Posttranslational Modification of the 20S Proteasomal Proteins of the Archaeon Haloferax volcanii

Abstract

20S proteasomes are large, multicatalytic proteases that play an important role in intracellular protein degradation. The barrel-like architecture of 20S proteasomes, formed by the stacking of four heptameric protein rings, is highly conserved from archaea to eukaryotes. The outer two rings are composed of α-type subunits, and the inner two rings are composed of β-type subunits. The halophilic archaeon Haloferax volcanii synthesizes two different α-type proteins, α1 and α2, and one β-type protein that assemble into at least two 20S proteasome subtypes. In this study, we demonstrate that all three of these 20S proteasomal proteins (α1, α2, and β) are modified either post- or cotranslationally. Using electrospray ionization quadrupole time-of-flight mass spectrometry, a phosphorylation site of the β subunit was identified at Ser129 of the deduced protein sequence. In addition, α1 and α2 contained N-terminal acetyl groups. These findings represent the first evidence of acetylation and phosphorylation of archaeal proteasomes and are one of the limited examples of post- and/or cotranslational modification of proteins in this unusual group of organisms.

26S proteasomes are central proteolytic enzymes of the ubiquitin-mediated degradation pathway in eukaryotes. The 20S core particle of 26S proteasomes is responsible for the hydrolysis of peptide bonds, and the overall architecture of this complex is conserved from archaea to eukaryotes (12). The 20S proteasomes are chambered proteases with the proteolytic N-terminal Thr active site sequestered within a cylindrical complex composed of 28 proteins associated as four heptameric rings of α- and β-type subunits (2). The α-type subunits form the outer two rings, and the β-type subunits form the inner two rings that comprise the central proteolytic chamber.

In the halophilic archaeon Haloferax volcanii, there are two α-type (α1 and α2) subunits and one β-type proteasomal subunit (54). The N-terminal protein sequence of the β subunit purified from 20S proteasomes reveals an N-terminal Thr residue exposed via posttranslational cleavage which is likely to form the active site (54). In contrast, the α subunits are not amenable to N-terminal sequencing, suggesting that these proteins are modified at their N termini (54).

Two subtypes of 20S proteasome complexes have been identified in stationary-phase cells and are denoted α1β- and α1βα2-20S, with the latter containing both α-type subunits in a single 20S proteasome (20). Interestingly, both α1 and α2 proteins form homo-oligomeric rings and α1-α1 and α2-α2 contacts (versus α1-α2 contacts) are detected in 20S proteasome complexes (20). Thus, the α1βα2-20S core particle is proposed to be an asymmetric proteasome, with the stacked rings arranged in an α1:β:β:α2 configuration. The α1β-20S proteasome, in contrast, is in a symmetrical α1:β:β:α1 configuration. The levels of α2 protein increase severalfold, while the levels of α1 and β proteins remain relatively constant as cells enter stationary phase (41). This is likely to allow for the regulated control of different 20S proteasome subtypes in the cell.

Previous studies have shown that different post- and cotranslational modifications take place on the subunits of proteasomes and their regulatory particles in eukaryotic cells (33, 42). These modifications, which include phosphorylation, N-terminal acetylation, autocatalytic cleavage of β propeptides, and N myristoylation, are suggested to control a variety of functions. Although several of these events have been demonstrated, few phenotypes that are associated with these proteasomal modifications have been analyzed. Suggested biological functions include the steric hindrance of the 20S proteasome aperture, the association of regulatory particles with the 20S core, and the subcellular distribution of proteasomes (33, 42). In this study, we identified N-terminal acetylation and phosphorylation sites on 20S proteasome proteins of H. volcanii. This is the first evidence of post- and/or cotranslational modification of archaeal proteasomal proteins and will facilitate our understanding of how these types of modifications regulate fundamental cellular processes.

MATERIALS AND METHODS

Materials.

Biochemicals were purchased from Sigma-Aldrich (St. Louis, MO). Other organic and inorganic analytical-grade chemicals were from Fisher Scientific (Atlanta, GA) and Bio-Rad (Hercules, CA).

Strains and medium.

The strains used in this study, H. volcanii DS70 and DS70(pJAM204), were previously described (20, 52) and are derived from the parent strain DS2. These strains were grown in ATCC 974 medium (42°C, 200 rpm) supplemented with 0.1 mg of novobiocin per liter as needed.

Enrichment of 20S proteasomes.

One liter of H. volcanii DS70(pJAM204), expressing a gene encoding a C-terminal hexahistidine variant of the α1 subunit (α1-His), was grown to early stationary phase (an optical density at 600 nm of 1.5). The parent strain DS70 was included as a negative control. Cells were lysed by French press (1,000 lb/in²) in 20 mM Tris buffer, pH 7.5, and 2 M NaCl (buffer A) supplemented with 5 mM imidazole and phosphatase inhibitor cocktail (Sigma-Aldrich). Lysates were clarified by centrifugation (10,000 × g at 4°C for 15 min) and passed through a 0.45-μm syringe filter (Nalgene, Rochester, NY). Filtered protein samples were applied to a 5-ml nickel-Sepharose Fast Flow column (HisTrap HP; Amersham, Piscataway, NJ) preequilibrated in buffer A with 5 mM imidazole. The nickel-charged column was washed sequentially in buffer A with 5 mM imidazole and 60 mM imidazole. Protein samples were eluted in buffer A with 500 mM imidazole in 2-ml fractions. These fractions were assayed for 20S proteasomal proteins by reducing 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and peptidase activity by using the peptide substrate N-Suc-LLVY-Amc (N-succinyl-Leu-Leu-Val-Tyr-7-amido-4-methylcoumarin or AMC) as previously described (54). Protein concentration was determined by a Bradford assay with bovine serum albumin as a standard (Bio-Rad).

Two-dimensional gel electrophoresis.

Proteins were separated by two-dimensional gel electrophoresis (2-DE) as previously described (24), with the following modifications. Fractions containing 20S proteasomes were precipitated in 75% (vol/vol) acetone (10 min at 4°C) and washed three times in 100% cold acetone. Protein pellets were resuspended in an isoelectric focusing-compatible buffer composed of 250 mM glycerol, 10 mM triethanolamine, and 4% (wt/vol) CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}. Samples were boiled prior to dilution in rehydration buffer (7 M urea, 2 M thiourea, 4% [wt/vol] CHAPS, 2 mM tributylphosphine, 0.2% [vol/vol] Bio-Lyte 3/5 and 3/10 ampholyte mixture [2:1], and 0.001% [wt/vol] bromophenol blue). Protein (25 to 45 μg) was applied to isoelectric focusing strips (11 cm, pI 3.9 to 5.1) in 185 μl of rehydration buffer and incubated overnight at room temperature (16 h, 25°C). Samples were focused at 250 V for 15 min, 8,000 V for 60 min with ramping, and 8,000 V for 50,000 V · h, with all steps at 20°C. A second dimension was performed by separating proteins via reducing 12% SDS-PAGE at 200 V (75 min, 16°C). Proteins were stained in the gel with SYPRO Ruby and visualized by a molecular imager FX (Bio-Rad).

Mass spectrometry.

Fractions (3 to 6 μg protein) from nickel affinity chromatography were directly separated by reducing 12% SDS-PAGE. Gels were stained with Bio-Safe Coomassie (Bio-Rad). Protein bands corresponding to the molecular weights of the subunits of 20S proteasomes were excised from the gel and destained with 100 mM NH₄HCO₃ in 50% (vol/vol) acetonitrile (4°C, overnight). Protein samples were reduced, alkylated in-gel, and digested either with trypsin (Promega), sequentially with trypsin and chymotrypsin (Sigma-Aldrich), or with Glu-C endoprotease (Staphylococcus V8 protease; Roche) (50). Capillary reverse-phase high-performance liquid chromatography (HPLC) separation of the protein digests was performed using a PepMap C18 column (75-μm inside diameter, 15-cm length; LC Packings, San Francisco, CA) in combination with an Ultimate capillary HPLC system (LC Packings, San Francisco, CA) operated at a flow rate of 200 nl per min. A gradient (90 or 120 min) from 5 to 50% acetonitrile in 0.1% acetic acid was used. Tandem mass spectrometry (MS/MS) analysis was performed using a hybrid quadrupole time-of-flight (QTOF) instrument (QSTAR) equipped with a nanoelectrospray source (Applied Biosystems, Foster City, CA) and operated with Analyst QS 1.1 data acquisition software. The synthetic phosphopeptide RGEDMSMQALpSTL (where pS represents phosphoserine) used for comparison to a phosphopeptide of the β protein was generated using fluorenylmethoxycarbonyl amino acid chemistry on an ABI 432A peptide synthesizer.

Database searching.

Tandem mass spectrometry data generated by information-dependent acquisition via QSTAR were searched against the deduced proteome of the H. volcanii DS2 genome sequence (communicated by J. Eisen, TIGR) by using the Mascot (Matrix Science, Boston, MA) database search engine (37). Probability-based Mascot scores were determined by a comparison of search results against estimated random match population and are reported as ∼10 × log₁₀(p), where p is the absolute probability. Individual Mascot ion scores greater than 32 were considered to indicate identity or extensive homology (P < 0.05). Scores above this default significant value were considered for protein identification in addition to validation by manual interpretation of MS/MS spectra. The 26-amino-acid N-terminal peptide fragment of α2 generated by a Glu-C endoprotease digest, which had a Mascot ion score of 24, was also included in this analysis. Carbamidomethylation of cysteine was used as a fixed modification based on sample preparation. Variable modifications that were implemented in the database search included deamidation of asparagine and glutamine, N-terminal acetylation, oxidation (single and double) of methionine, and phosphorylation of serine, threonine, and tyrosine.

RESULTS

Isoforms of 20S proteasome subunits.

Many subunits of 26S and 20S proteasomes from eukaryotes have isoforms that are generated by posttranslational modification and can be separated by 2-DE. In order to investigate this in archaea, an H. volcanii strain that synthesizes a C-terminal histidine-tagged α1 protein (α1-His) of 20S proteasomes (20) was used to enrich for proteasomal complexes by nickel affinity chromatography. Based on our previous work (20), samples enriched for proteasomes using this method and strain are proteolytically active and are composed primarily of 20S proteasomes and heptamers of α-type subunits. Less-abundant assemblies of α1 and α2 with molecular masses smaller than heptameric rings are also observed. Therefore, this general approach was used to enrich for proteasomes from log-phase cells for further analysis. Samples were separated by 2-DE by using an immobilized pH gradient from 3.9 to 5.1 prior to reducing 12% SDS-PAGE (Fig. 1). Protein spots which migrated in molecular mass in a manner similar to those of previously described α1, α2 and β subunits of 20S proteasomes (37.5, 34.5, and 30 kDa, respectively, based on SDS-PAGE) (54) were digested in-gel by trypsin, and the corresponding peptide mixture was analyzed by tandem mass spectrometry. All of these spots were identified as either α1, α2, or β 20S proteasomal proteins of H. volcanii, as indicated in Fig. 1.

FIG. 1.

Isoforms of 20S proteasomal proteins. 20S proteasomal proteins associated with α1-His were purified from recombinant H. volcanii and separated by 2-DE (45 and 25 μg total protein are shown in panels A and B, respectively). Isoforms of α1 and/or α1-His (a to h), β (i to m), and α2 (n and o) proteins were identified by MS analysis of protein spots. (A) Molecular mass standards (on the left) and a pI range of 3.9 to 5.1 (on top) are indicated. (B) Magnified region of a 2-DE gel.

Based on these results, all three of the H. volcanii 20S proteasomal proteins (α1, α2, and β) migrated as isoform chains (Fig. 1). These proteasome fractions were composed primarily of α1 and β subunits, consistent with the low levels of α2 protein present in log-phase H. volcanii cells (20, 54). Approximately eight isoforms of the α1 protein were identified (Fig. 1). Although the α1 isoforms were most likely a mixture of α1 proteins with and without the C-terminal histidine tag, this alone did not account for the large number of α1-specific isoforms detected in the chain. In addition to α1, four to five distinct isoforms of the β protein were detected (Fig. 1). Based on SDS-PAGE, four of these isoforms (indicated as i, j, k, and m) were similar in migration to the mature β subunit versus that of the β precursor. This differentiation is based on a previous study (54) that used Edman degradation to determine the N-terminal residue of the 20S proteasomal β subunit to be Thr49 (residue number according to the deduced protein sequence). Thr49 is highly conserved with the active-site Thr of other β-type proteasomal proteins, which undergo autocatalytic processing to remove an N-terminal propeptide and expose the active-site Thr during 20S proteasome assembly (32, 46). The propeptide of the H. volcanii β precursor is rather large and accounts for 5,427 Da of the molecular mass of the polypeptide. This difference in molecular mass between the β subunit and β precursor protein enables the separation of these two proteins by SDS-PAGE. Thus, the three most basic spots of the β isoform chain (i, j, and k) and the highly acidic spot (m) are attributed to β subunits which have undergone processing. The spot with a slightly higher molecular mass (30.1 kDa; spot l) may be a partially processed β subunit or a non-β protein which copurifies during the proteasome enrichment procedure and migrates too close to the mature β subunit to differentiate it via MS analysis of the protein spot. Although the levels of α2 are lower in the cell prior to stationary phase than those of the α1 and β proteins, two distinct isoforms of α2 were detected (Fig. 1). Based on these results, the number of α- and β-type isoforms detected was significantly greater than the three polypeptides deduced from the complete genome sequence (α1, α2, and β). The identification of these multiple α1, α2, and β isoforms suggested that the proteasomal polypeptides are modified in the cell by mechanisms which alter their pI values.

It should be noted that the uppermost band of 66 kDa directly associated with the nickel-Sepharose column, based on its purification from not only the His tag expression strain (DS70[pJAM204]) but also the parent strain DS70. This protein was recently identified as PitA (1) and contains an N-terminal chlorite dismutase-related domain and a C-terminal antibiotic biosynthesis monooxygenase-related domain linked by a central histidine region, which most likely accounts for its divalent metal affinity. In contrast to the PitA protein, the purification of the predominant lower bands composed of α1, α2, and β was specific for the α1-His expression strain.

Analysis of proteasome isoforms by mass spectrometry.

To enhance the number of peptide fragments detected by mass spectrometry and increase the likelihood of detecting post- or cotranslational modifications of the proteasome isoforms, the proteasome-enriched fractions were separated by reducing one-dimensional (versus two-dimensional) SDS-PAGE. The α1, α2, and β proteins were excised from the gel as bands, reduced, alkylated, and cleaved either with trypsin, sequentially with trypsin and chymotrypsin, or with Glu-C endoprotease. This approach enhanced the protein sequence coverage relative to analysis of protein spots from two-dimensional gels (data not shown), most likely due to the larger amount of protein isolated from the one-dimensional gels. The peptide fragments resulting from the protease digests were separated by reverse-phase HPLC and analyzed by electrospray ionization (ESI)-QTOF MS/MS. Using this approach, the trypsin digest of the α1-His-purified sample resulted in 37.9% coverage of β, 47.8% coverage of α2, and 45.2% coverage of α1 (Table 1), with Mascot total protein scores of 2,162, 338, and 161, respectively (coverage and scores are based on complete polypeptide deduced from GenBank accession numbers AF126262, AF126261, and AF126260) (54). In combination with a trypsin and chymotrypsin sequential digest, total coverage of α1 and β was increased to 54.7% and 56%, respectively. No additional peptides were detected for α2 by using this sequential approach. The Mascot total protein scores for the sequential tryptic and chymotryptic digests of β, α2, and α1 were 763, 422, and 328, respectively. Summaries of these Mascot search results are presented in Table 1. The Glu-C endoprotease was used to enhance coverage of the N terminus of α2 (see below for details).

TABLE 1.

Fragments of α₁, α₂, and β identified by QSTAR from trypsin digest and sequential trypsin and chymotrypsin digest

Proteina	Res. no.b	Δc	Missd	Mascot ion score	Sequence (variable modification)
α1-T, 45.2%	1−12	0.04	0	76	MQGQAQQQAYDR (N-acetyl; oxidation [M])
	13−22	0.01	0	48	GITIFSPDGR
	23−30	0.03	0	44	LYQVEYAR
	35−43	0.01	1	32	RGTASIGVR
	44−55	0.03	1	32	TPEGVVLAADKR
	56−68	0.03	1	38	SRSPLMEPTSVEK (oxidation [M])
	72−88	0.06	0	58	ADDHIGIASAGHVADAR
	89−95	0.01	0	48	QLIDFAR
	105−116	−0.02	0	34	YGEPIGIETLTK
	150−163	−0.01	0	36	LYETDPSGTPYEWK
α2-T, 47.8%	12−22	0.02	0	59	GTSLFSPDGR
	23−29	0.03	0	44	IYQVEYAR
	43−53	0.03	0	61	TADGVVLAALR
	54−67	0.04	0	34	STPSELMEAESIEK
	104−115	0.03	0	49	YGEPIGVETLTK
	116−148	0.13	0	42	TITDNIQESTQSGGTRPYGASLLIGGVENGSGR (deamidation [NQ])
	149−162	0.00	0	45	LFATDPSGTPQEWK
β-T, 37.9%	21−36	−0.03	0	55	SNVFGPELGEFSNADR
	57−68	0.02	0	45	TEEGVVLATDMR
	69−78	0.01	0	47	ASMGYMVSSK (oxidation [M])
	83−109	0.05	0	65	VEEIHPTGALTIAGSVSAAQSLISSLR
	119−137	0.04	1	57	RGEDMSMQALSTLVGNFLR (double oxidatione [M])
	236−243	0.02	0	44	HQNFEGLE
α1-TC, 32.1%	13−22	−0.03	1	48	GITIFSPDGR
	58−68	−0.06	1	33	SPLMEPTSVEK
	72−88	−0.01	0	43	ADDHIGIASAGHVADAR
	89−95	−0.04	2	43	QLIDFAR
	105−116	−0.10	2	37	YGEPIGIETLTK
	126−149	−0.06	3	62	TQVGGARPFGVALLIGGVENGTPR
	135−149	−0.06	2	46	GVALLIGGVENGTPR
α2-TC, 34%	43−53	−0.05	2	59	TADGVVLAALR
	54−67	−0.05	1	48	STPSELMEAESIEK
	71−87	−0.05	2	74	LDDALGAATAGHVADAR
	104−115	−0.06	2	49	YGEPIGVETLTK
	116−133	−0.05	0	69	TITDNIQESTSGGTRPY
	134−148	−0.04	2	56	GASLLIGGVENGSGR
β-TC, 49.4%	57−68	−0.07	1	68	TEEGVVLATDMR
	69−78	−0.05	1	38	ASMGYMVSSK
	83−109	−0.04	3	98	VEEIHPTGALTIAGSVSAAQSLISSLR
	119−128	−0.06	1	52	RGEDMSMQAL
	119−131	0.06	2	57	RGEDMSMQALSTL (deamidation [Q]; phospho [S])
	129−137	−0.06	3	40	STLVGNFLR
	138−159	−0.09	3	55	SGGFYVVQPILGGVDETGPHIY
	214−235	−0.07	1	99	DLASGNGINIAVVTEDGVDIQR
	236−243	−0.03	2	45	HQNFEGLE

^aProteasomal α1, α2, and β proteins are shown with the percent coverage of the deduced protein sequences. -T, trypsin digested; -TC, trypsin and chymotrypsin digested.

^bRes. no., residue numbers of deduced protein sequences.

^cThe difference between the calculated and observed molecular mass (Δ value) is reported in daltons.

^dMiss, number of missed cleavages.

^eDouble oxidation of a single Met to sulfone (+32).

α1 is modified by N^α acetylation.

One of the ions detected by ESI-QTOF MS/MS was an N-terminal tryptic fragment of α1 which included an acetyl group (Fig. 2). It was a doubly charged ion (m/z, 741.343) with a Mascot individual ion score of 76 and an expect (E) value of 5.5e−7. It was composed of amino acid residues 1 to 12 (MQGQAQQQAYDR) and had a monoisotopic neutral mass of 1,480.67 Da. The peptide was oxidized and acetylated at the N-terminal Met residue. The MS/MS fragmentation of the ion produced a complete y ion series as well as several key b ions, including b(1) with a mass of 190.05 Da (Fig. 2). This mass corresponded to a methionine with an oxidation to methionine sulfoxide (147.04 Da) plus an acetylation (42 Da). Additional b and a ions support the reported peptide sequence which includes the oxidation and acetylation modifications. There were several internal fragment ions resulting in major peaks in the spectrum. The major peaks are labeled and listed in the inset table of Fig. 2. It should be noted that the N-terminally acetylated ion was obtained in three experiments, each time with two to three oxidation states of the same peptide (data not shown).

FIG. 2.

MS/MS fragmentation of acet-M_oxQGQAQQQAYDR, an N-terminal fragment of the α1 protein of 20S proteasomes. The m/z of the doubly charged parental ion was 741.34 Da, which corresponds to the mass of the defined fragment plus oxidation (16.00 Da for methionine sulfoxide [M_ox]) and acetylation (42.01 Da for acet-) of the N terminus with a delta value of <0.1 Da. The individual Mascot ion score was 76 with an E value of 5.5e−7. Labeled peaks include b and y ions as well as dominate internal ion fragments. The loss of ammonia (*, 17.03 Da) is indicated.

In addition to the acetylated α1 peptide fragment, a tryptic peptide ion of α1 was also detected which spanned residues 2 to 12 (QGQAQQQAYDR). It was a doubly charged ion (m/z, 646.8) with a Mascot individual ion score of 43 and an E value of 1 e−3. The MS/MS fragmentation spectrum contained a complete y ion series as well as four unique b ions supporting the N-terminal cleavage event. This result suggests that not all of the α1 proteins in the cell are acetylated, and a subset may instead be cleaved by methionine aminopeptidase. Thus, N-terminal acetylated and cleaved variants of α1 appear to exist in the cell simultaneously.

α2 is modified by N^α-acetylation.

Based on the finding that the α1 protein was N^α acetylated, it was speculated that α2 may also include this modification. In order to enhance the probability of observing peptides containing the N terminus of α2, this protein was digested with Glu-C endoprotease. The rationale for using Glu-C (versus trypsin) was based on the abundance of basic residues in the N-terminal region of α2 (e.g., Arg3 and Lys6). Trypsin was predicted to cleave carboxyl to these basic residues and generate short N-terminal peptide ions that would not be readily detected by liquid chromatography-MS/MS.

The Glu-C endoprotease digest of α2 protein generated a quadruply charged ion (m/z, 776.6) that contained the first 26 amino acids of the deduced protein sequence (MNRNDKQAYDRGTSLFSPDGRIYQVE). This peptide had a Mascot individual ion score of 24 and an E value of 0.1. The MS/MS fragmentation spectrum of this N-terminal peptide (see Fig. S1 in the supplemental material) contained primarily b-type ions, including a 185.1-Da ion which corresponded to an acetylated methionine and several y-type ions. The peptide contained an acetyl group on the N terminus, and the fourth amino acid (Asn4) was deamidated. The low Mascot ion score and corresponding high E value were likely consequences of the length and amino acid composition of the peptide. However, the ability to manually assign all of the peaks in the spectrum and the consistency of the data support the validity of this identification as an acetylated N-terminal peptide of α2. In contrast to α1, all of the N-terminal fragments of α2 detected by MS/MS were acetylated.

β subunit is phosphorylated.

By combining trypsin and chymotrypsin to digest the β subunit, several phosphopeptides (with different reverse-phase HPLC elution times) were detected by ESI-QTOF MS/MS. Only one of these peptides enabled the identification of a phosphosite. This phosphopeptide had an individual Mascot score of 57 and an E value of 2.5e−4. The peptide was 13 amino acids in length and spanned residues 119 to 131 (RGEDMSMQALSTL) (numbered according to the deduced protein sequence). It was a doubly charged ion (m/z, 760.340) corresponding to a monoisotopic neutral mass of 1,518.66 Da. The peptide contained a deamidation on the eighth residue and a phosphorylation on the eleventh residue (Gln126 and Ser129 of the deduced β protein sequence, respectively). The MS/MS fragmentation contained primarily b and a ions versus y ions (Fig. 3). The characteristic collision-induced dissociation neutral loss of H₃PO₄ and subsequent formation of dehydroalanine at residue 11 verified the phosphorylation site at Ser129. The predominance of a b ion series (and the corresponding lack of a y ion series) can be attributed to the presence of an arginine residue at the N terminus of the peptide. It has previously been observed that the presence of the basic amino acid residues arginine, lysine, and histidine can alter collision-induced fragmentation patterns seen in mass spectrometry (49). Based on these results, which included the spectrum shown in Fig. 3, Ser129 was determined to be the actual site of phosphorylation. A synthetic peptide (RGEDMSMQALpSTL) was synthesized and analyzed by direct infusion on QSTAR. The b ion series was again favored over the y ion series, and the characteristic b11 and b12 neutral-loss sequence ions (−98 Da) were observed (see Fig. S2 in the supplemental material).

FIG. 3.

MS/MS fragmentation of RGEDMSMQALpSTL, an internal phosphopeptide of the β subunit of 20S proteasomes. The fragment spans residues 119 to 131 of the deduced β protein and was generated by a sequential trypsin and chymotrypsin digest of the β subunit purified in complex with α1-His. The mass of the parent ion (1,518.66 Da) corresponds to the mass of the peptide (1,438.63 Da), deamination (1 Da) of glutamine [Q_(E)], and phosphorylation (79.97 Da) of serine (pS). The individual Mascot ion score was 62 with an E value of 6.8e−4. A predominant b ion series was detected due to the N-terminal arginine of the peptide. The characteristic collision-induced dissociation neutral loss of H₃PO₄ and subsequent formation of dehydroalanine at residue 11 verified the phosphorylation site at Ser129. b(11) − H₃PO₄ and b(12) − H₃PO₄ are b(11) and b(12) fragmentation ions with a neutral loss of H₃PO₄ and subsequent formation of dehydroalanine (loss of 97.99 Da total), respectively. The loss of ammonia (*, 17.03 Da) and H₂O (^o, 18.01 Da) is indicated. Noise filtering was used.

In addition to the β subunit phosphopeptide, an unphosphorylated form of the peptide fragment, spanning residues 119 to 131 (RGEDMSMQALSTL), was detected as doubly charged ions with various degrees of methionine oxidation (m/z, 719.82, 727.81, and 735.81); however, the respective Mascot ion scores were low at 23.5, 30, and 24, which significantly increased the E values up to 4.2.

Copurifying nonproteasomal proteins.

Three additional proteins were identified by tandem spectrometric analysis of the proteasome-enriched samples. These included ORF00357-, ORF02338-, and ORF02898-TG_gha_446 (see Table S1 in the supplemental material for details) with calculated molecular masses of 19,377.70, 20,752.38, and 32,702.80 Da, respectively. Whether these proteins copurify with α1-His or purify by nickel affinity chromatography alone remains to be determined. Proteins of molecular masses in the range of 15 to 40 kDa were not detected by SYPRO Ruby-stained SDS-PAGE gels of proteins purified by nickel affinity chromatography from the parent strain DS70. However, mass spectrometry can be more sensitive at protein detection than gel stains due to the diversity of protein chemical properties.

The high content of His residues in ORF02338- and ORF02898-TG_gha_446 as well as the prediction that these proteins bind metal (see Table S1 in the supplemental material) is consistent with the likelihood that these proteins bind the nickel column and not the α1-His protein. The likelihood is less clear regarding ORF00357-TG_gha_446, which is not predicted to bind metal. This latter protein contains a DUF293 domain and is a distant homolog of the heat shock transcriptional repressor HrcA, recently shown to bind and regulate a bacterial GroEL (53). Thus, it is tempting to speculate that ORF00357-TG_gha_446 may associate with proteasomes. Interestingly, this protein was also found to have an N-terminal serine residue modified by an acetyl group based on the detection of a modified tryptic peptide spanning residues 2 to 11 (acet-SSIELTSSQK) (see Fig. S3 in the supplemental material).

DISCUSSION

In this study, we demonstrate that all three of the H. volcanii 20S proteasomal proteins (α1, α2, and β) are modified by post- and/or cotranslational mechanisms. These modifications included N^α acetylation of α1 and α2 as well as the phosphorylation of Ser80 of the mature β subunit (Ser129 of the deduced polypeptide). In addition, ORF00357-TG_gha_446, a distant homolog of HrcA, was modified by the acetylation of an N-terminal serine residue.

Acetylation.

Cotranslational N^α acetylation is common in eukaryotes, with 50 to 90% of all proteins acetylated on N-terminal serine, alanine, glycine, or threonine residues after the removal of the initiating methionine (9, 14, 40). In contrast, based primarily on studies of Escherichia coli (48, 51), N-terminal acetylation is presumed to be posttranslational and rare in prokaryotes. Very few archaeal proteins have been found to be modified by N-terminal acetylation; however, a systematic characterization has not been performed to fully understand the prevalence of this type of modification among prokaryotes. Thus, it is unclear whether the acetyl groups that modified the N termini of the proteasomal α1, α2, and ORF00357-TG_gha_446 proteins, as detected in this study, are common or rare events in H. volcanii.

In the yeast Saccharomyces cerevisiae, N^α acetylation is catalyzed by at least three N-terminal acetyltransferases, NatA, NatB, and NatC, which contain Ard1p, Nat3p, and Mak3p catalytic subunits, respectively (40). These enzymes act cotranslationally on separate groups of substrates, each with degenerate motifs (38-40). NatA typically acts on N-terminal residues, such as Ser, Ala, Gly, and Thr, after the initiator Met is removed cotranslationally by methionine aminopeptidase (34). In contrast, NatB and NatC often modify the N-terminal initiator Met when it precedes either Glu, Asp, Asn, or Met residues (NatB) or the bulky hydrophobic residues Ile, Leu, Trp, or Phe (NatC).

For E. coli, three N-terminal acetyltransferases have been identified (RimI, RimJ, and RimL); however, these enzymes are only known to modify the ribosomal proteins S18, S5, and L12 with N-terminal Ala-Arg, Ala-His, and Ser-Ile residues, respectively (48, 55). There is no evidence that the Rim proteins act cotranslationally. Instead, at least the RimL enzyme appears to act posttranslationally based on the finding that the ribosomal proteins L7 and L12 have identical N termini and yet only L12 is acetylated (48). Thus, RimL is thought to recognize a certain protein structure and not just the extreme N terminus of the substrate protein.

Nothing is known regarding the mechanism or function of N-terminal acetylation in archaea. Furthermore, only a few proteins are known to be modified in this manner, including glutamate dehydrogenase-2 (GDH-2) (28) and Alba1 chromatin proteins of Sulfolobus solfataricus (4); ribosomal proteins S7P, L31e, and S19P of Haloarcula marismortui (18, 22, 26); and halocyanin of Natronobacterium pharaonis (31). The majority of these modifications are on N-terminal serine residues, exposed by a presumed methionine aminopeptidase and preceding either Ser or Ala (Ser-Ser/Ala) residues. Halocyanin represents an example for a relatively large group of predicted archaeal lipoproteins in which an internal Cys appears to be modified by a diphytanyl (glycerol) diether, converted to the N-terminal residue via cleavage by a signal peptidase II-like protease, and then N^α acetylated (31). GDH-2 is the single example (prior to this study) of an archaeal protein acetylated directly on its initiating N-terminal methionine residue (28). Although the presence of six internal N-epsilon-methyl-lysine residues is speculated to be responsible for the extreme thermostability of GDH-2 (28), the role of the N-acetyl group remains to be determined. Regarding the archaeal enzymes that catalyze N^α acetylation, recent phylogenetic analysis suggests that a family of uncharacterized bacterial and archaeal acetyltransferases may be responsible (40). In contrast to N^α acetylation, in which the modifying enzymes and function(s) remain to be elucidated, the acetylation of Alba1 on an internal lysine residue (Lys16 of the mature protein) is known to be mediated by Sir2 (4) and alleviate repression of minichromosome maintenance activity in Sulfolobus solfataricus (29).

In this study, ORF00357-TG_gha_446 follows the consensus of the majority of archaeal proteins characterized to date, in which it is proposed that the substrate is first cleaved by a methionine aminopeptidase and then acetylated at an N-terminal serine residue of a Ser-Ser/Ala motif (similar to those of eukaryal NatA substrates). In contrast, the acetylation of the α1 and α2 proteins at initiating N-terminal methionine residues is less common among archaea (to date) yet consistent with the S. solfataricus GDH-2. The consensus of these three archaeal proteins is acetyl-Met-Glu/Gln/Asn and, thus, somewhat similar to eukaryal NatB substrates. Regarding H. volcanii α1, a mixed population of N-terminal acetylated and nonacetylated proteins (the latter of which appear to be cleaved by a methionine aminopeptidase) was detected, suggesting that these modifications are regulated or vary between complexes of different quaternary structures. Although all of these modifying enzymes remain to be characterized in H. volcanii, the genome does have the coding capacity for a methionine aminopeptidase (ORF02385-TG_gha_446) in addition to members of the uncharacterized bacterial and archaeal N-terminal acetyltransferase family (e.g., ORF03062- and ORF02272-TG_gha_446) (see Table S2 in the supplemental material for details).

Whether the acetylation is specific to mature H. volcanii 20S proteasomes or lower-molecular-mass complexes and what the function of this modification is remain to be determined. In S. cerevisiae, which forms a single 20S proteasome of seven different α-type and seven different β-type subunits, all seven of the α-type subunits and two of the β-type subunits (β3 and β4) are modified by N-terminal acetylation (23). The yeast 20S proteasome purifies in a latent state that is activated by mild chaotropic agents, such as SDS and heat. Presumably, these agents selectively denature the N termini of the α-type subunits, which form tightly closed constrictions on each end of the cylinder (17), resulting in an opening of the channel of the 20S proteasome cavity. Although the role(s) of proteasomal acetylation remains to be determined, 20S proteasomes purified in the latent state from a nat1 mutant of S. cerevisiae (which lacks the NatA activity required for the acetylation of α1, α2, α3, α4, α7, and β3) were slightly more active in catalyzing chymotrypsin-like peptidase activity than those purified from the parental strain (23). Thus, the deletions of six of the nine N-terminal acetyl groups of the yeast 20S proteasome (five of which are localized to the α ring) are speculated to change the higher-order structure, possibly opening the central channel in the absence of mild chaotropic agents (23) and/or regulatory ATPases.

Early studies suggested that archaeal 20S proteasomes are not gated. This suggestion was based on the disordered configuration of the N termini of the α subunits of the crystal structure of a Thermoplasma acidophilum 20S proteasome purified from recombinant E. coli (27) and the ability of archaeal 20S proteasomes to hydrolyze short peptides in the absence of chaotropic agents (13, 32). However, the activities of most, if not all, archaeal 20S proteasomes are stimulated by heat and/or SDS (13, 32), and the deletion of residues 2 to 12 of the recombinant T. acidophilum α protein generates a central pore in the α rings evident by electron microscopy that increases the capacity of 20S proteasomes to degrade proteins with little tertiary structure (e.g., β-casein) (5, 47). Based on this, the N termini of archaeal α-type subunits do appear to function as a gate that prevents entry into 20S proteasomes of polypeptides seven residues or larger. Thus, regulated acetylation of the H. volcanii α1 and α2 subunits may control the configuration of the 20S gate and ultimately modulate substrate entry and/or the association of regulatory ATPases.

Phosphorylation.

Several studies have demonstrated that archaeal proteins are phosphorylated; however, relatively little is known regarding the identities of the proteins that are modified or the sites of modification (15, 21). Furthermore, the impact phosphorylation has on the functional properties of the archaeal proteins that are modified or the kinases/phosphatases that regulate these events is poorly understood. Thus, the identification of a phosphorylation site for the β subunit of 20S proteasomes is a significant advance in our understanding of archaeal protein phosphorylation.

The Ser129 residue of the H. volcanii β protein that is phosphorylated is not highly conserved among archaea. Of 37 archaeal β-type proteins analyzed, the vast majority (>85%) had an alanine residue at the analogous position, with only close archaeal homologs (e.g., Haloarcula marismortui Q5V4S6) (16) retaining this residue. Broadening the BLAST search revealed that the Ser129 residue is more widespread in eukaryotes than in archaea and was conserved in all eukaryal β5 and β5i (LMP7i) proteasomal subunits examined, ranging from yeast to humans. β5 is responsible for the chymotrypsin-like peptidase activity of the housekeeping 20S proteasomes (36), and β5i replaces β5 to form immunoproteasomes after gamma interferon induction (25).

In this study, the phosphorylated form of the H. volcanii β protein was isolated in complex with the His-tagged α1 subunit. Unphosphorylated peptides, which spanned this same phosphosite (β Ser129) and copurified with α1, were also detected. Since all of the β proteins present in the α1-His-enriched samples were in complex with α1 and appeared to be similar to mature (processed) β subunits, the results suggest that both phosphorylated and unphosphorylated forms of β are present in 20S proteasomes. Automated modeling (45) of the β Ser129 to crystal structures of analogous 20S proteasomes suggests that this residue is located at the interface of the β subunits responsible for the association into heptameric rings and distant from the interface of the two β rings and the β-α ring interface. Thus, the phosphorylation of β Ser129 may influence the formation of the heptameric β ring as it assembles on the α ring scaffold.

Although the phosphorylation of archaeal 20S proteasomes has not previously been described, a number of phosphosites of the eukaryotic counterpart have been mapped, including α3 Ser248 of Candida albicans (35); α7 Ser243 and Ser250 of monkey (8); α7 Ser243, α7 Ser250 (10), and α2 Tyr120 of rat (6); and α5 Ser56 (3), α7 Ser249 (3, 11), α2 Tyr23, α2 Tyr97, α7 Tyr160, β2 Tyr154, and β7 Tyr102 (43) of human (Fig. 4 and 5). Although not identified in this study, several of these phosphorylation sites are conserved in H. volcanii 20S proteasome subunits, including the human α2 Tyr23, α5 Ser56, and β2 Tyr154 residues; rat α2 Tyr120 residue; and C-terminal Ser/Thr residues of a number of α-type subunits. It is possible that these modifications do occur in H. volcanii and may account, in part, for the additional proteasomal isoforms detected by two-dimensional SDS-PAGE in this study (Fig. 1).

FIG. 4.

Multiple-sequence alignment of select α-type 20S proteasome proteins. Phosphorylation sites determined by biochemical studies are boxed and indicated by a closed circle below the alignment. These include Hs_a7 Ser249 (2, 10); Hs_a7 Tyr160, Hs_a2 Tyr23, and Hs_a2 Tyr97 (42); Rat_a2 Tyr120 (5); Hs_a5 Ser56 (2); Rat_a7 and monkey Ser243 and Ser250 (7, 9); and Cal_a3 Ser248 (34). Hs, Homo sapiens; Cal, Candida albicans; Rat, Rattus norvegicus; Hvo, Haloferax volcanii. Swiss-Prot or GenBank accession numbers are as follows: Hs_a1, P60900; Hs_a2, P25787; Rat_a2, P17220; Hs_a3, P25789; Cal_a3, 46441895; Hs_a4, O14818; Hs_a4-like, Q8TAA3; Hs_a5, P28066; Hs_a6, P25786; Hs_a7, P25788; Rat_a7, 203207; Hvo_a1, Q9V2V6; Hvo_a2, Q9V2V5. Conserved residues are highlighted in gray and black, with amino acid position numbers indicated on the right and left.

FIG. 5.

Multiple-sequence alignment of the central region of select β-type 20S proteasome proteins. Phosphorylation sites determined by biochemical studies are boxed and indicated by a closed circle below the alignment. These include Hs_b7 Tyr102 and Hs_b2 Tyr154 (42) and Hvo_b Ser129 (this study). Hs, Homo sapiens; Hvo, Haloferax volcanii. Swiss-Prot or GenBank accession numbers are as follows: Hs_b1, P28072; Hs_b1i, P28065; Hs_b2, Q99436; Hs_b2i, P40306; Hs_b3, P49720; Hs_b4, P49721; Hs_b5, P28074; Hs_b5i, P28062; Hs_b6, P20618; Hs_b7,P28070; and Hvo_b, Q9V2V4. Conserved residues are highlighted in gray and black, with amino acid position numbers indicated on the right and left.

In an analogy to eukaryotic cells, it is quite likely that the phosphorylation site of the H. volcanii 20S proteasome β protein detected in this study regulates biological function. A number of studies with eukaryotes suggest that phosphorylation regulates the subcellular distribution and assembly of 26S proteasomes. For example, the phosphorylation of the regulatory particle Rpt6 subunit appears to promote its association with the α2 subunit of 20S proteasomes to form the 26S proteasome (44). Likewise, the levels of phosphorylated α7 and α3 are higher in 26S proteasomes than in free 20S proteasomes (30). The phosphorylation at either one of the two α7 sites (Ser243 and Ser250) is essential for the association with 19S regulatory complexes, and the ability to undergo phosphorylation at both sites gives the most efficient incorporation of α7 into 26S proteasomes (42). In addition, gamma interferon treatment decreases the level of phosphorylation of proteasomes, which coincides with a decrease in the levels of 26S proteasomes and an increase in PA28-containing proteasomes, thus suggesting that phosphorylation may play a role in regulating the formation of proteasomal complexes in animal cells (8). Thus, the assembly and disassembly of 26S proteasomes appear to be regulated by kinase(s) and/or phosphatase(s). Proteasomal phosphorylation also appears to be important in development and subcellular distribution. For example, the α4 subunit of proteasomes is preferentially phosphorylated in immature versus mature oocytes of goldfish (19). Furthermore, the phosphorylation of the α2 Tyr120 residue of rat is important in nuclear localization of proteasomes (6). Future studies are needed to elucidate the role of the acetylation and phosphorylation of the H. volcanii proteasomal proteins observed in this study. The identification of several of these modification sites now lays a foundation for these studies.