View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

The first characterization of a eubacterial : the 20S complex of Rhodococcus Tomohiro Tamura*t, Istvin Nagy**, Andrei Lupast, Friedrich Lottspeicht, Zdenka Cejkat, Geert Schoofs*, Keiji Tanaka§, Rene De Mot* and Wolfgang Baumeistert tMax-Planck-lnstitute for Biochemistry, D-82152 Martinsried, Germany. F.A. Janssens Laboratory of Genetics, Catholic University of Leuven, B-3001 Heverlee, Belgium. §Institute for Research, University of Tokushima, Tokushima 770, Japan.

Background: The 26S proteasome is the central chymotryptic substrate Suc-Leu-Leu-Val-Tyr-AMC in of the ubiquitin-dependent pathway of protein degrada- the presence or absence of 0.05 % SDS. Purified prepara- tion. The proteolytic core of the complex is formed by tions reveal the existence of four subunits, two of the the 20S proteasome, a cylinder-shaped particle that in a-type and two of the 3-type, the genes for which we archaebacteria contains two different subunits ( and 3) have cloned and sequenced. Electron micrographs show and in eukaryotes contains fourteen different subunits that the complex has the four-ringed, cylinder-shaped (seven of the a-type and seven of the 3 -type). appearance typical of . Results: We have purified a 20S proteasome complex Conclusions: The recent description of the first eubac- from the nocardioform actinomycete Rhodococcus sp. strain terial ubiquitin, and our discovery of a eubacterial pro- NI86/21. The complex has an apparent relative mol- teasome show that the ubiquitin pathway of protein ecular mass of 690 kD, and efficiently degrades the degradation is ancestral and common to all forms of life.

Current Biology 1995, 5:766-774

Background proteasome at 3.4 A resolution [15,16] and the identifica- tion of its catalytic nucleophile, a threonine residue Proteasomes are ubiquitous multisubunit of (PThrl) of the 13 subunit [16-18]. eukaryotic organisms. The 26S proteasome is the central protease of the ubiquitin-dependent pathway of protein Recently, a scan of the GenBank database revealed the degradation [1-4] and plays a key role in many cellular existence of several proteins from eubacteria with processes including the cell cycle [5,6], transcriptional sequences that show significant similarity to those of regulation [7,8] and antigen presentation [9]. The protein proteasome -type subunits [19]. Although most of them is formed by two, asymmetric, 19S cap complexes occur in an operon with a ClpX-related ATPase, and attached to the ends of a barrel-shaped 20S particle [10]. may form structures more similar to the Clp protease The 20S particle, generally referred to as the multi- than to proteasomes, one protein, from Mycobacterium catalytic proteinase or 20S proteasome, consists of four leprae, is significantly more similar to the Thermoplasma seven-membered rings that contain fourteen related but subunit and may form a true proteasome. In this article, different subunits [11,12]. These fall into two families, we report the discovery, purification and characteriza- with oa-type subunits forming the outer rings and 13-type tion of a 20S proteasome complex from a nocardioform subunits the inner rings of the complex. actinomycete, Rhodococcus sp. strain NI86/21, that is closely related to Mycobacterium. This discovery extends Until their discovery in the archaebacterium Thermo- the occurrence of proteasomes to the third kingdom plasma acidophilum [13], proteasomes were thought to and shows that the proteasome is an ancestral particle of occur exclusively in eukaryotes. Following this discovery, universal distribution. searches were made for proteasomes in other archaebac- teria and in eubacteria without success [14], leading to speculation that a Thermoplasma-like organism may have Results been the precursor of the eukaryotic cytosol. Although structurally indistinguishable from eukaryotic protea- Cloning and sequencing of an operon containing somes at a resolution of 2 nm, the proteasome of Thermo- proteasome-related genes plasma is much simpler, being formed of only two During the analysis of the thc gene cluster of Rhodococcus subunits, a and [3 [13]. This simplicity has allowed the sp. strain NI86/21, which is required for degradation of determination of the structure of the Thermoplasma thiocarbamate herbicides and the s-triazine herbicide

*The first two authors contributed equally to this work. Correspondence to: W. Baumeister. E-mail address: [email protected]

766 © Current Biology 1995, Vol 5 No 7 Characterization of a eubacterial proteasome Tamura et al. RESEARCH PAPER 767 Characterization of a eubacterial proteasome Tamura et al. RESEARCH PAPER 767 atrazine [20,21], four open reading frames (ORFs) were chromatography on Sepharose 6B (Fig. 2a) and tested identified downstream of thcR (Fig. la). One of the the eluted fractions for Suc-Leu-Leu-Val-Tyr-AMC- encoded proteins, PrcB, is clearly related to proteasomal hydrolyzing activity in the presence and absence of 3-type subunits, and a second, PrcA, shows a similarity 0.05 % SDS. Two peaks of activity, eluting at apparent of unclear statistical significance to or-type subunits. molecular masses of 500-700 kD and 200-300 kD, were Database searches revealed a region with a strikingly sim- obtained in the absence of SDS. In the presence of SDS, ilar gene organization in the genome of the related the low molecular weight fraction lost all activity, whereas nocardioform actinomycete Mycobacterium leprae (Collab- the high molecular weight fraction showed no change of orative Research Inc., GenBank accession U00017; Fig. activity, indicating that it may contain proteasomes. We 1c). The similarity between the regions is highly signifi- cant at the protein level (61-81 % sequence identity), the main difference being a large amino-terminal deletion in Rhodococcus Orf6 relative to Mycobacterium C1_172. At the DNA level, similarities include a putative ribosome- found in front of the start codons of orf7 (AAGGAGG) and oX (AGGAGG), the unusual GUG start codon of prcB, and the apparent translational coupling of the of7-prcB-prcA and ofx-prcB-C3_260 genes, as inferred from the two-base overlaps of the stop and start codons of the gene pairs orf7/odX-prcB and prcB-prcA/C3_260 (data not shown).

Purification and biochemical characterization of a 20S proteasome complex In order to isolate the proteasome complex from Rhodococcus, we tested the hydrolyzing activity of crude cell-extract fractions using the substrate Suc-Leu-Leu- Val-Tyr-AMC, which is a typical fluorigenic substrate for chymotrypsin-like activity. Previous studies had shown that low concentrations of SDS do not affect the chymotryptic activity of proteasomes from Thermoplasma (unpublished data) and indeed activate the chymotryptic activity of most eukaryotic proteasomes [22,23]. We sub- jected crude cell extracts of Rhodococcus to gel filtration

Fig. 2. Identification of an SDS-resistant, high molecular weight Fig. 1. Gene organization of (a) the DNA region downstream of protease in Rhodococcus sp. strain N186/21. (a) Crude cell thcR in Rhodococcus sp. strain N186/21, (b) the homologous extracts (163 mg) were chromatographed on a Sepharose 6B col- region in the same strain, and (c) the equivalent region in umn as described in Materials and methods, and total protein Mycobacterium leprae cosmid B2126 (GenBank accession No. was detected by absorbance at 280 nm (red line). The column U00017). The black arrowheads indicate the position of a poten- fractions were assayed for the hydrolysis of the synthetic peptide tial transcription terminator (stem-loop structure with Suc-Leu-Leu-Val-Tyr-AMC in the presence (white circles) or AG =-160 kJ). The sequence of of61, up to the Sacl site, was absence (grey circles) of 0.05 % SDS. (b) Tricine-SDS-PAGE determined previously 20]. The hybridization probes used for analysis and (c) peptidase activity of fractions obtained by Super- identification of overlapping fragments are shown as black bars. ose 6 FPLC column chromatography. This column represents the One major scale division represents 500 bp. Only restriction sites last step in the purification procedure. The column was cali- mentioned in the text are shown. The sequences of the 3 751 bp brated with the markers thymoglobulin (669 kD), apoferritin BamHI-Bglll fragment (accession No. U26421) and the 3 554 bp (443 kD) and alcohol dehydrogenase (150 kD). In panel (b), a BstXI-Pstl fragment (accession No. U26422) have been submitted lane carrying Thermoplasma caand [3 proteins (T.A.) has been to GenBank. added for comparison. 768 Current Biology 1995, Vol 5 No 7

therefore subjected this fraction to further purification were determined by Edman degradation. Internal using a sequence of conventional chromatography steps. sequence information was obtained by digesting each band in situ using Lys-C, separating the products by high The putative Rlhodococcus proteasome was purified to over pressure liquid chromatography (HPLC), and subjecting 95 % homogeneity by the method outlined in Table 1. them to Edman degradation. The 31 kD and 29 kD bands The overall purification was approximately 440-fold, and were amino-terminally blocked. Two internal peptide the yield was 46 %. The native molecular mass of the parti- sequences from the 31 kD band were found to be identi- cle was estimated to be approximately 690 kD by fast pro- cal to sequences encoded by the prcA gene, whereas four tein liquid chromatography (FPLC) (Fig. 2c). The substrate internal sequences from the 29 kD band, covering about a specificity of the particle was similar to that of the archae- third of the protein, were found to be related but non- bacterial proteasome [24] and showed high activity against identical to the prcA gene product (Fig. 3). We concluded the chymotryptic substrates Suc-Leu-Leu-Val-Tyr-AMC, that the two proteins did not arise from one precursor by Z-Gly-Gly-Leu-AMC and Suc-Ala-Ala-Phe-AMC, posttranslational processing, but were the products of dif- but no tryptic or peptidyl-glutamyl-peptidase activities ferent genes. We also concluded that both proteins, which (Table 2). we named al (31 kD) and oa2 (29 kD), were a-type sub- units and that the 31 kD band corresponded to the prcA Tricine-SDS-polyacrylamide gel electrophoresis (Tricine- gene product. Thus the two 0a-type subunits of the SDS-PAGE) of the putative proteasome yielded four Rlhodococcus proteasome are similar to oa-type subunits bands of similar intensity with estimated molecular from eukaryotes and archaebacteria, which are also weights of 31 kD, 29 kD, 25 kD and 24 kD (Fig. 2b). blocked at their amino termini. This was a surprising result, as the Thernioplasnma protea- some is formed by only two subunits ( and A), whereas The two other bands had amino-terminal sequences of eukaryotic proteasomes are formed by fourteen subunits TTIVAISY (25 kD band) and TTIVALTYKG (24 kD (seven of the oa-type and seven of the P-type). The four band). The amino-terminal sequence of the 24 kD band bands were blotted and their amino-terminal sequences corresponded to a sequence encoded by the prcB gene, whereas the amino-terminal sequence of the 25 kD band, as well as three other internal sequences, were similar, but not identical, to the prcB gene product (Fig. 3). As in the case of the two o-type subunits, we concluded that the two proteins were encoded by different genes, that they were both P-type subunits and that the 24 kD protein corresponded to the prcB gene product. In addition, we also concluded that 11 and B2 contain a pro-peptide that is cleaved during the assembly of the proteasome. The proteolytically active P-type subunits of eukaryotes and archaebacteria also contain a pro-peptide that is cleaved at a conserved threonine. This threonine is the catalytically active nucleophile and its free amino group may be important for as a proton acceptor/donor [16,18].

In an immunological cross-reactivity test of the Rllodococcus proteasome subunits with antibodies raised against the ao subunit, the e subunit or the entire proteasome complex from Tlermoplasma, no cross-reactivity was obtained as judged by western immunoblotting after Tricine-SDS- PAGE (data not shown). This result is not unexpected Characterization of a eubacterial proteasome Tamura et al. RESEARCH PAPER 769

Fig. 3. Aligned sequences of the Rhodococcus proteasome subunits. Peptides sequenced by Edman degradation are indicated by lines above or below the sequences. The arrow shows the site of proteolytic processing in the a subunits.

given the large sequence divergence between the protea- some subunits of Rhodococcus and Thermoplasma.

Cloning and sequencing of a second operon containing proteasome-related genes Following the discovery of two a and two , subunits in Rhodococcus, we searched for a second operon of protea- some-related genes by hybridization of total DNA prepa- rations with an internal prcA probe. We identified a second gene region (Fig. lb) with an organization stri- kingly similar to the regions identified previously in Rhodococcus (Fig. la) and Mycobacterium (Fig. 1c). We named the genes in this region, which code for subunits cx2 and 2, orf62 -otf72-prcB2 -prcA2 , and renamed the two prc genes of the initially identified region prcA and prcB l, as they code for subunits cxl and 31. We were sur- prised to find that the homologous proteins in Rhodococcus are all more similar to each other than to the equivalent proteins in Mycobacterium (Fig. 4a), at both the protein and the DNA level. This indicates that the gene duplica- tion in Rhodococcus occurred very recently on an evolu- tionary timescale, and that it is unlikely, therefore, to represent a general feature of eubacterial proteasomes. The second region of Rhodococcus proteasome genes had a GC content (63 %) intermediate between that of the first region (67 %) and of the region from Mycobacterium (59 %), and an orf6-homologue (orf62) of the same length as Mycobacterium C1_172. Further similarities to the two other proteasome gene regions included a putative ribosome-binding site in front of orf72 , a GUG start codon for prcB2, and two-base overlaps in the coding regions of orf72, prcB2 and prcA2.

Electron microscopy of the 20S complex Electron micrographs of negatively stained proteasome preparations revealed the two views characteristic of archaebacterial and eukaryotic proteasomes: ring-shaped Fig. 4. (a) Dendrograms showing the relatedness of Rhodococcus and4(R) Myobatenum S(Ml) proteins in % sequence identity. end-on views approximately 11 nm in diameter, represent- (b) Dendrogram of proteasome sequences derived from the ing projections along the cylinder axis, and rectangular alignment in Fig. 6. side-views - projections perpendicular to the cylinder axis 770 Current Biology 1995, Vol 5 No 7

Fig. 5. Electron micrograph of protea- somes from Rhodococcus sp. strain N186/21. The inset shows an averaged side view.

(Fig. 5). Correlation averages of the latter show that, as in eukaryotes. The second very highly conserved region in Thermoplasma, the Rhodococcus proteasome is a stack of four a-type subunits (residues 130-143) contains an invariant rings collectively forming a barrel-shaped structure (Fig. RPXG motif (in which X represents a large hydrophobic 5, inset). Because of the difficulties in assigning symmetry residue). This sequence is part of a loop that constricts based on projection images of large negatively stained the central pore through which unfolded polypeptides molecules [25], we are unable at this point to determine penetrate into the inner proteolytic compartment whether the rings are composed of six or seven members. [16,26]. In Rhodococcus and Mycobacterium, the equivalent sequence is KPYE and the loop is similar in size to other a subunits, indicating that the pore has similar properties Discussion and dimensions.

Sequence analysis and phylogeny In [3-type subunits, several residues critical to the shape Despite their significant divergence, the subunits of the and nature of the are conserved between 31, eubacterial proteasome are clearly related to or-type and 132 and other proteasome subunits. Most important is 3-type proteasomal subunits. Together with PrcB of residue Thrl of mature subunits, which provides the cat- Mycobacterium, 31 and 32 can be assigned unambiguously alytic nucleophile [16,18]. The amino-terminal group of to the family of 13-type subunits on the basis of sequence this residue is freed by cleavage of a pro-peptide in three similarity. The similarity of aol, a2 and Mycobacterium of the seven eukaryotic 3-subunit types and in Thermo- C3_260 to at-type subunits is more difficult to establish plasma 13 during assembly of the complex. The four from sequence data, and only two sequence blocks, other eukaryotic 3-subunit types do not contain this corresponding to residues 23-52 and 130-143 in Thermo- threonine and/or are not cleaved at this site, indicating plasma or, can be aligned significantly with other a sub- that they are proteolytically inactive. Both Rhodococcus units. Nevertheless, the biochemical data and the 3-type subunits are cleaved at this internal threonine, occurrence of a few highly conserved residues indicate resulting in the release of pro-peptides of 65 (31) and 59 that they are true a subunits and will assume the same (P32) residues. The presumed role of the free amino basic fold as all other a-type and 3-type subunits. Based terminus of Thrl is to provide a proton acceptor/donor on these highly conserved residues and on the predicted secondary structure, we have constructed the alignment Fig. 6. (opposite) Multiple alignment of proteasome sequences presented in Figure 6. The first region of significant simi- from eukaryotes (Hs, human; Rn, rat), archaebacteria (Ta, Ther- larity in a-type subunits (residues 23-52) encompasses a moplasma acidophilum) and eubacteria (R, Rhodococcus sp.; MI, Mycobacterium leprae; Ec, Escherichia coli; Ph, Pasteurella helix that is highly conserved between eukaryotic and haemolytica; Bs, Bacillus subtilis; Ae, Alcaligenes eutrophus). The Thermoplasma at sequences, but is less conserved in upper part of the alignment shows a-type subunits and the lower eubacterial subunits. This helix lies on the top outer sur- part 1-type subunits. Eukaryotic members are represented by face of a rings [16] and probably interacts with regula- human sequences where known; the 3-type sequences are shown tory factors such as the 19S cap (PA700) or the 11S in their mature form (where known) and include the MHC- encoded inducible subunits LMP2 and LMP7 which can replace regulator (PA28); its conservation indicates that these the constitutive subunits and upon y-interferon stimulation. factors may be equally universal in occurrence despite The numbering and secondary structure (H, -helix; S, 3-strand) the fact that they have only been detected, so far, in are those of the Thermoplasma proteasome [16]. Characterization of a eubacterial proteasome Tamura et al. RESEARCH PAPER 771 772 Current Biology 1995, Vol 5 No 7

function for the initial nucleophilic attack, analogous to Inclusion of this sequence shifted PrcB out of the the amino terminus of the catalytic serine in penicillin eubacterial HslV branch and moved this branch closer acylase [27]. to the root. Evidently, this tree must be viewed as provi- sional and likely to change in its bacterial branches with A second highly conserved residue in the active site is the addition of further sequences. Only minor changes Lys33, which is conserved throughout all catalytically are likely to occur in the eukaryotic branches. active subunits, including all eubacterial 3-type subunits. Lys33 forms a salt-bridge to 3Asp17 [16], and this Models for the oligomeric structure of the Rhodococcus residue is equally conserved in all eubacterial 3-type sub- proteasome units. Finally, two glycine-rich loops (GSG, residues Several models can be imagined that account for the 130-132, and SGG, residues 170-172) that are conserved increased subunit complexity of the eubacterial protea- in all active [3-type subunits (but not in inactive 3-type some relative to the archaebacterial one. The simplest subunits) and that frame the active site [16] are also con- model is that of two proteasome populations in Rhodococcus served in 31,.132 and PrcB as GSG and TGG. - one formed by oil and 31, the other by 2 and 132 (Fig. 7a) - although it would then be a surprising coinci- A distance-based dendrogram of proteasome sequences dence that we would obtain approximately equal (Fig. 4b) shows that the 13subunits of Rhodococcus and of amounts of both complexes during purification. Two Mycobacterium form a common branch that originates other models incorporate equimolar amounts of each close to the root of 1-type sequences. The two proteins subunit into the complex, either by proposing four dif- are clearly more closely related to the active than to the ferent homo-oligomeric rings per particle (Fig. 7b), or inactive subunits of eukaryotes. As discussed previously two hetero-oligomeric rings of each type (Fig. 7c). The [12], the shape of the dendrogram is dependent on details latter model requires an even number of subunits per of the alignment, on the choice of scoring matrix and on ring. Although at this point we cannot distinguish other evolutionary assumptions. The main difference between these models, experiments are in preparation to between this tree and similar trees published previously differentiate between them. The Rhodococcus proteasome [12,19] is the fact that PrcB was considered part of the may develop into an extremely useful research tool, eubacterial HslV branch prior to the inclusion of a fairly should it prove to have a complexity intermediate divergent ORF sequence from Alcaligenes eutrophus. between the archaebacterial and eukaryotic complexes.

Fig. 7. Models for the organization of the Rhodococcus proteasome. Model (a) assumes two distinct proteasome populations; model (b) assumes one proteasome population formed by four distinct homo-oligomeric rings. The dotted line indicates our inability to determine whether the rings have six or seven members. Model (c) assumes one proteasome population formed by a and 13rings with alternating subunits, and only appears likely in the case of six subunits per ring. Characterization of a eubacterial proteasome Tamura et al. RESEARCH PAPER 773

Conclusions using an automated sequencer (A.L.F., Pharmacia). The ASSEMGEL program of PCGENE (IntelliGenetics) was used The ubiquitin pathway of protein degradation, in which to assemble the subsequences. Potential coding regions were identified with the programs GCWIND [30] and FRAME [31]. the proteasome functions as the central protease, was orig- inally believed to occur only in eukaryotes. More recently, proteasomes and ubiquitin were also found in the archae- Purification of the Rhodococcus proteasome All purification steps were performed at 4 C. Harvested cells bacterium Thermoplasma acidophilum [13,28]. Following (5-6 g per preparation) were washed twice with homogenate the identification of eubacterial sequences significantly buffer (50 mM Hepes, 1 mM DTT, pH 8.0), resuspended with related to proteasomes in GenBank [19], we have now 2 vol (mass/vol) of the same buffer and sonicated. The isolated the first proteasome from a eubacterium. homogenate was centrifuged for 1 h at 61 700 xg. Proteolytic activity was tested using the synthetic peptide Suc-Leu- The data presented in this article show that the high- Leu-Val-Tyr-AMC as a substrate. The 61 700 x g supernatant molecular weight protease which we identified in was loaded onto a Sepharose 6B column (3 x 90 cm; Pharma- Rhodococcus sp. strain NI86/21 is a true 20S proteasome. cia) equilibrated with buffer A (50 mM Tris-HC1, 1 mM DTT, The apparent molecular mass of the particle (690 kD) is pH 7.0). Fractions of 5 ml were collected and assayed for pro- comparable to that of the archaebacterial proteasome teolytic activity. The active, high-molecular mass fractions (500-700 kD) were pooled and loaded onto a DEAE-Sephacel (700 kD), as are its insensitivity to 0.05 % SDS and its column (2 x 10 cm; Pharmacia) equilibrated with buffer A. chymotryptic substrate specificity. In electron micro- Bound proteins were eluted using a 0-400 mM NaC1 linear graphs, the particle has the same dimensions and the gradient in buffer A. Fractions of 3 ml were collected and same four-ringed, cylinder-shaped appearance as the assayed for proteolytic activity. The active fractions, obtained archaebacterial and eukaryotic proteasomes. Finally, the at approximately 250 mM NaC1, were pooled and dialyzed sequences of its subunits are significantly related to those against 10 mM potassium phosphate buffer, 1 mM DTT, pH of ot-type and [3-type proteasome subunits. In conjunc- 7.0. The dialyzed sample was applied to a hydroxylapatite col- tion with the recent description of ubiquitin in the umn (1.4 x 6 cm; Bio-Rad), equilibrated with 10 mM potas- eubacterium Anabaena variabilis [29], this discovery shows sium phosphate buffer. Bound proteins were eluted using a that the ubiquitin pathway is ancestral and common to all 10-300 mM phosphate linear gradient. Fractions of 1.5 ml forms of life. were collected and assayed for proteolytic activity. The active fractions, obtained at approximately 60 mM phosphate, were pooled and dialyzed against buffer A. The dialyzed sample was Materials and methods loaded onto a Mono-Q column (5 mm x 5 cm; Pharmacia) equilibrated with buffer A. Bound proteins were eluted using a Strains, media, and culture conditions 0-500 mM NaCl linear gradient in buffer A. Fractions of 0.6 ml were collected and assayed for proteolytic activity. The Rhodococcus sp. strain N86/21 was obtained from the active fractions, obtained at approximately 350 mM NaCl, National Collection of Agricultural and Industrial Micro- were pooled and concentrated by centrifugation in a Centri- organisms (NCAIM, Budapest, Hungary) and grown in Luria sample was loaded onto a broth at 30 C. con-30 (Amicon). The concentrated Superose 6 HR column (1 x 30 cm, Pharmacia) equilibrated with 25 mM Tris-HCI, 1 mM DTT, pH 7.5. Fractions of Cloning and sequence analysis of proteasome structural 0.5 ml were collected and assayed for peptidase activity. The genes protein content of the fractions was analyzed by 16.5 % We have reported previously the cloning and partial sequence Tricine-SDS-PAGE [32]. of a 14 kb Sad DNA fragment of Rhodococcus sp. strain NI86/21 in LambdaGEM-12 (FAJ2028) [20]. This fragment carries the thcRBCD gene cluster involved in herbicide degradation. In Enzyme assays order to characterize the DNA region downstream of thcR, an Fluorigenic substrate peptides were obtained from Bachem. overlapping 2.8 kb BamHI fragment was identified by Southern For peptidase assays, 10 nmoles of peptide (final concentration hybridization using the 306 bp PstI-Sad fragment of the 100 p1M) were incubated with the enzyme sample for XFAJ2028 insert as a probe. Subsequently, a pUC18 clone with 15-60 min at 37 °C in 50 mM Tris-HCI buffer (pH 8.0) in the presence or absence of 0.05 % SDS, as described [33]. The final this BamHI fragment (pFAJ2291) was isolated by colony 1 hybridization from an enriched library containing BamHI frag- reaction volume was 100 p l. The reaction was stopped by 1 ments of the estimated size. The region further downstream was adding 100 pI 10 % SDS, and the fluorescence of the reaction cloned into pUC18 as a 1.6 kb BgIII fragment overlapping with products was measured. the aforementioned BamHI fragment. This clone (pFAJ2406) was identified using the 357 bp SmaI-BamHI fragment of the Amino-acid sequence analysis pFAJ2291 insert as a hybridization probe. Using the same Purified proteasomes were subjected to Tricine-SDS-PAGE, probe, a second hybridizing band (BglII fragment of about electroblotted onto a siliconized glass-fiber sheet (Glassybond, 5.7 kb) was detected by Southern hybridization and cloned into Biometra, G6ttingen, Germany), detected by Coomassie blue pUC18 (pFAJ2450). The 595 bp SmaI-BglII fragment from staining and either subjected directly to Edman degradation pFAJ2450 was used as a probe in plaque hybridization to screen [34,35] or digested in situ using Lys-C prior to degradation in an EMBL3 library of Sau3A-digested genomic DNA. This order to obtain internal sequences. A second set of internal allowed the identification of an EMBL3 clone (FAJ2029) with sequences was obtained by subjecting purified proteasomes to a DNA insert spanning the BglII fragment of pFAJ2450 and Tricine-SDS-PAGE, staining, excising the bands, digesting in flanking DNA regions. DNA sequencing of both strands on situ with Lys-C and separating the resulting peptides by HPLC overlapping fragments subcloned in pUC18 was carried out prior to Edman degradation [36]. 774 Current Biology 1995, Vol 5 No 7

Sequence alignment and phylogeny structure of the 20S proteasome from the archaeon T. acidophilum were extracted from GenBank using ENTREZ at 3.4 A resolution. Science 1995, 268:533-539. Sequences 17. Seemuller E, Lupas A, Zohl F, Zwickl P, Baumeister W: The protea- (NCBI, Bethesda MD, USA) and aligned using PileUp (GCG some from Thermoplasma acidophilum is neither a cysteine nor a Package, Genetics Computer Group, Madison WI, USA) and . FEBS Lett 1995, 359:173-178. MACAW [37]. Final adjustments to the alignment were made 18. SeemLiller E, Lupas A, Stock D, Liiwe J, Huber R, Baumeister W: manually in MALIGNED [38] using the known or predicted Proteasome from Thermoplasma acidophilum: a threonine protease. Science 1995, 268:579-581. secondary structure of the subunits. Secondary structure 19. Lupas A, Zwickl P, Baumeister W: Proteasome sequences in predictions were obtained by submitting pre-aligned sequences eubacteria. Trends Biochem Sci 1994, 19:533-534. to the PHD server ([email protected]) [39]. 20. Nagy I, Schoofs G, Compernolle F, Proost P, Vanderleyden J, De Dendrograms were computed using the DARWIN server Mot R: Degradation of the thiocarbamate herbicide EPTC (S-ethyl dipropylcarbamothioate) and biosafening by Rhodococcus sp. ([email protected]) [40] and the FITCH module of PHYLIP strain N186/21 involve an inducible cytochrome P-450 system and 3.5p (distributed by the author, J. Felsenstein, joe@genetics. aldehyde dehydrogenase. J Bacteriol 1995, 177:676-687. washington.edu). 21. Nagy I, Compernolle F, Ghys K, Vanderleyden J, De Mot R: A single cytochrome P-450 system is involved in degradation of the herbicides EPTC (S-ethyl dipropylcarbamothioate) and atrazine by Electron microscopy Rhodococcus sp. strain N186/21. Appl Environ Microbiol 1995, Purified proteasomes were stained with uranyl acetate. 61:2056-2060. Conditions for recording electron micrographs and image 22. Wilk S, Orlowski M: Evidence that pituitary cation-sensitive neutral analysis were as detailed in [41]. is a multicatalytic protease complex. I Neurochem 1983, 40:842-849. 23. Dahlmann B, Rutschmann M, Kuehn L, Reinauer H: Activation of Acknowledgements: The authors thank P. De Neve for cloning of the multicatalytic proteinase from rat skeletal muscle by fatty acids pFAJ2406, J. Vanderleyden for stimulating discussions, D. Shields or sodium dodecyl sulphate. Biochem J 1985, 228:171-177. (Trinity College, Dublin, Ireland) for the GCWIND program, and 24. Dahlmann B, Kuehn L, Grziwa A, Zwickl P, Baumeister W: G. Kleman (Ohio State University, USA) for the FRAME pro- Biochemical properties of the proteasome from Thermoplasma gram. T.T. acknowledges a Postdoctoral Fellowship for Research acidophilum. EurJ Biochem 1992, 208:789-797. 25. Marco S, Urena D, Carrascosa JL, Waldmann T, Peters J, Hegerl R, et Abroad from the Japan Society for the Promotion of Science. al.: The molecular chaperone TF55: assessment of symmetry. FEBS R.D.M is a Senior Research Associate with the National Fund for Lett 1994, 341:152-155. Scientific Research (Belgium). 26. Wenzel T, Baumeister W: Conformational constraints in protein degradation by the 20S proteasome. Nature Struct Biol 1995, 2:199-204. References 27. Duggleby HJ, Tolley SP, Hill CP, Dodson EJ,Dodson G, Moody PCE: Penicillin acylase has a single-amino-acid catalytic center. Nature 1. Waxman L, Fagan J, Goldberg AL: Demonstration of two distinct high 1995, 373:264-268. molecular weight proteases in rabbit reticulocytes, one of which 28. Wolf S, Lottspeich F, Baumeister W: Ubiquitin found in the archae- degrades ubiquitin conjugates. J Biol Chem 1987, 262:2451-2457. bacterium Thermoplasma acidophilum. FEBS Lett 1993, 326:42-44. 2. Hough R, Pratt G, Rechsteiner M: Purification of two high molecu- 29. Durner J, Boger P: Ubiquitin in the prokaryote Anabaena variabilis. lar weight proteases from rabbit reticulocyte lysate. Biol Chem J Biol Chem 1995, 270:3720-3725. 1987, 262:8303-8313. 30. Shields DC, Higgins DG, Sharp PM: GCWIND: a microcomputer 3. Ganoth O, Leshinsky E, Eytan E, Hershko A: A multicomponent program for identifying open reading frames according to codon system that degrades proteins conjugated to ubiquitin. Resolution positional G+C content. Comput Appl Biosci 1992, 8:521-523. of factors and evidence for ATP-dependent complex formation. 31. Bibb MJ, Findlay PR, Johnson MW: The relationship between base Biol Chem 1988, 263:12412-12419. composition and codon usage in bacterial genes and its use for the 4. Hershko A, Ciechanover A: The ubiquitin system for protein simple and reliable identification of protein coding sequences. degradation. Annu Rev Biochem 1992, 61:761-807. Gene 1984, 30:157-166. 5. Glotzer M, Murray AW, Kirschner MW: Cyclin is degraded by the 32. Schagger H, von Jagow G: Tricine-sodium dodecyl sulfate- ubiquitin pathway. Nature 1991, 349:132-138. polyacrylamide gel electrophoresis for the separation of proteins in 6. Ghislain M, Udvardy A, Mann C: Saccharomyces cerevisiae 26S the range from 1 to 100 kDa. Anal Biochem 1987, 166:368-379. protease mutants arrest cell-division in G2/metaphase. Nature 33. Tanaka K, Yoshimura T, Ichihara A, Ikai A, Nishigai M, Morimoto M, 1993, 366:358-362. et al.: Molecular organization of a high molecular weight multi- 7. Dubiel W, Ferrell K, Rechsteiner M: Peptide sequencing identifies protease complex from rat liver. J Mol Biol 1988, 203:985-996. MSS1, a modulator of HIV Tat-mediated transactivation, as subunit 34. Eckerskorn C, Mewes W, Goretzki H, Lottspeich F: A new 7 of the 26S protease. FEBS Lett 1993, 323:276-278. siliconized-glass fiber as support for protein-chemical analysis of 8. DeMartino GN, Moomaw CR, Zagnitko OP, Proske RJ, Chu-Ping M, electro-blotted proteins. Eur] Biochem 1988, 176:509-519. Afendis SJ, et al.: PA700, an ATP-dependent activator of the 20S pro- 35. Eckerskorn C, Jungblut P, Mewes W, Klose J, Lottspeich F: Identifi- teasome, is an ATPase containing multiple members of a nucleotide- cation of mouse brain proteins after two-dimensional electro- binding . J Biol Chem 1994, 269:20878-20884. phoresis and electroblotting by microsequence analysis and amino 9. Gaczynska M, Rock KL, Goldberg AL: Role of proteasomes in acid composition analysis. Electrophoresis 1988, 9:830-838. antigen presentation. Enzyme Protein 1993, 47:354-369. 36. Eckerskorn C, Lottspeich F: Internal amino acid sequence analysis of 10. Peters JM, Cejka Z, Harris JR, Kleinschmidt JA, Baumeister W: proteins separated by gel electrophoresis after tryptic digestion in Structural features of the 265 proteasome complex. J Mol Biol polyacrylamide matrix. Chromatographia 1989, 28:92-94. 1993, 234:932-937. 37. Schuler GD, Altschul SF, Lipman DJ: A workbench for multiple 11. Tanahashi N, Tsurumi C, Tamura T, Tanaka K: Molecular structures alignment construction and analysis. Proteins 1991, 9:180-190. of 205 and 26S proteasomes. Enzyme Protein 1993, 47:241-251. 38. Clark SP: MALIGNED: a multiple sequence alignment editor. Corn- 12. Lupas A, Koster AJ, Baumeister W: Structural features of 265 and put App/ Biosci 1992, 8:535-538. 20S proteasomes. Enzyme Protein 1993, 47:252-273. 39. Rost B, Sander C, Schneider R: PHD - an automatic mail server for 13. Dahlmann B, Kopp F, Kuehn L, Niedel B, Pfeifer G, Hegerl R, protein secondary structure prediction. Comput Appl Biosci 1994, Baumeister W: The multicatalytic proteinase (prosome) is ubiqui- 10:53-60. tous from eukaryotes to archaebacteria. FEBS Lett 1989, 40. Benner SA, Badcoe I, Cohen MA, Gerloff DL: Bona fide prediction 251:125-131. of aspects of protein conformation. Mol Biol 1994, 235: 14. Puhler G, Pitzer F, Zwickl P, Baumeister W: Proteasomes: multi- 926-958. subunit proteases common to Thermoplasma and eukaryotes. 41. Baumeister W, Dahlmann B, Hegerl R, Kopp F, Kuehn L, Pfeifer G: Systematic Appl Microbiol 1994, 16:734-741. Electron microscopy and image analysis of the multicatalytic 15. Jap B, Phler G, Lcke H, Typke D, Lwe J, Stock D, et al.: proteinase. FEBS Lett 1988, 241:239-245. Preliminary X-ray crystallographic study of the proteasome from Thermoplasma acidophilum. J Mol Biol 1993, 234:881-884. Received: 10 April 1995; revised: 12 May 1995. 16. LOwe J, Stock D, Jap B, Zwickl P, Baumeister W, Huber R: Crystal Accepted: 12 May 1995.