySulfoethyl A column (4.6 ϫ 100 mm, 5 m, 300 Å). Sample was dissolved in 1 ml of strong cation exchange loading buffer (25% (v/v) ACN, 10 mM KH 2 PO 4 adjusted to pH 3 with phosphoric acid), loaded, and washed for 20 min at 1 ml/min. Peptides were eluted in a two-step gradient (0 -500 mM KCl (25% (v/v) ACN, 10 mM KH 2 PO 4 adjusted to pH 3 with phosphoric acid)) over 15 min at a flow rate of 1 ml/min with fractions collected from 20 min onward. Yeast Two-hybrid Analysis-Cells were harvested and washed once with KPP (potassium-phosphate) buffer (pH 6.5, 4°C). After washing, cells were resuspended in 500 l of KPP buffer, glass beads were added, and the mixture was vortexed for 90 s at 4°C. 1 ml of cold KPP was added, and cell debris were separated by centrifugation (4000 rpm, 10 min). Protein concentration was determined using a microbiuret assay. ␤-Galactosidase was assayed according to Ref. 46.

Construction of C. albicans SNF1-tagged
Strain-For the analysis of the CaSnf1p kinase complex assembly, a TAP tag cassette was chromosomally fused to the C terminus of CaSnf1p (Fig. 1A). To show the functionality of the CaSNF1-TAP construct and to avoid any influence of CaSnf1p, the remaining wild-type gene CaSNF1 was additionally deleted, resulting in strain CPK141-28 (supplemental Table I , Fig. 1B). Correct integration of the TAP cassette and the deletion of the second allele were shown using Southern hybridization ( Fig.  2A). Strain CPK141-28 showed wild-type growth on glucose and on ethanol, which proved that the recombinant tagged CaSNF1 gene, transcribed from its native promoter, was fully functional (data not shown). Western blot analysis following SDS-PAGE separation (using the peroxidase-anti-peroxidase antibody directed against the protein A part of the TAP tag) (Fig. 2B) showed an apparent molecular mass of the fusion protein of 90 kDa. The observed value fits well with the calculated value of CaSnf1p (69 kDa) fused to the calmodulin binding peptide (5 kDa) and protein A (16 kDa) separated by a TEV cleavage site.
Tandem Affinity Purification of the CaSnf1p Kinase Complex-To identify the interaction partners of CaSnf1p we performed tandem affinity purifications of the TAP-tagged protein from cells grown on different carbon sources. Following TAP, the protein complexes were separated via SDS-PAGE, and the corresponding band patterns were compared (Fig. 3A). Subsequently, all gels were entirely sliced and subjected to an in-gel digestion protocol. Proteins were identified via PMF and nano-LC MS/MS. When cells were grown on glucose, the complex displayed three intense bands and one faint band (Fig. 3A, lane 1), whereas cells grown on ethanol displayed three intense bands (Fig. 3A, lane 2). In both experiments we could identify the most intense band (band 2) as the bait (CaSnf1p) fused to the calmodulin-binding protein (Figs. 3B and 4A). The protein had an apparent molecular mass of 74 kDa, which fits well to the calculated value of CaSnf1p (69 kDa) fused to the calmodulin binding peptide (5 kDa).
Using both PMF and nano-LC MS/MS analysis, all four bands could be unambiguously assigned to C. albicans proteins. Band 4 could be assigned to hypothetical protein CaO19.13191 (NCBI accession number 46439783). The migration behavior of CaO19.13191 during SDS-PAGE fits to the calculated molecular mass of 41 kDa (Fig. 3A, band 4). After PMF, 45% of the amino acid sequence could be covered, and subsequent LC MS/MS analyses enhanced the sequence coverage to 56% with well assigned peptide sequences. As a representative example, the MS/MS spectrum of a very hydrophobic peptide (1672.99 Da), which elutes at 56 min (42% (v/v) ACN), is shown (Fig. 4B). From the nearly complete y-and b-ion fragmentation series of the parent ion, the peptide sequence LINVYEAVDILALVK could be determined with a Spectrum Mill score Ͼ15 and a scored peak intensity Ͼ70%, (scores Ͼ15 lie within the category of highest confi-dence). The identified hypothetical protein CaO19.13191 shows 64% identity to the ␥ subunit Snf4p from S. cerevisiae and is now named CaSnf4p (Fig. 5).
Protein band 3 ( Fig. 3A) with an apparent molecular mass of 50 kDa could be identified as CaO19.4084 (NCBI accession number 46435808) with a calculated molecular mass of 46 kDa. Using PMF, 24% of the amino acid sequence was covered, which was enhanced to 29% using high quality MS/MS data (Fig. 4C). This protein shows striking similarities to the S. cerevisiae ␤ subunits ScGal83p and ScSip2p, having 49 and 45% homology, respectively (Fig. 5C). With this degree of similarity it was not possible to clearly attribute CaO19.4084 to one of the S. cerevisiae proteins. Furthermore, CaO19.4084 also showed homologies to the ASC and KIS domains of the third S. cerevisiae ␤ subunit, ScSip1p, being 27% identical and 43% similar in the respective regions.
Protein band 1 (Fig. 3A) migrating at 100 kDa was identified as CaO19.12464 (NCBI accession number 46444282) with a calculated molecular mass of 81 kDa. Using PMF, 33% of the amino acid sequence was covered, which was increased to 38% when LC MS/MS fragmentation data of two additional peptides were considered. To substantiate this result the MS/MS spectrum of the peptide LLYIINNEYR is shown (Fig. 4D). Using homology prediction, the protein shows similarities to the ASC domains of the S. cerevisiae ␤ subunits, ScSip2p having 42% identity and 65% similarity, ScGal83p with 64% identity and 80% similarity, and ScSip1p with 45% identity and 78% similarity (Fig. 5D). In comparison, CaO19.12464 possesses only a partial KIS domain, having 33% identity to ScSip2p, 36% iden- tity to ScGal83p, and 40% identity to ScSip1p within the conserved region. CaO19.12464 co-purified from ethanol-grown cells was not visible using SDS-PAGE but could be identified using PMF. We were able to match two peptides (one with one missed cleavage site) within 25 ppm mass accuracy to a myristoylated N-terminal tryptic peptide of CaO19.12464 (Fig. 6). The leader sequence already indicates a high probability for N-terminal myristoylation, which was also shown for the S. cerevisiae ␤ subunits ScSip2p (47)   gal83,sip1,sip2; the genes encoding all scaffold proteins were deleted. This mutant cannot grow with ethanol as the sole carbon source. Independent transformation of CaKIS1 and CaKIS2 recovered the ability of the triple deletion mutant to grow on ethanol (data not shown). This clearly confirmed that CaKis1p and CaKis2p have scaffold functionality.
Additionally, the physical interactions of the identified C. albicans proteins were measured with yeast two-hybrid analysis (see Table I). CaKis1p interacted with CaSnf4p in both directions of the assay, whereas an interaction with CaSnf1p was only found when CaKis1p was fused to the binding domain. Interestingly, the interaction of CaSnf4p with the scaffold proteins, in comparison to CaSnf1p, was severalfold increased on both carbon sources. This analysis confirmed the interactions of CaSnf1p with CaKis1p and CaKis2p, respectively. As already reported for S. cerevisiae (49), the interaction of CaSnf4p and CaSnf1p was only weak after growth on ethanol but increased ϳ5-fold under glucose repression conditions. Surprisingly CaKis2p fused to a DNA-binding domain itself shows strong activity in the yeast two-hybrid assay (pPK343, Table I). This suggests strong transcriptional activation without the presence of an activation domain, which was so far not reported for any of the ␤ subunits from S. cerevisiae.
Carbon Source-dependent Assembly of C. albicans Scaffold Subunits Using iTRAQ TM Quantification-Because ScSnf1 is involved in cell signaling upon starvation, we followed the complex composition when cells were grown on different carbon sources. Using SDS-PAGE, we analyzed the banding pattern. The ratios of intensities changed, with a high reproducibility, depending on the carbon source provided. Grown on glucose, four bands became visible after Coomassie Blue staining, of which CaSnf1p was the most intense. Grown on ethanol, only three bands were visible, all with similar intensities. The amount of CaKis2p incorporated in the complex decreased when cells were grown on ethanol. To study the varying protein abundance, we used iTRAQ TM (ABI, Sciex), a multiplexed peptide quantification method. This amine-specific labeling reagent enabled us to per- Yeast two-hybrid analysis of intact C. albicans proteins. Values show an average ␤ galactosidase activity in Miller units (nmol ϫ min ϫ mg Ϫ1 ) for two transformants each measured at least twice. S.E. were less than 10%. The values obtained were found from cells grown in liquid SC -trp -leu medium containing either 4% glucose (upper case) or 3% ethanol (lower values). form multiplexed, relative peptide quantification (50). After purification of the complex (from cells grown on different carbon sources) and in-solution digestion, each sample was labeled using the iTRAQ TM labeling reagents. The samples were combined and then fractionated via strong cation exchange. The resulting fractions were desalted and separated using reversed phase chromatography, and the labeled peptides were detected online using the QTrap 2000 hybrid tandem mass spectrometer. For each scan of the mass spectrometer the three highest signals were automatically chosen for fragmentation, and the corresponding fragment ions displayed isotope-specific signals. The area below the signals was used to quantify the relative abundance of peptides derived from a protein in the sample (Table II). The iTRAQ TM analysis clearly proved that the amount of CaKis2p was 3-fold decreased on ethanol when compared with glucose. The other subunits of the CaSnf1 kinase complex did not respond to the carbon source provided. This result suggests that the CaKis2p/Snf1p kinase complex could be of major importance for the regulation of carbohydrate metabolism in C. albicans.

DISCUSSION
Because of its specificity and sensitivity, tandem affinity purification combined with mass spectrometry analysis provides an excellent technique for studying protein complexes (34). Here we describe an approach to genomically tag C. albicans proteins at their C terminus, so that the fusion construct is still expressed from its native promoter, whereas the other allele was deleted. To avoid any artificial changes in subunit composition the unchanged expression rate of the respective bait is of major importance.
Here we investigated the CaSnf1 protein complex, which functions as a metabolic sensor upon metabolic changes in various yeasts. In S. cerevisiae three distinct forms of the Snf1 kinase complex exist in vivo (30). The ScSnf1 kinase is associated with the ␥ subunit ScSnf4p and one of the three ␤ subunits, ScGal83p, ScSip1p, and ScSip2p.
As C. albicans is a diploid organism, we deleted the remaining wild-type allele to exclude its possible influence on the complex composition (Fig. 1). Cells showed a normal growth behavior after the genomic integration of a TAP cassette at the C terminus of the CaSnf1 protein and after deletion of the remaining wildtype CaSNF1 gene. This clearly proved the functionality of the CaSnf1-TAP fusion protein. Using our adapted tandem affinity purification protocol for C. albicans, we were able to show the interaction of CaSnf1p with its predicted ␥ subunit, CaSnf4p (Fig. 3). Transposon mutagenesis in a large scale loss-of-function genetic screening identified CaSnf4p as being involved in filamentous growth (51). Furthermore, we identified two ␤ subunits of the CaSnf1 kinase complex. Both proteins showed homologies to the three known S. cerevisiae ␤ subunits, Gal83p, Sip1p, and Sip2p. Because of their homology they could not be clearly assigned to any of the S. cerevisiae proteins and were named CaKis1p and CaKis2p. CaKis1p was homologous to all three S. cerevisiae ␤ subunits. The primary structure of the CaKis2 protein is different from all previously described ␤ subunits of the Snf1 kinase complexes. Although in all currently known ␤ subunits the KIS and ASC domains are located adjacent to each other, these two domains are separated in CaKis2p by a spacer region of ϳ200 amino acid residues. Furthermore, the CaKis2p ASC domain is also interrupted by a spacer region of 73 amino acid residues (Fig. 5).
The migration behavior of CaKis2p in a denaturing gel does not fit to the calculated molecular mass, suggesting possible posttranslational modification of the protein (Fig. 3). From the PMF data we were able to match, with a tolerance better than 25 ppm, two peptides (one with a missed cleavage site) that correspond to a fatty acid attached to the glycine residue following the removal of the N-terminal methionine (Fig. 6). These data support our hypothesis that CaKis2p is N-terminal myristoylated. The CaKis2 protein is the first Snf1 ␤ subunit for which the myristoylation has been shown biochemically. Such myristoylations were also suggested for the S. cerevisiae scaffold proteins Sip1 (48) and Sip2 (47), and the myristoyl moiety seems to direct ScSip2p to the plasma membrane (18) and ScSip1p to the vacuole (48). In this case, myristoylation may function as an anchor to retain the kinase complex in the cytoplasm.
The newly identified C. albicans Kis1p and Kis2p were tested for functionality as ␤ subunits in S. cerevisiae. After transformation of the respective genes into a gal83,sip1,sip2 triple deletion mutant, both genes could complement the growth failure of this mutant on ethanol. This clearly shows that CaKis1p and CaKis2p both function as scaffold proteins in the ScSnf1 kinase complex although the primary structure of CaKis2p is significantly different as compared with the other known ␤ subunits.
Interaction studies on different carbon sources using the yeast two-hybrid system showed that the CaSnf1p interacts with the two ␤ subunits CaKis1p and CaKis2p (Table I). We show that CaSnf1p weakly interacts with CaSnf4p in the presence of ethanol as it is described for S. cerevisiae (49). Surprisingly, CaSip2p fused to the DNA-binding domain demonstrated a very strong activity. We could map the activation domain to the middle region of the protein between amino acid residues 265-584 using N-and C-terminal deletions (data not shown). To date, no such transcriptional activation has been shown for any other known Snf1 ␤ subunits, and its possible physiological relevance needs further investigation.
Because ScSnf1p is one key regulator when cells are starved for glucose, we purified the complex from cells grown in the presence of ethanol as a non-fermentable carbon source. Using the iTRAQ TM reagents, we were able to monitor relative protein abundance in the complex purified from cells grown on different carbon sources. The ␤ subunit composition changed upon glucose starvation with a 3-fold decrease of CaKis2p on ethanol. In S. cerevisiae, the Snf1 kinase complex is localized in the cytosol in the presence of glucose. In contrast, under derepressing conditions such as on ethanol, the Snf1 kinase is needed in the nucleus. Because CaKis2p is myristoylated, it may function as an anchor and therefore retain the complex in the cytosol as described for S. cerevisiae (18,48).
The C. albicans Snf1 complex differs from that of S. cerevisiae in many respects. Snf1 is an essential protein in C. albicans, but not in S. cerevisiae. The scaffold proteins, especially CaKis2p, differ in their primary sequences. In addition, the C. albicans Snf1 composition is less complex when compared with the Snf1 complex of S. cerevisiae where six associated proteins could be identified. 4 Therefore, the investigation of the C. albicans Snf1 kinase complex is of particular interest for Relative protein quantification using iTRAQ TM amd nano-LC MS/MS analysis. Counts represent the number of MS/MS experiments used to quantify the ratios. Changes in protein abundance are described as a n-fold change with respect to cells grown on glucose. S.D. is the standard deviation calculated from corresponding MS/MS experiments. understanding its essential role in the regulation of cellular metabolism. The TAP method presented here is applicable to virtually any protein in C. albicans and provides an efficient way to focus research on metabolic pathogenicity factors in the protein networks.