Genetic and Proteomic Analyses of a Proteasome-Activating Nucleotidase A Mutant of the Haloarchaeon Haloferax volcanii

The halophilic archaeon Haloferax volcanii encodes two relatedproteasome-activating nucleotidase proteins, PanA and PanB,with PanA levels predominant during all phases of growth. Inthis study, an isogenic panA mutant strain of H. volcanii wasgenerated. The growth rate and cell yield of this mutant strainwere lower than those of its parent and plasmid-complementedderivatives. In addition, a consistent and discernible 2.1-foldincrease in the number of phosphorylated proteins was detectedwhen the panA gene was disrupted, based on phosphospecific fluorescentstaining of proteins separated by 2-dimensional gel electrophoresis.Subsequent enrichment of phosphoproteins by immobilized metalion and metal oxide affinity chromatography (in parallel andsequentially) followed by tandem mass spectrometry was employedto identify key differences in the proteomes of these strainsas well as to add to the restricted numbers of known phosphoproteinswithin the Archaea. In total, 625 proteins (approximately 15%of the deduced proteome) and 9 phosphosites were identifiedby these approaches, and 31% (195) of the proteins were identifiedby multiple phosphoanalytical methods. In agreement with thephosphostaining results, the number of identified proteins thatwere reproducibly exclusive or notably more abundant in onestrain was nearly twofold greater for the panA mutant than forthe parental strain. Enriched proteins exclusive to or moreabundant in the panA mutant (versus the wild type) includedcell division (FtsZ, Cdc48), dihydroxyacetone kinase-linkedphosphoenolpyruvate phosphotransferase system (EI, DhaK), andoxidoreductase homologs. Differences in transcriptional regulationand signal transduction proteins were also observed, includingthose differences (e.g., OsmC and BolA) which suggest that proteasomedeficiency caused an up-regulation of stress responses (e.g.,OsmC versus BolA). Consistent with this, components of the Fe-Scluster assembly, protein-folding, DNA binding and repair, oxidativeand osmotic stress, phosphorus assimilation, and polyphosphatesynthesis systems were enriched and identified as unique tothe panA mutant. The cumulative proteomic data not only furtheredour understanding of the archaeal proteasome system but alsofacilitated the assembly of the first subproteome map of H.volcanii.

INTRODUCTION

Top

INTRODUCTION

Phosphorylation is one of the most important and widespreadposttranslational modifications of proteins. Its usefulnesscomes in the form of its strong perturbing forces, which modulatestructure and function in a profoundly effective manner (27).Phosphorylation and dephosphorylation of proteins can occurquite rapidly, making this mechanism ideal for controlling adaptiveresponses to environmental cues and changing intracellular conditions.These appealing characteristics have made phosphorylation themodification of choice for regulating a number of vital cellularprocesses, many of which overlap between the branches of life.It is estimated that in eukaryotes, as much as 30% of the proteomeis phosphorylated, with serine constituting 90% of the modifiedresidues and threonine and tyrosine representing the other 10%(12, 55). Many of these modified proteins represent immune cellsignaling pathways, circadian regulatory systems, cell divisioncontrol components, and metabolic networks within plant and/ormammalian systems (19, 42, 54). In bacteria, phosphorylationof "two-component" signal transduction and phosphotransferasesystem (PTS) elements broadens the occurrence of protein phosphorylationto include histidine and aspartic acid residues (17, 55). Althoughless common, serine, threonine, and tyrosine phosphorylationis also emerging as a regulatory device of bacteria (17).

To date, there are only a modest number of proteins with confirmedphosphorylation sites among archaea (11, 27). Several of thesemirror eukaryotic and/or bacterial counterparts in terms ofmodification status and function. Based on traditional radiolabelingexperiments, such proteins as Halobacterium salinarum CheA andCheY are phosphorylated at sites identical to those of theirtwo-component signal transduction bacterial homologs (45). Thearchaeal translation initiation factor 2 ${alpha}$ of Pyrococcus horikoshiiis phosphorylated at Ser48 by a kinase (PH0512) related to thehuman double-stranded RNA-dependent protein kinase (51). Inboth Crenarchaeota and Euryarchaeota, the initiator proteinCdc6 (Orc) is autophosphorylated by a DNA-regulated mechanismat a conserved Ser residue (10, 16). Likewise, the β subunitof 20S proteasomes from Haloferax volcanii is phosphorylatedat a Ser residue (21). A zinc-dependent aminopeptidase, witha leucine zipper motif that associates with a CCT-TRiC familychaperonin, has also been shown to be phosphorylated in Sulfolobussolfataricus (8).

The advent of modern high-throughput phosphoenrichment and analyticaltools provides the opportunity now to identify archaeal phosphoproteinson a large scale, including those that lack similarity to membersof the bacterial or eukaryotic clades or for which modificationshave not been predicted previously. This study, in which a comparativeproteomic analysis of wild-type H. volcanii cells and thoselacking proteasomal function through deletion of a primary proteasome-activatingnucleotidase (PanA) was performed, exemplifies such a scenario.The analysis included a combination of high-throughput phosphoenrichmentmethods, subenrichment strategies, and tandem mass spectrometry(MS-MS).

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Chemicals and reagents. Immobilized metal ion affinity chromatography (IMAC) and metaloxide affinity chromatography (MOAC) materials were suppliedby Qiagen (Valencia, CA) and Perkin-Elmer (Waltham, MA), respectively.Trizol reagent and Pro-Q Diamond phosphoprotein stain were purchasedfrom Invitrogen (Carlsbad, CA). Mevinolin was from Merck (WhitehouseStation, NJ). Two-dimensional gel electrophoresis (2-DE) materialswere supplied by Bio-Rad (Hercules, CA). Reversed-phase C₁₈spin columns and trypsin were purchased from The Nest Group,Inc. (Southborough, MA) and Promega (Madison, WI), respectively.Restriction enzymes and other DNA-modifying enzymes were fromNew England BioLabs (Ipswich, MA). Hi-Lo DNA and Precision Plusprotein molecular mass standards were from Minnesota Molecular,Inc. (Minneapolis, MN) and Bio-Rad, respectively. All otherchemicals and reagents were supplied by Sigma-Aldrich (St. Louis,MO) and Fisher Scientific (Pittsburgh, PA).

Strains, media, growth conditions, and plasmids. Escherichia coli strains DH5 ${alpha}$ and GM2163 (New England Biolabs)were, respectively, used for routine cloning and purificationof plasmid DNA for transformation of H. volcanii according tothe methods of Cline et al. (7). E. coli strains were grownin Luria broth supplemented with 100 mg ampicillin, 50 mg kanamycin,and/or 30 mg of chloramphenicol per liter as needed. For geneticanalysis, H. volcanii strains were grown at 42°C in YPCmedium as described in reference 1. H. volcanii medium was supplementedwith 4 to 5 mg mevinolin per ml and/or 0.1 mg novobiocin perliter as needed. H. volcanii DS70 (56) served as the parentstrain for the generation of the panA mutant strain GG102. Thechromosomal copy of GG102 panA has a 187-bp deletion and aninsertion of a modified Haloarcula hispanica hmgA gene (hmgA*)encoding 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase,which renders H. volcanii cells resistant to mevinolin (56).The following approach was used to generate GG102. Vent DNApolymerase was used for PCR amplification of the panA gene fromH. volcanii genomic DNA using primer 1 (5'-CATATGATGACCGATACTGTGGAC-3'and primer 2 (5'-GAATTCAAAACGAAATCGAAG GAC-3') (NdeI and EcoRIsites are boldfaced). H. volcanii genomic DNA was prepared forPCR from colonies of cells freshly grown on YPC plates. In brief,cells were transferred into 30 µl distilled H₂O usinga toothpick, boiled (10 min), chilled on ice (10 min), and centrifuged(for 10 min at 14,000 x g). The supernatant (10 µl) wasused as the template for PCR. The 1.28-kb PCR fragment was clonedinto pCR-BluntII-TOPO (Invitrogen) to generate plasmid pJAM636.The fidelity of this insert was confirmed by DNA sequencingat the University of Florida Interdisciplinary Center for BiotechnologyResearch. The 1.5-kb NotI fragment of plasmid pMDS99, whichcarries hmgA* (56), was inserted into the BbvCI-to-NruI sitesof panA carried on pJAM636 by blunt-end ligation to generateplasmid pJAM906. This suicide plasmid (pJAM906, which carrieshmgA* and panA in the same orientation, as determined by restrictionmapping using KpnI and AgeI) was transformed into H. volcaniiDS70, and recombinants were selected on YPC solid medium supplementedwith mevinolin. Isolated colonies were restreaked for isolationon fresh YPC solid medium plus mevinolin and were screened byPCR using primers that annealed outside the panA region of suicideplasmid pJAM906 (primer 3 [5'-TACGATAAGGACTCGGCGTCGCAGC-3']and primer 4 [5'-TACGTC GCGTTCGCGGCGTAGTCAC-3']). Clones thatgenerated the appropriate PCR product were further screenedby Western blotting using anti-PanA, anti-PanB, anti- ${alpha}$ 1, andanti- ${alpha}$ 2 antibodies as described previously (44). For complementationstudies, H. volcanii-E. coli shuttle plasmids carrying panAand panB genes were transformed into mutant strain GG102. Theseplasmids, pJAM650 and pJAM1012, encode C-terminal polyhistidine-taggedPanA and PanB (PanA-His₆ and PanB-His₆), respectively, underthe control of the Halobacterium cutirubrum rRNA P2 promoterand were constructed as follows. The 1.2-kb NdeI-HindIII fragmentof pJAM642 and the 1.24-kb NdeI-XhoI fragment of pJAM1006 (44)were separately blunt-end ligated into the NdeI-BlpI site ofpJAM202 (24) to replace psmB-his6 with panA-his6 (pJAM650) andpanB-his6 (pJAM1012), respectively.

For growth rate and proteomic analyses, log-phase H. volcaniicells were grown in 25 ml (125-ml foil-capped Erlenmeyer flasks,200 rpm, 42°C) of ATCC 974 medium and minimal medium (1),respectively. For proteomic analysis, cells were grown to anoptical density at 600 nm (OD₆₀₀) of 1.0 to 1.2, harvested bycentrifugation (at 10,000 to 14,000 x g and 4°C), flashfrozen in liquid nitrogen, and stored at –80°C.

Protein extraction and quantification. For IMAC analysis, protein was extracted from cells accordingto a previous study (25). For all other procedures, proteinwas extracted using Trizol according to protocols describedpreviously (28). Samples were dried through vacuum centrifugationand stored at –80°C. Protein concentrations were determinedby a Bradford protein assay using bovine serum albumin as astandard according to the supplier's instructions (Bio-Rad).

2-DE. 2-DE was carried out as described previously with 11 cm immobilizedpH gradient strips and Criterion 12.5% resolving Tris-HCl gels(Bio-Rad) (28). Prior to separation, protein samples were reconstitutedin an isoelectric focusing-compatible buffer consisting of 250mM glycerol, 10 mM triethanolamine, and 5% (wt/vol) 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate(CHAPS). Proteins separated by 2-DE were stained in the gelwith Pro-Q Diamond phosphoprotein stain according to the supplier'sinstructions (Invitrogen). Gels were imaged with the MolecularImager FX scanner with a 532-nm excitation laser and a 555-nmLP emissions filter (Bio-Rad). Acquired images were analyzedwith PDQuest (version 7.0.1) software (Bio-Rad).

Protein reduction, alkylation, and tryptic digestion. Dried protein samples (300 µg) were resuspended in 100µl of 50 mM NH₄HCO₃ (pH 7.5). Samples were reduced bythe addition of 5 µl of 200 mM dithiothreitol (DTT solution)(for 1 h at room temperature or 21°C). Samples were alkylatedby the addition of 4 µl of 1 M iodoacetamide (for 1 hat 21°C). Alkylation was stopped by the addition of 20 µlof DTT solution (for 1 h at 21°C). Samples were digestedwith a 1:20 ratio of trypsin to protein for 18 to 24 h at 37°C.Digested peptides were purified using 300-µl C₁₈ spincolumns and dried under vacuum centrifugation. In-gel proteinswere reduced, alkylated, and digested with trypsin by usingan automated platform for protein digestion (ProGest; GenomicsSolutions, Ann Arbor, MI).

Peptide methyl esterification. A 2 M concentration of methanolic acid was generated by adding60 µl of acetyl chloride dropwise to 300 µl of anhydrousmethanol in a glass tube. The mixture was sealed and incubated(for 10 to 15 min at 21°C). Dried tryptic peptides weremixed with 30 µl of the methanolic acid reagent and incubatedin a sealed jar containing desiccative material for 90 min at21°C. Methyl esterified samples were dried under vacuumcentrifugation and reconstituted in mobile phase B (0.1% [vol/vol]acetic acid, 0.01% [vol/vol] trifluoroacetic acid, and 95% [vol/vol]acetonitrile) for MS-MS analysis.

IMAC. Phosphoprotein enrichment by IMAC was performed using a phosphopurificationsystem according to the supplier's instructions (Qiagen) withthe following modifications. Extracted protein (2.5 mg) wasresuspended in 25 ml of the supplied lysis buffer to a finalconcentration of 0.1 mg of protein per ml. Six 500-µlfractions per sample, as opposed to the suggested four, werecollected for analysis.

Titanium dioxide phosphopeptide enrichment. MOAC of H. volcanii DS70 and GG102 peptides was performed usingthe Phos-Trap phosphopeptide enrichment kit (catalog no. PRT301001KT;Perkin-Elmer) with the following modifications. Samples wereagitated in the presence of 40 µl of elution buffer asopposed to the recommended 20 µl. Samples were incubatedwith the TiO₂ resin for 10 min instead of 5 min and were incubatedwith elution buffer for 15 min instead of 10 min. The resultingsamples were dried under vacuum centrifugation at 42°C for30 min and reconstituted in mobile phase B for MS-MS analysis.

RP-HPLC coupled with nanoelectrospray ionization-QTOF (QStar) MS-MS. Separation of peptides (desalted with a PepMap C₁₈ cartridge)by capillary reversed-phase high-performance liquid chromatography(RP-HPLC) was performed using a PepMap C₁₈ column (15 cm [length]by 75 µm [inner diameter]) and an Ultimate Capillary HPLCsystem (LC Packings, San Francisco, CA). A linear gradient of5% to 40% (vol/vol) acetonitrile for 25 min at 200 nl·min^–1was used for separation. MS-MS analysis was performed onlineusing a hybrid quadrupole time-of-flight (QTOF) instrument (QStarXL hybrid LC/MS-MS) equipped with a nanoelectrospray source(Applied Biosystems, Foster City, CA) and operated with AnalystQS (version 1.1) data acquisition software. Information-dependentacquisition was employed in which each cycle consisted of afull scan from m/z 400 to 1,500 (1 s) followed by MS-MS (3 s)of the two ions that exhibited the highest signal intensity.In the full-scan acquisition mode, ions were focused throughthe first quadrupole by focusing and declustering potentialsof 275 V and 55 V, respectively, and were guided to the TOFregion via two quadrupole filters operated in radio frequency(rf)-only mode. Ions were orthogonally extracted, acceleratedthrough the flight tube (plate, grid, and offset voltages were340, 380, and –15 V, respectively), and refocused to a4-anode microchannel plate detector via an ion mirror held at990 V. The same parameters were utilized with the MS-MS modeof operation; however, the second quadrupole was employed tofilter a specific ion of interest while the third quadrupoleoperated as a collision cell. Nitrogen was used as the collisiongas, and collision energy values were optimized automaticallyusing the rolling collision energy function based on m/z andthe charge state of the peptide ion.

LCQ ion trap MS. IMAC samples separated in the gel were analyzed with a ThermoLCQ Deca quadrupole ion trap MS (LCQ Deca MS) in line with a5-cm (length) by 75-µm (inner diameter) Pepmap C₁₈ 5-µm-particle-size(300-Å mean pore diameter) capillary column (LC Packings).The HPLC device operating upstream of the MS system was runwith a 60-min gradient from 5% to 50% mobile phase B with aflow rate of 12 µl·min^–1. MS parent ion scanswere followed by four data-dependent MS-MS scans.

MS data and protein identity analyses. Spectra from all experiments were converted to DTA files andmerged to facilitate database searching using the Mascot searchalgorithm (version 2.1; Matrix Science, Boston, MA) againstthe deduced H. volcanii proteome (assembly, 26 May 2006; http://archaea.ucsc.edu/).Search parameters included trypsin as the cleavage enzyme. Carbamidomethylationwas defined as the only fixed modification in the search, whilemethionine oxidation, pyro-Glu from glutamine or glutamic acid,acetylation, and phosphorylation of serine, threonine, and tyrosineresidues were set as variable modifications. Mass tolerancesfor all LCQ analyses were 2 Da for MS and 1 Da for MS-MS. Masstolerances for all QStar analyses were 0.3 Da for both MS andMS-MS. Protein identifications for which a probability-basedmolecular weight search (MOWSE) score average of 30 or abovewas not assigned were excluded. Proteins were considered uniqueto GG102 panA or DS70 only if protein identities were exclusivefor at least 2 samples per strain. Transmembrane spanning heliceswere predicted using TMHMM, version 2.0 (32). Phosphosites werepredicted using NetPhos, version 2.0 (5). Proteins were categorizedinto clusters of orthologous groups (COG) using COGNITOR (53).

All phosphopeptides predicted by MS via the Mascot databasesearch algorithm were filtered to include only those peptidesthat were top-ranking hits and had E values less than 3.0. Theresulting list of phosphopeptide spectra was reviewed manuallywith Analyst QS (version 1.1) acquisition software (AppliedBiosystems, Foster City, CA), with Scaffold software (version1.6; Proteome Software, Portland, OR), and with spectral datagenerated through the Mascot search algorithm. The peptide fragmentmasses detected were compared with those in a theoretical peptidefragmentation spectrum generated by Protein Prospector, version4.0.8 (P. R. Baker and K. R. Clauser; http://prospector.ucsf.edu).Mass additions of 80 Da (HPO₃) or neutral losses of 98 Da (H₃PO₄)were confirmed mathematically through analysis of peptide fragmentationand correlation of detected diagnostic ions present in the correspondingspectrum. Consideration was also given to spectral quality,including peak intensity and completeness of generated ion series,in the screening of spectral data for confirmation of modification.Theoretical lists of modified internal ions were also generated,searched, and assigned in the actual spectra.

RESULTS AND DISCUSSION

Top

RESULTS AND DISCUSSION

Construction of the H. volcanii panA mutant GG102. In addition to three 20S proteasomal proteins ( ${alpha}$ 1, ${alpha}$ 2, and β),H. volcanii encodes two proteasome-activating nucleotidase proteins(PanA and PanB), which are 60% identical. Of these, the ${alpha}$ 1, β,and PanA proteins are relatively abundant during all phasesof growth (on rich medium at moderate temperatures, 37 to 42°C)(44). In contrast, the levels of PanB and ${alpha}$ 2 are relatively lowin the early phases of growth and increase severalfold duringthe transition to stationary phase (44). To further investigatethe function of the predominant Pan protein, PanA, a mutationwas generated in the chromosomal copy of the panA gene. A suicideplasmid, pJAM906, was used to generate this mutation, in whichpanA had a 187-bp deletion in addition to an insertion of theH. hispanica mevinolin resistance marker. Recombinants wereselected on rich medium supplemented with mevinolin and screenedby colony PCR using primers which annealed outside the chromosomalregion that had been cloned onto the suicide plasmid. Approximately10% of the clones that were screened generated the expectedpanA mutant PCR product of 2.50 kb compared to the parent strainPCR product of 1.28 kb (data not shown). Of the PCR-positiveclones, one (GG102) was selected for further analysis by Westernblotting using anti-PanA antibodies. The PanA protein was readilydetected by Western blotting for the parent strain, DS70 (Fig.1). In contrast, PanA was not detected under any growth conditionsfor the panA mutant GG102, thus confirming the mutation.

View larger version (21K):
[in this window]
[in a new window]

FIG. 1. Western blot analysis of the H. volcanii panA mutant strain GG102. Cell lysates of the parent strain, DS70, and the panA mutant, GG102, were analyzed by Western blotting using anti-PanA, -PanB, - ${alpha}$ 1, and - ${alpha}$ 2 polyclonal antibodies as indicated (where ${alpha}$ 1 and ${alpha}$ 2 are the ${alpha}$ -type subunits of 20S proteasomes). Total protein (2, 4, and 8 µg) from DS70 (lanes 1, 3, and 5) and GG102 (lanes 2, 4, and 6) was separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis prior to analysis.

20S proteasome and PanB protein levels are not altered by the panA mutation. To determine whether the panA mutation influenced the levelsof 20S proteasome and/or Pan proteins, DS70 and GG102 were analyzedby Western blotting using polyclonal antibodies raised separatelyagainst the ${alpha}$ 1, ${alpha}$ 2, and PanB proteins (where ${alpha}$ 1 and ${alpha}$ 2 are subunitsof 20S proteasomes). No differences in the levels of any ofthese three proteasomal proteins were detected between thesetwo strains when they were grown to log phase in minimal medium(Fig. 1). Thus, H. volcanii does not appear to have a feedbackmechanism to increase the levels of PanB or 20S proteasomesafter a loss of PanA under the growth conditions examined inthis study.

Growth phenotype of the panA mutant GG102 compared to its parental and complemented strains. Comparison of GG102 panA to its parent, DS70, revealed significantdifferences in growth in rich medium (Fig. 2). This includedan increase in the doubling time from 4.1 to 5.2 h as well asa decrease in the overall cell yield from an OD₆₀₀ of 2.1 to1.8. Both the doubling time and the overall cell yield wereat least partially restored by complementation with plasmidpJAM650 or pJAM1020, encoding PanA-His₆ or PanB-His₆ protein,respectively. The ability of the pHV2-derived plasmid pJAM1020to partially complement the panA mutation for growth may bedue, at least in part, to the high-level expression of PanBfrom this plasmid, which includes both a strong rRNA P2 promoterof H. cutirubrum and a T7 terminator to mediate transcriptionof the panB-his6 gene fusion. This is likely to alter the levelsof PanB protein, normally low in early phases of growth andincreased severalfold during the transition to stationary phase.Although the PanA and PanB proteins may have overlapping functionsbased on this partial complementation for growth, these proteinsdiffer significantly in primary sequence (by 40%) and growthphase-dependent regulation (44) and thus are likely to degradeproteins differentially in terms of substrate preference and/orrate.

View larger version (12K):
[in this window]
[in a new window]

FIG. 2. Reduced growth phenotype of H. volcanii GG102 panA compared to its parent and complemented strains. The DS70 parent strain ( ${blacksquare}$ ), GG102 panA ( ${square}$ ), GG102 panA(pJAM650) ( ${circ}$ ), and GG102 panA(pJAM1012) (•) were inoculated (1%) from log-phase cultures into 25 ml ATCC 974 medium at 42°C (200 rpm) in 125-ml Erlenmeyer flasks and were monitored for growth (OD₆₀₀). Plasmids pJAM650 and pJAM1012 encode PanA-His₆ and PanB-His₆, respectively. Experiments were performed in biological triplicate.

Comparative 2-DE analysis of the panA mutant and its parent. Expression of the panA- and panB-his6 gene fusions using anrRNA P2 promoter on a high-copy-number plasmid is likely toalter the proteome composition of the complemented panA mutantfrom that of the "wild type." To avoid this complexity, theparental strain, DS70, was used as the "wild type" for proteomecomparison to the panA mutant GG102. Analysis of cell lysatesof DS70 and GG102 by 2-DE revealed comparable numbers of totalprotein spots as detected by the Sypro Ruby protein stain: 333(±17) and 314 (±14) spots, respectively. However,further analysis of the 2-DE-separated proteins by Pro-Q Diamond,a phosphoprotein-specific fluorescent dye developed by Steinberget al. (50), revealed more than twice as many "phosphoprotein"spots for the panA mutant (64 ± 7) as for the wild type(31 ± 6) (Fig. 3). As many as 40 to 44% of the Pro-QDiamond-stained phosphoproteins were not detected by Sypro Rubyand thus are likely to be present at relatively low levels inthe cell. Interestingly, the limited impact of the panA mutationon the total number of protein spots detected by Sypro Rubycontrasts with the >1.7-fold increase after addition of the20S proteasome-specific inhibitor clasto-lactacystin-β-lactone(cLβL) (29). This may reflect the central role of the N-terminalThr proteolytic active site of the 20S β subunit, whichis irreversibly inactivated by cLβL, in all proteasome-mediatedprotein degradation in this archaeon. In contrast, PanA andother closely related AAA+ ATPases are more likely to play upstreamroles, including differential substrate recognition, unfolding,and translocation into the 20S proteasome core for degradation.

View larger version (19K):
[in this window]
[in a new window]

FIG. 3. Total proteins of the H. volcanii parent strain, DS70 (A), and GG102, its panA mutant (B), separated by 2-DE and stained with the phosphospecific fluorescent protein stain Pro-Q Diamond. Protein molecular mass standards (Bio-Rad Low Range) and the range of the immobilized pH gradient strip used for isoelectric focusing are indicated on the left and at the top, respectively.

Phosphoprotein and phosphopeptide enrichments. IMAC and MOAC were used in four separate experiments (experimentsA to D) to enrich the cell lysates of the panA mutant (GG102)and its parent (DS70) for phosphoproteins and/or phosphopeptides.A similar 2-DE pattern of proteins was detected by Pro-Q Diamondstaining before and after IMAC enrichment (data not shown).As an initial approach to protein identification, IMAC-purifiedsamples were separated by 12% sodium dodecyl sulfate-polyacrylamidegel electrophoresis (1-dimensionally), and bands that appeareduniquely or in discernibly different amounts in the mutant wereexcised and digested in the gel with trypsin prior to hybridMS-MS analysis (LCQ Deca MS) (experiment A). This analysis produceda modest number of protein identifications (a total of 10) withMOWSE scores of 30 or above and thus was not further pursuedas a viable approach for high-throughput analysis of these strains.To overcome these limitations, biological triplicate samplesof GG102 and DS70 were subjected to three separate approaches:(i) IMAC enrichment of proteins followed by cleavage with trypsin,(ii) MOAC enrichment of methyl esterified tryptic peptides,and (iii) IMAC enrichment of proteins followed by MOAC enrichmentof methyl esterified tryptic peptides (experiments B to D, respectively).Peptides from all three of the latter approaches were separatedby C₁₈ RP-HPLC and analyzed online using electrospray ionization-QTOFhybrid MS.

In total, these approaches resulted in the identification of625 proteins with MOWSE scores of 30 or above and with at least2 peptides identified per protein (see Table S1A in the supplementalmaterial; detailed proteomic data are available upon request).This corresponds to 15.4% of the deduced proteome of H. volcanii,based on the DS2 genome sequence minus proteins encoded by pHV2(the plasmid cured from DS70 and its derivatives). The proteinsidentified ranged from 4.9 to 186 kDa, with an average molecularmass of 41.6 kDa. The majority of the proteins identified (523proteins, or 83% of the total) were encoded on the 2.848-Mbchromosome. The coding sequences for the remaining 202 proteinswere distributed among pHv4 (52 proteins), pHv3 (43 proteins),and pHv1 (7 proteins). This represents 8 to 17% of the codingcapacity for each element (18% of the 2.848-Mb chromosome, 8.2%of pHv4, 11.2% of pHv3, and 7.9% of pHv1). Of the total proteinsidentified, 145 were based on high-probability scores and multiplepeptide identifications within a single biological replicateand thus were not assigned as unique or common to DS70 or GG102.The remaining proteins were categorized as either common toGG102 and DS70 (328 proteins), unique to DS70 (57 proteins),or unique to GG102 (98 proteins) (see Table S1A in the supplementalmaterial). The number of proteins identified by MS-MS in thesephosphoprotein-enriched fractions that were exclusive to GG102was nearly twofold higher than the number of identified proteinsexclusive to DS70. This relative difference in MS-MS-identifiedstrain-specific proteins is consistent with the differencesobserved in the number of Pro-Q Diamond-stained 2-DE gel spotsfrom the proteomes of these two strains.

The 480 proteins that were categorized as either common to bothstrains or exclusive to one strain (DS70 or GG102 panA) wereidentified based on an average detection per protein of 4.7tryptic peptides in 5 biological replicates and an average MOWSEscore of 76. Among the proteins common to both strains, 8 to9% were estimated to be more abundant in either GG102 (15 proteins)or DS70 (13 proteins) based on spectral counts. Of the total625 proteins identified, only a small portion (32 proteins,or 5.1%) were predicted to form transmembrane spanning helices(TMH), compared to the approximately 22% putative TMH proteinsencoded on the genome (see Table S1A in the supplemental material).The TMH proteins were detected primarily in GG102 (more than80%), with nearly 50% classified as unique to this strain. Alower proportion of TMH proteins was detected in DS70 (53%)with only 2 of these (6%) grouped as exclusive to this strain.While the significantly higher number of identified predictedmembrane proteins in the panA deletion strain than in the wildtype is intriguing, the overall number identified is low, andthey are likely to require preparation of membranes to enhancetheir detection (30).

Summaries of proteins identified as either unique to or moreabundant in either GG102 or DS70 are given in Tables 1 and 2.At this stage in our understanding, it is not clear whetherthese differences in proteomes between the two strains reflecta change in the phosphorylation status of the protein or inoverall protein abundance. However, the identification of thesedifferences was reproducible. As a confirmation of the approaches,the PanA protein was exclusively and reproducibly detected inthe DS70 parent strain across all proteomic analyses. The PanAprotein was not detected in GG102, consistent with the targetedknockout of the encoding panA gene from the chromosome of thisstrain (Fig. 1).

View this table:
[in this window]
[in a new window]

TABLE 1. Proteins identified as unique to either the panA mutant GG102 or the parent strain, DS70^a

View this table:
[in this window]
[in a new window]

TABLE 2. Proteins identified as common but with higher spectral counts, suggesting greater abundance, for either the panA mutant GG102 or DS70

Categorization of "unique" proteins into COG. The 156 proteins identified as exclusive to either GG102 orDS70 (listed in Table 1) were categorized into COG as indicatedin Fig. 4 (for details, see Table S1B and C in the supplementalmaterial). Overall, the largest percentage (36%) of these proteinswas equally distributed between two groups: (ii) those of unknownor general function (R/S) and (ii) those involved in amino acidtransport and metabolism (E). The remaining proteins clusteredinto a diversity of functional COG groups ranging from translationto signal transduction as outlined in Fig. 4. Comparison ofthe panA mutant to its parent (GG102 versus DS70) revealed notabledifferences in the distribution of "unique" proteins withinmany of these groups. In particular, the percentages of proteinsexclusive to GG102 panA were considerably higher than thoseunique to DS70 in energy production and conversion (C), posttranslationalmodification, protein turnover, and chaperones (O), cell envelopebiogenesis and outer membrane (M), and inorganic ion transportand metabolism (P). In contrast, the percentages of proteinsthat clustered to transcription (K), translation, ribosomalstructure, and biogenesis (J), and nucleotide transport andmetabolism (F) were higher in DS70 than in GG102. The increasednumber of proteins involved in translation, nucleotide biosynthesis,and transcription for DS70 may be a reflection of the highergrowth rate and overall cell yield of this strain compared tothose of GG102 panA. Although the increased number of GG102-uniqueproteins clustering to energy production and conversion (C)would seem to contradict this suggestion, the vast majorityof the latter proteins were oxidoreductases (10 out of 11),including a homolog of "Old Yellow Enzyme," amounts of whichhave been shown to increase during periods of oxidative stressin Bacillus subtilis (14). The remaining GG102-unique proteinof this cluster was a glycerophosphoryl diester phosphodiesterase(ugpQ), an enzyme that can be an indicator of phosphate starvation(2). In contrast to GG102, the two DS70-unique proteins thatclustered to this general COG group (C) were predicted to beinvolved in the generation of proton motive force, includingan ATP synthase subunit and plastocyanin-/cytochrome-like protein.

View larger version (17K):
[in this window]
[in a new window]

FIG. 4. Proteins identified by MS-MS as exclusive to either the panA mutant (GG102) or parent (DS70) grouped according to the COG database. "Members of COG" represents the number of exclusive proteins which group to a particular COG as a percentage of total exclusive proteins identified by MS-MS for GG102 (solid bars), DS70 (open bars), and the sum of GG102 and DS70 (striped bars). COG groups listed include E (amino acid transport and metabolism), R/S (general function prediction only/function unknown), C (energy production and conversion), G (carbohydrate transport and metabolism), K (transcription), J (translation, ribosomal structure, and biogenesis), O (posttranslational modification, protein turnover, chaperones), L (DNA replication, recombination, and repair), F (nucleotide transport and metabolism), M (cell envelope biogenesis, outer membrane), P (inorganic ion transport and metabolism), N/U (cell motility and secretion/intracellular trafficking), T (signal transduction mechanisms), V (defense mechanisms), D/B (cell division and chromosome partitioning), I (lipid metabolism), and H (coenzyme metabolism).

One of the most notable differences between the protein profilesof the panA mutant and its parent was linked to phosphorus assimilationand synthesis of polyphosphate (the PHO regulon, many membersof which cluster to group P) (summarized in Fig. 5). Proteinsof the PHO regulon identified exclusively in GG102 includedhomologs of phosphate/phosphonate ABC-type transporters (pstB2,pstS2, and phnC), PHO regulators (abrB_phoU2), polyphosphatekinase (ppk), ABC transporter glycerol-3-phosphate binding protein(ugpB), and glycerophosphoryl diester phosphodiesterase (ugpQ)(discussed above). Many additional proteins unique to GG102were encoded within these genomic regions (i.e., hisH, HVO_A0476,trxB, metK, and cysK). In addition, a transcriptional regulator(HVO_0730) unique to DS70 was found to be divergently transcribedfrom a gene encoding inorganic pyrophosphatase (ppa). Componentsof the ABC-type Pst2 transporter uniquely identified in GG102are encoded on pHV4 and appear ancillary to a paralogous ABC-typePst1 transporter common to both DS70 and GG102, which is encodedon the 2.848-Mb chromosome. The up-regulation of componentsof phosphate uptake and metabolism typically corresponds tocellular stress (13, 52), with the levels of polyphosphate impactingcell survival (6). Proteomic and microarray analyses of archaealspecies such as Methanosarcina acetivorans have revealed thePHO regulon to be expressed at higher levels during growth onthe lower-energy-yielding substrate acetate than on methanol(38).

View larger version (28K):
[in this window]
[in a new window]

FIG. 5. Organization of H. volcanii genes encoding proteins linked with phosphorus transport and metabolism that are altered by the panA mutation. Genes encoding relevant paralogs are also included. ORFs are numbered according to the genome assembly available at http://archaea.ucsc.edu/. ORFs encoding proteins unique to GG102 (solid arrows), unique to DS70 (shaded arrows), common to DS70 and GG102 (hatched arrows), and not detected in this study (open arrows) are indicated. Proteins include leucyl-tRNA synthetase (leuS); heat shock proteins (hsp); prephenate dehydratase (pheA); the glutamine amidotransferase subunit of imidazole glycerol phosphate synthase (hisH); phosphate binding protein (hbp); phosphonate/phosphate ABC transporter ATPase (phnC) and permease (phnE); regulatory proteins of phosphate assimilation and metabolism (phoU1 and phoU3); an protein homolog ambiactive regulator of transcription during transition between log and stationary phase fused to PhoU (abrB_phoU2); phosphate ABC transporter ATP-binding (pstB1 and pstB2), permease (pstA1, pstA2, pstC1, and pstC2), and solute-binding (pstS1 and pstS2) components; a conserved protein (HVO_0476, which is linked to all haloarchaeal pst operons identified to date); superoxide dismutase (SOD)transmembrane helix (TMH); thioredoxin reductase (trxB); inositol monophosphatase/NADPH phosphatase (suhB); glycerophosphoryl diester phosphodiesterase (ugpQ); glycerol-3-phosphate ABC transporter solute binding (ugpB), permease (ugpA and ugpE), and ATP-binding (ugpC) components; cysteine synthase (cysK); thiamine biosynthesis protein (thiI); polyphosphate kinases (ppk1 and ppk2); S-adenosylmethionine synthase (metK); adenylate cyclase (cyaA); cytochrome oxidase (cox1); zinc/cadmium cation efflux system (ZnCad efflux); histidine triad DNA-binding protein (His triad); methionine aminopeptidase type II (map); serine/threonine protein phosphatase (prpA); and inorganic pyrophosphatase (ppa). Proteins encoded by genes abbreviated as abrB_phoU2 and thiI_tRNA modification may be multifunctional. Bar, 1,000 bp of genomic DNA.

In addition to alterations in the PHO regulon, the panA knockoutresulted in increased enrichment and identification of proteinslinked to protein folding, Fe-S cluster assembly, and the oxidativestress response compared to those for the parent strain, DS70.These included components of a thiosulfate sulfurtransferase(tssA, tssB) and Fe-S assembly system (nifU, sufB, sufC), thioredoxin/disulfidereductases (trxB1 and trxB2), topoisomerase A (topA), replicationA-related single-stranded DNA binding protein (RPA), excinucleaseABC ATPase subunit (uvrA), Old Yellow Enzyme (yqjM; discussedabove), Hsp20 molecular chaperone (ibpA), OsmC-like regulator(osmC), peptidyl-prolyl cis-trans-isomerase (slyD), S-adenosylmethioninesynthase, and cysteine biosynthetic enzymes (e.g., dehydrogenase,O-acetylhomoserine sulfhydrylase) (Tables 1 and 2). Interestingly,in E. coli the osmotically inducible OsmC is proposed to usehighly reactive cysteine thiol groups to elicit hydroperoxidereduction (35) and is downstream of a complex phosphorelay systemthat includes regulation by the AAA+ proteases Lon and HslUV(33, 41).

A number of additional differences in transcription and signaltransduction proteins were also observed between DS70 and GG102panA. For example, subunits of RNA polymerase were altered bythe panA mutation, with Rpb4 more abundant in DS70 fractionsand RpoA and RpoB ( ${alpha}$ and β) more abundant in those of GG102(Table 2). The altered levels of Rpb4 in DS70 versus GG102 areconsistent with the recent finding that the levels of this proteinare directly correlated with eukaryotic cell growth (i.e., cellsexpressing lower levels of Rpb4 grow more slowly than cellsexpressing higher levels) (48). In addition to the differencesnoted above, a BolA-like protein and conditioned medium-inducedprotein 2 (Cmi2) were identified exclusively in DS70. BolA triggersthe formation of osmotically stable round cells when overexpressedin stationary-phase E. coli (46), and Cmi2, a putative metal-regulatedtranscriptional repressor, may be regulated by quorum sensingin H. volcanii (G. Bitan-Banin and M. Mevarech, GenBank accessionno. AAL35835). A number of putative DNA-binding proteins andhomologs of methyl-accepting chemotaxis (MCP) proteins werealso found to be altered. In addition, a number of mandelateracemase/muconate lactonizing enzymes related to the E. colistarvation-sensing protein RspA (20) were altered by the panAmutation.

Some archaea do not encode Pan proteins (e.g., species of Thermoplasma,Pyrobaculum, and Cenarchaeum). Thus, it is speculated that Cdc48/VCP/p97AAA+ ATPases, which are universal to archaea, may function with20S proteasomes as in eukaryotes (22). Interestingly, two Cdc48-likeAAA+ ATPases were identified as either exclusive or more abundantin GG102 panA than in DS70 (HVO_1327 and HVO_2700 [Tables 1and 2, respectively]), suggesting that these proteins may compensatein part for the loss of the AAA ATPase function of PanA. A TMHhomolog of prohibitin (HVO_0035) known to regulate AAA proteasefunction (3) was also identified as exclusive to GG102.

Intriguing similarities also emerged between the proteomic profilesof this study and our previous work in which cells were treatedwith the 20S proteasome-specific inhibitor cLβL and analyzedby 2-DE (29). Specifically, three proteins that accumulatedmore than fivefold in cells treated with cLβL were alsodiscovered in notably higher abundance in the panA mutant strainthan in the parent strain of this study. This subset of proteinsincluded the DJ-1/ThiJ family protease, the translation elongationfactor EF-1 ${alpha}$ , and the Fe-S cluster assembly ATPase SufC (Table2). In addition, FtsZ cell division protein homologs and componentsof the putative phosphoenolpyruvate PTS which appear to be coupledto dihydroxyacetone kinase (PtsI, DhaK, DhaL) were also foundat higher levels in GG102 panA and cLβL-treated cells thanin the wild type.

Phosphopeptide identification. Based on favorable likelihood scores using the Mascot searchalgorithm and careful manual analysis of the peptide spectra,nine individual phosphosites, including four serine, four threonine,and one tyrosine modification site, were tentatively identifiedand mapped to a total of eight proteins (Table 3; see also Fig.S1A to H in the supplemental material). Five of these phosphositeswere identified exclusively by the combined IMAC-MOAC approach(experiment D). The remaining four sites were identified bythe combined IMAC-MOAC process as part of an overlapping dataset with at least one other phosphoanalysis method within thisstudy (Table 3). Although neural networks are not yet availablefor archaeal protein phosphosite predictions based on the limiteddata set available for this domain, many of the putative phosphositesidentified in this study had NetPhos scores above the thresholdof 0.5, which is based on eukaryotic phosphosites.

View this table:
[in this window]
[in a new window]

TABLE 3. Putative phosphosites identified by MS-MS^a

Among the H. volcanii phosphoproteins identified, the Cdc48-relatedAAA+ ATPase (HVO_2700), which was more abundant in GG102 enrichedfractions than in DS70 based on spectral counting, was representedby two identical phosphorylated peptides occurring in separateexperiments for both strains (Table 3). Both inclusive and exclusivey ions and internal ion fragments containing a phosphorylatedThr45 residue were identified (numbering throughout is basedon the deduced polypeptide sequence [see Fig. S1 in the supplementalmaterial]). Similarly, a protein of unknown function (HVO_A0206)was also represented by duplicate peptides phosphorylated atSer88 in multiple experiments (Table 3). Although the functionof this ORF is unknown, it is transcribed in the same orientationas two genes encoding members of the clustered regularly interspacedshort palindromic repeats (CRISPR) family of proteins, whichwere identified by MS-MS as common to GG102 and DS70. A DNA-directedRNA polymerase subunit A' (RpoA') was determined to have a phosphomodificationat Tyr117 (Table 3). The discovery of the Tyr117 phosphosite,which had the highest NetPhos score (0.962) and the lowest expectvalue (1.4 e-4) of the phosphosites identified, was supportedby the appearance of three separate internal fragment ions containingthe modified tyrosine residue as well as an inclusive and exclusivey-ion series (see Fig. S1 in the supplemental material). Otherphosphoproteins included a Cdc6-1/Orc1-1 ortholog (HVO_0001)and pyruvate kinase (Table 3). Although both proteins were commonto DS70 and GG102 based on MS-MS identification, the phosphopeptidesof these proteins were detected only in GG102 panA. The reasonfor the exclusive appearance of the latter phosphopeptides inthe panA mutant strain remains to be determined; however, itdoes implicate the proteins corresponding to these phosphopeptidesas potential proteasomal substrates that depend on their phosphorylationstatus for destruction. Phosphosites were also identified oncomponents of membrane systems including an ATPase of an ABC-typetransporter as well as an integral membrane protein with a TrkA(NAD-binding) domain (Table 3). A hypothetical protein (HVO_C0059)was also identified as a putative phosphoprotein based on thedetection of one doubly modified peptide fragment at residuesSer136 and Thr139 (Table 3).

Many of the phosphorylated proteins identified and listed inTable 3 are supported by evidence of the phosphorylation andproteasome-mediated destruction of their orthologs. For example,eukaryal RNA polymerases II and III are phosphorylated at tyrosine,serine, and/or threonine residues (23, 49), and ubiquitin E3ligase Wwp2 targets the Rpb1 subunit of RNA polymerase II (relatedto H. volcanii RpoA') for destruction by 26S proteasomes (37).Eukaryal and archaeal Cdc6 proteins undergo autophosphorylationat serine residues as a presumed means of regulating their DNAhelicase loading activity (10, 16). Furthermore, the ubiquitin/proteasome-mediateddestruction of Cdc6 in both mammalian and yeast cells has beenobserved and is proposed to be a mechanism for uncoupling DNAreplication and cell division as part of a preprogrammed celldeath response (4). Cdc48/p97/VCP proteins undergo Akt-mediatedphosphorylation at multiple sites as a mode of ubiquitinatedprotein substrate release to facilitate degradation by proteasomes(31). In addition, these AAA+ ATPase proteins are phosphorylatedby DNA-protein kinase at serine residues as part of a proposedmechanism to enable chaperone activity and to aid in DNA repair(40).

Conclusions. Based on the results of this study, a global increase in thenumber of phosphoproteins appears to occur in a panA mutantcompared to wild-type cells. Although several of the proteinsenriched and identified as exclusive to GG102 panA are likelyinvolved in phosphorus assimilation and metabolism and thusmay bind phosphate noncovalently, this alone cannot accountfor the large-scale difference in phosphoprotein staining betweenthe panA mutant and its parent. Therefore, we propose that thephosphorylation of protein substrates may facilitate their recognitionfor proteasome-mediated destruction in organisms for which traditionalubiquitination pathways are absent (e.g., archaea, actinomycetes).It is also possible that the observed changes in phosphoproteincontent reflect a secondary effect of the maintained presenceof a kinase or enhanced digestion of a phosphatase.

It has been demonstrated in eukaryotes that protein phosphorylationcan serve as a precursor to ubiquitin tagging and subsequentdegradation by 26S proteasomes (26, 39). Furthermore, phosphorylationof conserved sequences rich in proline, glutamic and asparticacids, serine, and threonine (PEST sequences) is a well-knownexample of posttranslational modification as a destabilizingforce on substrate proteins (15, 43). The link between proteinphosphorylation and proteasomal degradation is also supportedthrough the function of accessory structures such as the COP9signalosome, which has been implicated as a coordinator of proteinkinases and ubiquitin ligases in plant systems (18).

The predominant proteins identified as exclusive to the panAmutant GG102 (versus DS70) were linked to protein folding, Fe-Scluster assembly, oxidative stress response, and phosphorusassimilation and polyphosphate synthesis. A number of additionaldifferences in transcription and signal transduction proteins(e.g., OsmC, BolA, Cmi2) were also observed between these twostrains, and these, together with the growth defect, suggestthat the panA mutant is undergoing stress and accumulating polyphosphate.Consistent with this, a distant relative of PanA, the ATPasering-forming complex of mycobacteria, is presumed to associatewith 20S proteasomes and serve as a defense against oxidativeor nitrosative stress (9). Proteasomal inhibition has also beenshown to hypersensitize differentiated neuroblastoma cells tooxidative damage (36). Interestingly, a polyphosphate-Lon proteasecomplex is proposed in the adaptation of E. coli to amino acidstarvation (34). Whether archaeal proteasomes are linked topolyphosphate remains to be determined; however, short-chainpolyphosphates identical to those of Saccharomyces cerevisiaehave been detected in H. volcanii cells grown under conditionsof amino acid starvation (47).

ACKNOWLEDGMENTS

Special thanks to S. Stevens, Jr., and S. McClung of the ProteomicsCore (Interdisciplinary Center of Biotechnology Research, Universityof Florida, Gainesville) for assistance with MS-MS protein analysis.Thanks also to A. Hartman, J. Eisen, and The Institute for GenomicResearch (TIGR) for communication of the H. volcanii DS2 genomesequence. Thanks also to M. L. Dyall-Smith for supplying H.volcanii DS70 and plasmid pMDS99.

This research was funded in part by a grant from the NationalInstitutes of Health (R01 GM057498) and Department of Energy(DE-FG02-05ER15650).

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology and Cell Science, University of Florida, Gainesville, FL 32611-0700. Phone: (352) 392-4095. Fax: (352) 392-5922. E-mail: jmaupin{at}ufl.edu

Published ahead of print on 26 October 2007.

Supplemental material for this article may be found at http://jb.asm.org/.

REFERENCES

Top