The 20S Proteasome of Streptomyces coelicolor

Abstract

20S proteasomes were purified from Streptomyces coelicolor A3(2) and shown to be built from one α-type subunit (PrcA) and one β-type subunit (PrcB). The enzyme displayed chymotrypsin-like activity on synthetic substrates and was sensitive to peptide aldehyde and peptide vinyl sulfone inhibitors and to the Streptomyces metabolite lactacystin. Characterization of the structural genes revealed an operon-like gene organization (prcBA) similar to Rhodococcus and Mycobacterium spp. and showed that the β subunit is encoded with a 53-amino-acid propeptide which is removed during proteasome assembly. The upstream DNA region contains the conserved orf7 and an AAA ATPase gene (arc).

Bacterial intracellular proteases have a variety of functions, including processing of precursor forms (such as removal of signal peptides in exported proteins), degradation of aberrant or damaged proteins (resulting from mutations or environmental stress), and inactivation of proteins that play key roles in regulatory processes. Such a diversity of tasks implicates the involvement of a complex set of proteolytic enzymes, as exemplified by the well-characterized Escherichia coli system (12). Some of these proteases, like Lon, have homologues in Archaea and Eucarya, whereas others seem to have a restricted interdomain distribution. Proteasomes were thought to be absent from Bacteria, until 20S proteasomes were discovered in the actinomycete Rhodococcus (42).

In eukaryotic cells, the 20S proteasome constitutes the proteolytic core of the 26S proteasome, which contains several accessory proteins (including ATPases) that enable selective proteolysis of ubiquitin-tagged substrates (9, 43), but such a 20S proteasome-associated regulatory entity has not yet been identified in Bacteria or Archaea. The 20S proteasome belongs to the group of self-compartmentalizing proteases, in which the active sites are confined to an interior compartment, created by self-assembly of a number of subunits (4, 21). The barrel-like structure of the 20S proteasome results from the stacking of four seven-membered rings: the two outer rings are composed of α-type subunits and the two inner rings are built from β-type subunits, in which the active sites are generated by autocatalytic processing during proteasome assembly. In eukaryotes such as yeast, different α-type and β-type subunits (seven of each) generate the 20S complex (14). In archaebacteria such as Thermoplasma acidophilum, the composition with only one α- and one β-type subunit is of minimal complexity (20), which has facilitated elucidation of the novel catalytic mechanism involving the N-terminal threonine of the β subunit generated upon autocatalytic maturation (37, 38). An intermediate degree of complexity is provided by the 20S proteasome of Rhodococcus erythropolis NI86/21, which is built from two different α- and β-type subunits (42, 49). The disclosure of proteasome-like genes by genomic sequencing of the related nocardioform actinomycetes Mycobacterium leprae and Mycobacterium tuberculosis (8, 22) and the subsequent characterization of the 20S proteasome genes in Mycobacterium smegmatis (18) revealed that the α₁₄β₁₄ subunit composition, as found in archaebacteria, also occurs in eubacteria. In this communication, we report the biochemical and genetic characterization of the 20S proteasome from a phylogenetically distant actinomycete, Streptomyces coelicolor strain A3(2).

Purification of 20S proteasomes from S. coelicolor.

S. coelicolor A3(2) was grown at 30°C for 3 days in medium containing casein (10 g/liter), yeast extract (5 g/liter), glucose (5 g/liter), glycine (5 g/liter), and 5 mM MgCl₂. Cells harvested from 3 liters of culture were washed with 50 mM HEPES buffer (pH 8.0) and resuspended in 100 ml of this buffer containing lysozyme (1 mg/ml). The cell suspension was kept on ice for 2 h. All further steps were carried out at 4°C, unless specified otherwise. DNase I (200 U) was added to the lysate, which was cleared by centrifugation at 61,700 × g for 1 h. Twenty milliliters of supernatant (containing about 230 mg of protein) was loaded on a Sepharose 6B column (3.2 by 86 cm; Pharmacia) and eluted with 50 mM Tris-HCl buffer (pH 7.5) containing 1 mM dithiothreitol (DTT) and 20% (vol/vol) glycerol (buffer A) at a flow rate of 46 ml/h. Fractions (5 ml) were collected and assayed for proteinase activity by using the synthetic substrate succinyl-Leu-Leu-Val-Tyr-7-amido-4-methylcoumarin (Suc-LLVY-AMC) (Bachem). The fluorigenic synthetic peptide (10 nmol) was incubated for 15 to 60 min at 37°C in 50 mM Tris-HCl buffer (pH 8.0) with the enzyme samples in a total reaction volume of 100 μl. The reaction was stopped by adding 100 μl of 10% (wt/vol) sodium dodecyl sulfate (SDS), and the fluorescence was measured to estimate the release of the 7-amido-4-methylcoumarin moiety. The active, high-molecular-mass fractions from three Sepharose 6B runs were pooled and loaded on a DEAE-Sephacel column (2.2 by 10 cm; Pharmacia) equilibrated with buffer A. Bound proteins were eluted with a 0 to 0.5 M NaCl linear gradient in 400 ml of buffer A. Fractions of 4 ml were collected. The fractions with proteolytic activity eluting at approximately 300 mM NaCl were pooled and dialyzed against 10 mM potassium phosphate buffer (pH 7.0) containing 1 mM DTT and 20% glycerol. The dialyzed sample was applied to a hydroxyapatite column (1.4 by 6 cm; Bio-Rad) equilibrated with 10 mM potassium phosphate buffer containing 20% (vol/vol) glycerol. A 10 to 300 mM potassium phosphate linear gradient (100 ml) was used for elution, and 1.5-ml fractions were collected. Fractions (1.5 ml) with proteolytic activity on Suc-LLVY-AMC and which eluted at approximately 85 mM potassium phosphate were pooled and dialyzed against 25 mM Tris-HCl (pH 7.5) containing 1 mM DTT and 20% glycerol (buffer B). This sample was further purified on a Q Sepharose column (1.2 by 6 cm; Pharmacia). Fractions of 1 ml were collected during linear gradient elution with 200 to 600 mM NaCl (50 ml). Fractions (1 ml) with proteolytic activity, eluted at about 470 mM NaCl, were again pooled and dialyzed against buffer B. The final purification step involved linear gradient elution (0 to 0.6 M NaCl in 40 ml) from a Mono Q column. The fractions with proteasomes, eluted at approximately 480 mM NaCl, were dialyzed against buffer A and used for further characterization. Table 1 presents an overview of the purification procedure.

TABLE 1.

Purification of 20S proteasomes from S. coelicolor

Purification step	Total protein (mg)	Total activity (nmol h−1)	Sp act (nmol h−1 mg−1)	Purifi-cation factor	Yield (%)
Crude extract	700	3,200	4.6	1	100
Sepharose 6B	76.7	3,080	40.2	9	96
DEAE-Sephacel	13.7	2,160	158	34	68
Hydroxyapatite	3.37	1,683	499	109	53
Q Sepharose	0.26	1,333	5,127	1,115	42
Mono Q	0.057	880	15,439	3,356	28

Characterization of the S. coelicolor 20S proteasome.

Electron micrographs of negatively stained proteasomes show the two characteristic views (end-on and side-on) of the barrel-like 20S proteasome (Fig. 1). SDS-polyacrylamide gel electrophoresis analysis showed that the 20S proteasome preparation was homogeneous, revealing two bands of equal intensities with estimated relative molecular weights of 24,400 (β subunit) and 29,700 (α subunit) (Fig. 2). Purification of 20S proteasomes from the archaea Methanosarcina thermophila (24) and Pyrococcus furiosus (2), together with analyses of complete archaeal genomes (Archaeoglobus fulgidus, Methanobacterium thermoautotrophicum, and Methanococcus jannaschii [overview at http: //www.tigr.org/tigr_home/tdb/mdb/mdb.html]), indicate that most archaeal 20S proteasomes have an α₁₄β₁₄ subunit composition. However, we noted that the ongoing genomic sequencing of Pyrococcus horikoshii has already disclosed two distinct β-type subunit genes. The increased level of subunit complexity reported for the R. erythropolis NI86/21 20S proteasome (42, 49) has not been observed for 20S proteasomes from other eubacteria (actinomycetes), namely Mycobacterium spp. (8, 18) and S. coelicolor (this work).

FIG. 1.

Electron micrograph of 20S proteasomes from S. coelicolor, negatively stained with uranyl acetate (3), showing ring-shaped end-on views representing projections along the cylinder axis and rectangular side views corresponding to projections perpendicular to the cylinder axis. Bar, 100 nm.

FIG. 2.

Tricine-SDS-polyacrylamide gel electrophoresis analysis of 2.28 μg of purified S. coelicolor 20S proteasomes (right lane). The sizes of the marker proteins (left lane) are 97.4, 66.2, 45.0, 31.0, 21.5, and 14.4 kDa (from top to bottom). The proteins were stained with Coomassie brilliant blue.

The substrate specificity of the Streptomyces 20S proteasome was determined at 37°C with various synthetic substrates (Bachem). The highest activity was obtained with Suc-LLVY-AMC (15.4 μmol h⁻¹ mg⁻¹ = 100%). Substantially lower activities were measured with two other chymotryptic substrates, benzoyl- oxycarbonyl-Gly-Gly-Leu-7-amido-4-methylcoumarin (0.78 μmol h⁻¹ mg⁻¹ = 5%) and succinyl-Ala-Ala-Phe-7-amido-4-methylcoumarin (0.15 μmol h⁻¹ mg⁻¹ = 1%), but no such activity was detectable with succinyl-Leu-Tyr-7-amido-4-methylcoumarin. Trypsin-like activity was assayed with benzoyloxy-carbonyl-Gly-Gly-Arg-7-amido-4-methylcoumarin, tert-butyl-oxycarbonyl-Leu-Arg-Arg-7-amido-4-methylcoumarin, ben-zoyloxycarbonyl-Ala-Arg-Arg-7-amido-4-methylcoumarin, and benzoyl-Val-Gly-Arg-7-amido-4-methylcoumarin, whereas benzoyloxycarbonyl-Leu-Leu-Glu-β-naphthylamide (Z-LLE-βNA) was used as a peptidyl-glutamyl peptidase substrate; however, neither of these hydrolytic activities were demonstrated for the 20S proteasome from Streptomyces. This substrate spectrum is quite similar to that reported for the 20S proteasome from Rhodococcus, with high activity on chymotryptic substrates but no other detectable peptidase activities (42). High chymotrypsin-like activity is also a feature of the proteasomes from T. acidophilum (1, 10) and P. furiosus (2). Remarkably, the M. thermophila proteasome displays about 40% higher activity on Z-LLE-βNA than on Suc-LLVY-AMC (23).

The effects of some known proteasome inhibitors on the activity of the Streptomyces proteasome were determined (Fig. 3). Lactacystin (Affiniti) was a more potent inhibitor of the Streptomyces 20S proteasome than acetyl-Leu-Leu-norleucinal (Ac-LLnL) (Sigma) and benzoyloxycarbonyl-Leu-Leu-Leu-vinylsulfone (Z-LLL-VS). Very similar inhibition patterns were reported for the chymotrypsin-like activities of eukaryotic proteasomes preincubated with these inhibitors (6). For the Thermoplasma 20S proteasome, lactacystin is a moderately potent inhibitor (15% residual activity with 100 μM lactacystin [41]). Unlike Ac-LLnL, both lactacystin and Z-LLL-VS irreversibly inactivate proteasomes by covalent modification of the active-site threonine (6, 11, 25). Remarkably, lactacystin is produced by a Streptomyces strain (29). It will be of interest to investigate what may be the physiological implications for Streptomyces cells that would produce both a major intracellular protease and a potent specific inhibitor of this enzyme. This is reminiscent of the regulation of activity of a trypsin-like protease that is essential for aerial mycelium formation by the autogenous inhibitor leupeptin (acetyl-leucine-leucine-arginal) and a leupeptin-inactivating enzyme in Streptomyces exfoliatus (17).

FIG. 3.

Inhibition of the S. coelicolor 20S proteasome by Ac-LLnL, Z-LLL-VS, and lactacystin. Purified proteasome (0.912 μg) was incubated with inhibitor in 100 μl of Tris-HCl (pH 8.0) at room temperature for 1 h. Subsequently, 20 nmol of Suc-LLVY-AMC was added. After incubation at 37°C for 1 h, the reaction was stopped (by addition of 100 μl of 10% SDS) and the amount of 7-amido-4-methylcoumarin released was measured.

The optimum temperature for Suc-LLVY-AMC hydrolysis by the Streptomyces 20S proteasome was 55°C (data not shown), which is substantially lower than the temperature optima reported for T. acidophilum (85°C to 91°C [1, 10]), P. furiosus (95°C [2]), and M. thermophila (70°C to 75°C [23]). This difference probably reflects the thermophilic nature of these archaebacteria, as opposed to Streptomyces with an optimal growth temperature of about 30°C.

Cloning and characterization of the 20S proteasome structural genes from S. coelicolor.

From S. coelicolor cells grown in glycine-containing medium and treated with lysozyme (1 mg/ml) at 30°C for 30 min, total DNA was isolated according to the method of Nagy et al. (26). A digoxigenin-labeled 960-bp SacI-EcoRI fragment from the R. erythropolis prcB₂A₂ operon (42) was used as a probe for Southern hybridization (hybridization temperature, 60°C; washing was performed at 60°C with 7.5 mM sodium citrate containing 75 mM NaCl and 0.1% SDS) with total S. coelicolor DNA restricted by several endonucleases. Two hybridizing fragments (5.1-kb KpnI and 8.0-kb SphI fragments), covering a 10.6-kb region with the proteasome genes and flanking DNA, were cloned in pUC18 (generating pFAJ2594 and pFAJ2609, respectively). These clones were selected by colony hybridization with the Rhodococcus DNA probe from size-fractionated sublibraries. Subcloned fragments were subjected to automated sequence analysis (A.L.F. sequencer; Pharmacia), and subsequences were assembled with the ASSEMGEL program of PCGENE (IntelliGenetics). Potential codon regions were identified with the GCWIND program (40), relying on codon usage bias in genomes with high G+C contents.

Two open reading frames (ORFs) (prcB, prcA) with high homology to known eubacterial proteasome genes and organized in a similar way were identified (Fig. 4). The N-terminal sequence determined for the smaller subunit of the purified 20S proteasome (TTIVAVTF) was identical to the deduced amino acid sequence of PrcB after removal of a 53-amino-acid propeptide, characteristic of β-type subunits (Fig. 5A). The calculated molecular mass of the mature PrcB protein (24,428 Da) is in excellent agreement with the estimated M_r of 24,400. The propeptides of eubacterial β subunits are considerably longer (up to 65 residues for PrcB₁ of R. erythropolis) than those of archaebacterial β subunits, which typically consist of fewer than 10 amino acids. Zühl et al. (48) demonstrated that the propeptides of the Rhodococcus proteasome β subunits play a crucial role: they support the initial folding of the β subunits and promote the maturation of the holoproteasomes, in which the N-terminal active-site residue, threonine, becomes available following autocatalytic release of the propeptides. Such a chaperone-like function has also been assigned to the propeptide of a eukaryotic β-type proteasome subunit (7). A comparison of the currently available eubacterial propeptide sequences (Fig. 5A), reveals that two residues (HG) preceding the processing site (G-T peptide bond) are strictly conserved, as well as several residues in the middle part of the propeptide, whereas little conservation is apparent in the remainder of the sequences. It is likely that the central box with the consensus sequence SSFX(D/E)(F/Y/L)LX₄PEXLP is important for the function of the eubacterial propeptide. Such a box is not obviously present in eukaryotic propeptides, although a number of them contain the SF or AF sequence. No N-terminal sequence was obtained for the larger subunit, indicating that the amino terminus was blocked, a general characteristic of α-type subunits (42). The estimated M_r of this subunit (29,700) is somewhat higher than the calculated molecular mass of 27,883 Da.

FIG. 4.

Gene organization of the proteasome structural genes (prcB, prcA) and the flanking regions from M. tuberculosis (MT) (GenBank accession no. Z97559; 14,030 bp), R. erythropolis (RE) (GenBank accession no. U26422 and this work; 13,000 bp), and S. coelicolor (SC) (this work; 10,612 bp). The fragment shown for Rhodococcus represents the second proteasome operon. The ORFs in the respective DNA regions are numbered independently. For M. tuberculosis, the ORF numbering of cosmid MTCY261 is used. The box labeled IS represents one of the sixteen copies of IS6110 present in the genome of M. tuberculosis H37Rv (8). Conserved ORFs and homologous genes are shown as filled arrows, and the extent of sequence conservation is indicated (numbers representing percentages of identical amino acids are circled).

FIG. 5.

Multiple sequence alignments of the proteasome β-subunit propeptides (A), the ORF7 sequences (B), and the N-terminal coiled-coil regions of ARC homologues (C) from the following actinomycetes: M. leprae (Ml), M. smegmatis (Ms), M. tuberculosis (Mt), R. erythropolis (Re1 and Re2, or Re), and S. coelicolor (Sc). Differential shading (three levels in panels A and B; two levels in panel C) reflects the degree of sequence conservation (identical or similar amino acids) at a given position. The propeptide cleavage sites are marked with an arrow, and the first 10 residues of the processed β subunits are shown in panel A. In panel C, the hydrophobic amino acids in positions 1 and 4 of the heptad repeats of the proposed coiled-coil structures of ARC proteins are labeled • and ▿, respectively, and proline residues flanking coiled-coil segments are marked (in black on a shaded background).

Organization of the proteasome genomic regions in actinomycetes.

Although purification and biochemical characterization of 20S proteasomes is currently limited to Rhodococcus (42) and Streptomyces (this work), genomic sequencing of M. tuberculosis (8) and M. leprae, and subsequent genetic analysis with M. smegmatis (18), now enables a comparison of the proteasome regions of representative actinomycetes to be made. To extend the known DNA sequence of the upstream region of prcB₂A₂ of R. erythropolis NI86/21, the flanking sequences of the arc gene (46) were determined from the original λFAJ2029 clone (42).

Figure 4 reveals a patchy distribution of conserved genes and ORFs (apparently restricted to actinomycetes) with several interjacent ORFs that are not conserved between the different actinomycete species. The 20S proteasome structural genes of S. coelicolor, prcB and prcA, are separated by 61 bp. However, in Rhodococcus (42) and Mycobacterium (8, 18) spp., the prcB stop codon overlaps the prcA start codon, suggesting a translational coupling. The small conserved orf7, which overlaps the respective prcB genes in Rhodococcus and Mycobacterium, ends 580 bp upstream of the Streptomyces prcB gene. The Streptomyces ORF7 (72 amino acids) is somewhat larger than its counterparts from Rhodococcus and Mycobacterium (63 to 64 amino acids) but is quite similar, particularly in the C-terminal half, which contains a high proportion of acidic residues (Fig. 5B). The calculated pI value for Streptomyces ORF7 (3.7) and those of its homologues (3.6 to 3.7) are very similar. At present, it is not known whether this conserved ORF7 may exert a proteasome-related function. A possible ORF (orf8) is present between orf7 and prcB and partially overlaps the latter, but no homologues were identified via database searches. Using a proteasome-specific probe, the proteasome genes have been mapped to the AseI-I fragment (cosmid I4) of the S. coelicolor genome (16, 32).

Other conserved ORFs in the upstream region of S. coelicolor are orf6, arc, and orf1 (partially sequenced). The arc gene product of R. erythropolis, ARC (for AAA ATPase forming ring-shaped complexes), was recently characterized (46). The linkage with proteasome genes and striking similarities in domain topology with those AAA-type ATPases forming part of the eukaryotic 26S proteasomes (5, 30) suggest a proteasome-related function for ARC, but this remains to be proven. A feature of the Rhodococcus ARC (and of the putative mycobacterial homologues) which it shares with proteasomal ATPases (33) is the presence of a potential N-terminal coiled-coil structure (46). By using the COILS server (http://www.isrec.isb-sib.ch/software/COILS_form.html), such an N-terminal segment was also predicted for Streptomyces ARC (Fig. 5C).

Unlike the gene organization in Mycobacterium and Rhodococcus, no ORFs are positioned between arc and another conserved (actinomycete-specific) ORF, orf6, in Streptomyces (Fig. 4). It is notable that the deduced ORF6 sequence has significant homology (about 40% identity) to the putative orf9 product (27). A close homologue of ORF9 (partially sequenced) appears to be present in S. coelicolor as well (Fig. 4). In M. tuberculosis, but not in M. leprae (27), the equivalent of orf9 is located in another part of the genome (cosmid MTCY49 [GenBank accession no. Z73966]). The presence of a copy of the promiscuous insertion sequence IS6110 downstream of the proteasome genes in M. tuberculosis (Fig. 4) suggests that a genomic rearrangement may have taken place.

The Streptomyces orf1 and its homologues in Rhodococcus (orf16) and Mycobacterium (orf14) encode putative methyltransferases with significant homologies to putative gene products from several archaebacteria (up to 37% identity for P. horikoshii [GenBank accession no. AB009528]). Collectively, these ORFs are distantly related to l-isoaspartyl protein carboxyl methyltransferases from various organisms; these are involved in the repair of l-isoaspartyl residues present in damaged proteins (45). For the remaining, nonconserved Streptomyces ORFs, homology searches revealed similarity to ferredoxins for ORF3, to LacI-related repressors (including S. coelicolor MalR [44]) for ORF4, and to transmembrane transporter proteins for ORF5, but the extent of the homology (<40% identity) did not allow reliable predictions of their possible functions.

Conclusions.

The presence of 20S proteasomes in Rhodococcus, Mycobacterium, and Streptomyces spp. suggests that this cytoplasmic protease is widespread among actinomycetes (gram-positive bacteria with high G+C contents). The presence of proteasome-like genes in other representative actinomycete species (Amycolatopsis methanolica, Brevibacterium linens, Clavibacter michiganense, and Corynebacterium glutamicum) was confirmed by Southern hybridization with a rhodococcal proteasome gene probe (28). Furthermore, a polyclonal antibody raised against Rhodococcus proteasomes and cross-reacting with the S. coelicolor β-type subunit revealed the presence of putative β-type subunits in the SDS-insensitive, Suc-LLVY-AMC-active fractions obtained by glycerol gradient centrifugation of cell extracts from these species (28). Recently, proteasome genes have also been identified in Frankia sp. strain ACN 14a/ts-r (31).

Genomic sequencing data support the notion that 20S proteasome genes are confined to actinomycetes. In many other bacteria, including gram-positive bacteria (such as Bacillus spp.), a structurally different self-compartmentalizing protease, HslVU, is present (19, 34). This protease is built from the Clp-related ATPase subunit HslU (13, 39) and from the proteolytic HslV subunit, which shows some sequence homology (22) and similarity in catalytic mechanism (6, 47) to β-type proteasome subunits. In E. coli, a synergistic action of HslVU with other ATP-dependent proteases (such as Lon) in controlling the turnover of ς³² (sigma factor directing transcription of heat shock genes) and abnormal proteins has been observed (15). Lon homologues have been identified in mitochondria, archaebacteria, and eubacteria, including the fast-growing, nonpathogenic M. smegmatis (35). To our surprise, we were unable to identify a gene for a Lon homologue in the genome of the slow-growing M. tuberculosis H37Rv (8). It will be of interest to see whether this may also be the case for other pathogenic mycobacterial species, such as M. leprae.

Recently, it was shown that specific inhibition of proteasomes with Z-LLL-VS severely affected cell viability of the archaeon T. acidophilum following exposure to heat shock (36). However, an M. smegmatis mutant lacking proteasomes showed no obvious phenotypic alterations (18), indicating that an efficient backup system may exist in actinomycetes. The elucidation of the in vivo function of 20S proteasomes in actinomycetes should benefit from the identification of this protease and its structural genes in Streptomyces, since molecular genetic tools are much more advanced for this genus than for other actinomycetes.

Nucleotide sequence accession numbers.

The nucleotide sequences of the S. coelicolor proteasome genes (including flanking DNA) and the nucleotide sequence of the arc-containing DNA region upstream of the R. erythropolis prcB₂A₂ genes have been deposited in the GenBank database under accession no. AF086832 and AF088800, respectively.