hUmp1 and its subcellular localisation

International Journal of Biological Macromolecules 38 (2006) 259–267

Possible tetramerisation of the proteasome maturation factor POMP/proteassemblin/hUmp1 and its subcellular localisation夽 Melanie M. Hoefer a , Eva-Maria Boneberg a , Stefan Grotegut b , Justine Kusch c , Harald Illges a,d,∗ a Biotechnology Institute Thurgau, Taegerwilen, Switzerland Institute for Biochemistry and Genetics, Department of Clinical and Biological Sciences, University of Basel, Basel, Switzerland c Institute of Biochemistry, Swiss Federal Institute of Technology (ETH) Zurich, Zurich, Switzerland d University of Applied Sciences, Department of Natural Sciences, Immunology and Cell Biology, von-Liebig-Str. 20, 53359 Rheinbach, Germany b

Received 30 November 2005; received in revised form 22 February 2006; accepted 6 March 2006 Available online 16 March 2006

Abstract The proteasome is a multisubunit complex with a central role in non-lysosomal proteolysis and the processing of proteins for presentation by the MHC class I pathway. The 16 kDa proteasome maturation protein POMP (also named proteassemblin or hUmp1) acts as a chaperone and is essential for the maturation of the 20S proteasome proteolytic core complex. However, the exact mechanism, timing and localisation of mammalian proteasome assembly remains elusive. We sought to investigate the localisation of POMP within the cell and therefore purified the protein and produced a polyclonal antibody. For immunisation, POMP was overexpressed and purified from a bacterial GST-system. Interestingly, after removal of the GST-tag, POMP was hardly detectable by Coomassie blue- and Ponceau red-staining. However, with a reverse zinc-staining, the protein could easily be visualised. POMP was gel-filtrated and eluted from a calibrated chromatography column with an apparent molecular weight of approximately 64 kDa, suggesting that it forms tetramers. Moreover, localisation studies by immunofluorescence stainings and confocal microscopy revealed that POMP is present in the cytoplasm as well as in the nucleus. © 2006 Elsevier B.V. All rights reserved. Keywords: Confocal microscopy; Proteasome maturation; Tetramerisation

1. Introduction The 26S proteasome is a large (∼2–2.5 MDa) intracellular adenosine 5 -triphosphate-dependent protease complex that identifies and degrades proteins tagged for destruction by the ubiquitin system. By generating peptides from intracellular antigens which are then presented to T-cells, the ubiquitin/proteasome system also plays a central role through the MHC class I pathway in the cellular immune response [1–4]. The 26S proteasome is composed of two types of complexes, the 20S proteolytically active core particle and the flanking two 19S regulatory particles. Four stacked heptameric rings of ␣ and ␤ subunits form the cylindrical 20S core complex arranged in a ␣7␤7␤7␣7 夽

This paper was presented at the Challenging Proteins Workshop in Paris, October 17–18, 2005. ∗ Corresponding author. Tel.: +49 2241 865570; fax: +49 2241 8658570. E-mail addresses: [email protected] (M.M. Hoefer), [email protected] (H. Illges). 0141-8130/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ijbiomac.2006.03.015

pattern. The 20S subcomplex may also be found unaccompanied by the 19S subcomplex or in association with alternative regulatory particles [3]. In mammalian cells 20S proteasomes can be present either as standard (or constitutive) 20S proteasomes, or as immunoproteasomes. The formation of immunoproteasomes is induced by IFN-␥ or TNF-␣ where ␤1i/LMP2, ␤2i/MECL1 and ␤5i/LMP7 replace their constitutive homologues ␤1, ␤2 and ␤5 [5–7]. For both types of proteasomes the proteasome maturation protein (POMP, also termed proteassemblin or hUmp1) is required for proper maturation of the 20S complex [8–10]. It acts as a chaperone and assists to assemble the two ␣7␤7 half proteasomes (13-16S precursor) and thereafter is itself degraded by the mature proteasome [3,6]. This activity was originally described in yeast with the POMP-homologue Ump1 [11]. Proteasome maturation is an essential process as revealed by RNAi studies on Ump1 where increased protein oxidation and cell death occurred rapidly after RNAi induction [12]. In mammalian cells, prolonged knock-down of POMP through siRNA caused induction of apoptosis [13].

260

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

The fact that cells rely on functional proteasomes and that transformed cells display greater susceptibility to proteasome inhibition than non-malignant cells has been used to develop new anti-cancer drugs. Proteasome inhibitors are already approved for treatment of recurrent multiple myeloma and are in phase II of clinical trials for a variety of B-lymphocyte malignancies including relapsed or refractory non-Hodgkin’s lymphomas [14–17]. Proteasomes are ubiquitous and abundant in various cell types. However, localisation seems to differ. In yeast it has been shown that about 80% of proteasome complexes reside within the nucleus [18]. Ump1-GFP localises mainly to the nucleus and around the nuclear envelope/rough endoplasmic reticulum (ER). These data, and the putative nuclear localisation sequences (cNLS) present in the proteasomal ␣ subunits led to the conclusion, that 13-16S precursor complexes are translocated together with Ump1 via the importin/karyopherin ␣␤ docking receptor to the nucleus where they mature [19]. In mammalian cells, the situation appears to be more complex. Here, proteasomes are localised throughout the cell; however, the relative abundance in cytoplasm and nucleus is highly variable. Although proteasomes diffuse freely within the cytoplasm, they are also found associated with centrosomes, cytoskeletal networks and the outer surface of the ER. In the nucleus, proteasomes are present throughout the nucleoplasm, however are sometimes also found to be associated with discrete subnuclear domains called PML (promyelocytic leukaemia oncoprotein) nuclear bodies (PMLNBs, also named ND10 bodies, POD domains, nuclear bodies) [20]. PML-NBs have been described to be enriched in components of the proteasomal system and represent proteolytic centres [21,22]. In this work, we describe the expression and purification of the human proteasome maturation protein POMP, which reveals unusual biochemical characteristics and staining properties. Furthermore, we depict its localisation within the cell and report that POMP can form tetramers.

(HSPC036), that was described as hUmp1 [8] and POMP [10]. The cDNA was cloned into the pGEX-4T-3 bacterial expression vector (Amersham Biosciences, Freiburg, Germany) using the following primers for PCR amplification: TG-104 (BamHI): 5 GCG GGA TCC CGG ATG AAT GCC AGA GGA CTT GGA-3 and TG-105 (XhoI): 5 -CCG CTC GAG CCT ATT ACA GTA AAC CAA GTT TAT ATT CCA CCA T-3 . The glutathione S-transferase (GST)-POMP fusion protein was expressed in E. coli BL21 and was partly soluble and partly insoluble in inclusion bodies. Therefore, the protein was purified either from the supernatant, and/or the pellet was dissolved in a small volume of urea buffer (8 M urea, 100 mM NaH2 PO4 , 10 mM Tris–HCl, pH 8.0) and quickly injected into a large volume of PBS supplemented with complete protease inhibitor cocktail (Roche, Rotkreuz, Switzerland), and subsequently purified over a glutathione sepharose column according to the manufacturer’s protocol (GSTrap Fast Flow column or batch purification with glutathione sepharose 4 Fast Flow). The GST-tag was removed by on-column digestion with thrombin (Amersham Biosciences), and subsequently five POMP containing fractions (1 ml each) were eluted with 50 mM Tris–HCl, pH 8.0. Afterwards, the GST-tag was eluted from the column in 5× 1 ml fractions using the same buffer as above but containing 10 mM reduced l-glutathione (Sigma, Buchs, Switzerland). Fraction numbers indicated in the figure legends refer to the respective elution step (i.e. eluate 1 (POMP) corresponds to the first ml eluted from the column after thrombin cleavage). Removal of the GST-tag through thrombin added three extra amino acids (Gly-Ser-Arg) N-terminal to the POMP protein. The resulting 16 kDa POMP protein was used for immunisation of a Chinchilla Bastard rabbit (Charles River, Sulzfeld, Germany). From the rabbit serum polyclonal antibody was purified over a protein G column (HiTrap G HP, Amersham Biosciences).

2. Materials and methods

The AF262975 sequence was cloned into the pQE-32 expression vector (Qiagen, Basel, Switzerland), adding a 6× histidinetag to the N-terminus of the POMP sequence. The following primers were used for amplification: TG-145 (BamHI): 5 -GCG GGA TCC CGA TGA ATG CCA GAG GAC TTG GA-3 and TG-146 (PstI): 5 -CCG CTG CAG CCT ATT ACA GTA AAC CAA GTT TAT ATT CCA CCA T-3 . 6×His-POMP was purified from E. coli SG13009[pREP4] under denaturing conditions with Ni-NTA agarose (both Qiagen). All buffers used were 8 M urea, 100 mM NaH2 PO4 , 10 mM Tris–HCl at different pH; pH 8.0 for lysis, pH 6.3 for washing, pH 5.9 for the first elution, pH 4.5 for the second elution step. For washing, 5× 2 ml fractions were collected, and for elution steps one and two, 7× 1 ml fractions were saved. Fraction numbers indicated in the figure legends refer to the respective washing or elution step (e.g. wash 1 corresponds to the first 2 ml collected from the washing step).

2.1. Cell lines Daudi, Raji (B-cells) and human embryonic kidney (HEK) 293 cells were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM) supplemented with 1% UltraGlutamineI, penicillin/streptomycin (100 U/ml/100 ␮g/ml) (all Cambrex, Verviers, Belgium) and 10% fetal calf serum (FCS) (Linaris, Wertheim-Bettingen, Germany). Cells were maintained at 37 ◦ C with 7.5% CO2 . 2.2. Purification of recombinant POMP and production of a polyclonal antibody against POMP The following POMP sequence representing the full length mRNA was used for cloning: GenBank accession number AF262975 (Homo sapiens voltage-gated potassium channel ␤ subunit 4.1, K+ v ␤4.1/KCNA4B [23]), which is identical within the coding sequence to AF077200 (HSPC014), that was described as proteassemblin [9], as well as to AF125097

2.3. 6×His-POMP purification

2.4. Gel-filtration chromatography of POMP Gel-filtration chromatography of POMP (after removal of the GST-tag by thrombin cleavage) was performed on a Superdex

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

261

75 pg HiLoad 16/60 column (Amersham Biosciences) equilibrated in TN buffer (50 mM Tris–HCl, 150 mM NaCl, pH ¨ 8.0). The column was run with an AKTAPrime chromatography system (Amersham Biosciences) at 0.8 ml/min at RT. Calibration proteins were the cytoplasmic domain of human CD21 (CD21cyto, 4.4 kDa, M. Hoefer, manuscript in preparation), ribonuclease A (13.7 kDa) and ovalbumin (43 kDa) (LMW gelfiltration calibration kit, Amersham Biosciences). Eluted fractions were collected and analysed on a 4–12% BisTris NuPage SDS gel (Invitrogen, Basel, Switzerland) under reducing conditions and blotted onto Protran nitrocellulose membranes (Schleicher & Schuell GmbH, Dassel, Germany). Immunoblots were developed as described below. 2.5. Immunoprecipitation, SDS-PAGE and immunoblotting Cells were collected by centrifugation and lysed in PBS supplemented with complete protease inhibitor cocktail by one freeze–thaw cycle. After centrifugation, the supernatant was used as cytosolic extract and the pellet was lysed for 30 min on ice in lysis buffer (1% NP-40, 50 mM NaCl, 50 mM NaF, 30 mM Na4 P2 O7 , 5 mM EDTA, 1 mM Na3 VO4 and complete protease inhibitor cocktail). After centrifugation, the supernatant was used as membrane enriched lysate. Immunoprecipitations were performed with the polyclonal POMP antibody and Protein A Sepharose (Sigma). Lysates or immunoprecipitations were then loaded on 12% or 4–12% Bis/Tris NuPAGE SDS gels (Invitrogen) and blotted onto Protran nitrocellulose membranes (Schleicher & Schuell GmbH). To verify transfer efficiency, the membranes were stained with Ponceau S red (Sigma) (0.5% (w/v) in 5% acetic acid). The same results were achieved with the MemCode Reversible Protein Stain for nitrocellulose membranes from Pierce (Pierce/Perbio Science, Lausanne, Switzerland) (data not shown), which was described to have a detection limit of 25 ng (250 ng for Ponceau S red). For immunoprecipitation and immunoblotting, the polyclonal POMP antibody was purified over a protein G column (Amersham Biosciences), except for Figs. 2C and 5A where the rabbit serum was used without further purification. The blots were developed with anti-POMP or a 5×His-specific mAb (Qiagen) and goat antirabbit or goat anti-mouse POD-labelled secondary antibody (both Sigma), and the chemiluminescent substrate SuperSignal West Pico from Pierce. 2.6. SDS-PAGE Coomassie- and zinc-stainings For Coomassie-stainings, the gels were fixed and stained at the same time with 0.2% (w/v) Coomassie Brilliant Blue R250 (Roth, Reinach, Switzerland) in 50% methanol, 10% acetic acid and destained with 20% methanol, 10% acetic acid. Zinc-stainings were performed with the imidazol-based zincstaining from Roth (Roti-white-staining). In some cases (shown in Fig. 1D), the gels were stained a second time after the first staining vanished (usually after 1–2 weeks when the gels were stored in ddH2 O at RT), which greatly reduced the background. In Fig. 2D, the Coomassie-stained gel was stained in a second step with a zinc-staining according to the protocol of

Fig. 1. Purification of POMP. (A) GST-POMP expression was induced in BL21 E. coli for 3 h with 1 mM IPTG and the protein was purified over a GSTcolumn and removal of the GST-tag with thrombin cleavage occurred directly on the column for 7 h at 30 ◦ C. The SDS gel was stained with Coomassie blue. M = molecular weight marker, 1 = lysate, 2 = cleared lysate, 3 = flowthrough, 4 = wash 1, 5 = wash 5, 6 = wash 10, 7 = flowthrough/thrombin load, 8 = eluate 1 (POMP), 9 = eluate 2 (POMP), 10 = eluate 1 (GST), 11 = eluate 2 (GST). (B) and (C) Another purification of GST-POMP (thrombin digest for 10 min at RT), the blotted membrane was stained with Ponceau red (B) and the same blot developed with the polyclonal anti-POMP serum (diluted 1:10,000, not protein G-purified) (C). M = molecular weight marker, 1 = uninduced culture lysate, 2 = induced culture lysate (3 h, 1 mM IPTG), 3 = flowthrough, 4 = lysate (load), 5 = flowthrough, 6 = wash 1, 7 = wash 10, 8 = eluate 1 (POMP), 9 = eluate 2 (POMP), 10 = eluate 3 (POMP), 11 = eluate 2 (GST). (D) Quantification of POMP with BSA on a zinc-stained SDS-PAGE (second zinc-staining; see Section 2), defined amounts of bovine serum albumin (BSA) were loaded and the calculated amounts of POMP are given in the last two lanes.

262

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

Fernandez-Patron et al. [24], but using the commercial zincstaining from Roth. For documentation, zinc-stained gels were scanned with a black background and the resulting images were further processed with the Adobe PhotoShop program, using the “auto-contrast” or “auto-levels” function. The sensitivity for this staining was given to be below 15 ng. 2.7. Immunofluorescence stainings and confocal microscopy HEK293 cells were grown overnight to a cell density of approximately 50% on poly-l-lysine (Sigma) coated cover slips. Daudi cells were washed twice with serum-free IMDM and then allowed to adhere on poly-l-lysine coated cover slips for 1 h at 37 ◦ C. Then, the cells were washed twice with PBS, fixed with 4% paraformaldehyde and permeabilised with 0.5% Triton X-100 (HEK293) or with 0.5% saponin (Daudi). Cells were stained with the purified anti-POMP antibody and Alexa Fluor 488-labelled secondary goat anti-rabbit F(ab)2 and nuclei were stained with DAPI (both Molecular Probes/Juro, Lucerne, Switzerland) using a standard protocol. The stained cells were mounted with mowiol on microscope slides and examined with a Zeiss LSM 510 Meta confocal microscope (Zeiss, Jena, Germany), using a Plan-Apochromat 63×/1.4 oil DIC objective. Deconvolved microscope images were obtained by collecting z-stack images with 0.2 ␮m thickness of the optical sections (10.8 ␮m in total) and by using AutoDeblur software (AutoQuant, Troy, USA) for deconvolution. 3D reconstructions were made with the ImarisXT software (Bitplane AG, Zurich, Switzerland) using the “surpass” function with maximum intensity projection (MIP). 3. Results 3.1. POMP is only weakly stained by Coomassie blue

Fig. 2. Purification of 6×His-POMP. 6×His-POMP was expressed in E. coli SG13009[pREP4] and purified on Ni-NTA agarose under denaturing conditions. Coomassie-staining (A), zinc-staining (B), and immunoblot with a Histag-specific antibody (C). M = molecular weight marker, 1 = uninduced culture lysate, 2 = induced culture lysate (2 h, 1 mM IPTG), 3 = lysate (8 M urea, pH 8.0), 4 = load (8 M urea, pH 8.0), 5 = flowthrough, 6 = wash 1 (8 M urea, pH 6.3), 7 = wash 5 (8 M urea, pH 6.3), 8 = 6×His-POMP elution 1 (8 M urea, pH 5.9), 9 = 6×His-POMP elution 2 (8 M urea, pH 5.9), 10 = 6×His-POMP elution 3 (8 M urea, pH 4.5), 11 = 6×His-POMP elution 4 (8 M urea, pH 4.5). In (B) only one third of the protein amounts loaded in (A) and (C) were loaded per lane. (D) Quantification of 6×His-POMP with defined amounts of BSA and CD21cyto. In the first step, the SDS-PAGE was stained with Coomassie blue, and in a second step the same gel was stained by a reversible zinc-staining.

POMP was purified as a GST- as well as a 6×His-tagged fusion protein from bacterial expression systems. GST-POMP was purified under native conditions and the GST-tag was removed through cleavage with thrombin. Kinetics of the digest were determined by incubating GST-POMP with thrombin for 5 s up to 22 h and complete cleavage was achieved after 10 min (data not shown). The eluted POMP protein (15.8 kDa) was hardly detectable when the SDS gel was stained with Coomassie blue, while the GST-tag (26 kDa) that was eluted from the column after POMP recovery was visible as well as the GSTPOMP-fusion protein (42 kDa) (Fig. 1A). The identity of POMP and GST was confirmed by peptide mass fingerprinting (data not shown). Staining of an immunoblot membrane with Ponceau red (Fig. 1B) or a commercially available turquoise blue stain (see Section 2, data not shown) did not result in visualisation of POMP. The blot was incubated with polyclonal serum raised against POMP and here, both the fusion- and the cleaved protein were detectable (Fig. 1C). The question then arose why was POMP visible as GST-fusion protein, but not after tag-removal. To answer this question, different approaches to determine the protein concentration of the eluted POMP were

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

263

Table 1 Physicochemical properties of POMP and Ump1 Protein

Number of amino acids

Calculated molecular weight (Da)a

POMP

141

15789.0

1280b , 1490c

5.01

Molar absorption coefficient (ε) at A280 (M−1 cm−1 )

Theoretical isoelectrical point (pI)a

Basic residues (%)

Acidic residues (%)

Hydrophobic residues (%)

9.2

13.5

42.5

GST-POMP

368

42237.7

42200b , 44350c

5.69

12.2

14.4

43.9

GST

227

26466.7

40680b , 42860c

6.32

14.1

14.9

42.8

17310.7

1280b , 1490c

5.95

9.1

12.3

40.8

16759.8

13940b , 13980c

5.01

10.1

14.2

35.9

6×His-POMP Ump1 a b c

154 148

Prediction by ProtParam [37]. Calculated after the method by Edelhoch [27]. Calculated after Pace et al. [25].

undertaken. The conventional colorimetric methods of protein concentration determination mostly rely on the Coomassie dye (e.g. Bradford assay) or on the presence of aromatic amino acids and cystines (e.g. bicinchoninic acid (BCA) and Lowry assay), or additionally on high protein concentrations (Biuret assay). All these methods could not be applied to POMP protein concentration determination since it contains one tyrosine only, one cysteine and possesses no tryptophans. Therefore, an alternative approach was used. In reverse zinc-stainings proteins appear as unstained, transparent bands and only the background is stained white through selective precipitation of the metal cations on the gel matrix. One third of the elution fractions E1 (lane 8) and E2 (lane 9) shown in Fig. 1A were loaded on a zinc-stained gel and the amounts were calculated as 13 and 2.4 ␮g, respectively (Fig. 1D). The protein concentration was determined spectroscopically by absorbance at 280 nm, using the molar extinction coefficient (ε) of 1490 M−1 cm−1 for a folded protein according to Pace et al. [25] (Table 1) and the Lambert–Beer law (c = A280 Mw ε−1 d−1 ). Interestingly, with this staining, elution fraction E1 appeared as 1–2 ␮g and elution E2 as ∼0.5 ␮g when compared to BSA. The detection limits for Coomassie blueand Ponceau red-stainings are given as 0.5–0.05 ␮g [24,26] and 0.25 ␮g [26], respectively. According to these guidelines, POMP should have been stained much better by Coomassie blue and Ponceau red. It can therefore be concluded that POMP has very weak staining properties with these two dyes. Because of the problems described above, a second approach to purify POMP was applied. POMP was cloned into the pQE-32 vector for His-tag based affinity protein purification. 6×HisPOMP was purified under denaturing conditions using 8 M urea buffer and elution occurred due to lowering the pH. Again, the protein was not detectable by Coomassie-staining (Fig. 2A). The amount of 6×His-POMP protein loaded was calculated with the ε = 1280 M−1 cm−1 for an unfolded protein according to the method of Edelhoch [27] as 6.7 ␮g for elution fraction 3 (lane 10) and 8.1 ␮g for elution 4 (lane 11). In a reverse zinc-staining, however, 6×His-POMP was easily detectable as inferred from lanes 7 to 11, although only one third of the protein amounts were loaded per lane (i.e. 2.3 ␮g in lane 10, 2.7 ␮g in lane 11) (Fig. 2B) compared to the Coomassie-stained gel (Fig. 2A). Moreover, immunodetection with a His-tag specific antibody showed that the protein was present and not degraded (Fig. 2C). To verify

the results obtained from the zinc-staining in Fig. 1D where the protein concentration was determined for a folded protein [25], two different amounts of 6×His-POMP calculated by the Edelhoch method [27] were loaded next to defined quantities of BSA as well as to GST-purified CD21cyto, a 4.4 kDa protein (after GST-tag removal; Hoefer, manuscript in preparation). The SDS-PAGE was Coomassie blue stained in the first step where 6×His-POMP remained undetectable, and a zinc-staining was applied on top of the first staining [24]. Here, 6×His-POMP could easily be visualised (Fig. 2D) thus confirming the results seen in Fig. 1D. 3.2. Tetramerisation of POMP Gel-filtration of GST-purified POMP under non-denaturing conditions at pH 8.0 resulted in a single peak consisting of POMP and corresponding to a molecular mass of ∼64 kDa, reminiscent of tetramers (Fig. 3A). The range of the Superdex 75 pg chromatography column was given to be from 3 to 70 kDa and the void volume (V0 ) was 45 ml. Surprisingly, POMP was eluted very early from the column with an elution volume of 50 ml instead of the expected 88 ml. Analysis of the collected fractions with SDS-PAGE under reducing conditions yielded a band of ∼16 kDa in Western blot analysis (Fig. 3B). In the loaded sample as well as in the eluted fractions, POMP dimers and tetramers were also faintly visible above 28 kDa and above 62 kDa (indicated with an asterisk). 6×His-POMP was further analysed on non-reducing gels, revealing two single bands (17–18 kDa and around 38 kDa), suggesting that POMP can dimerise under these conditions, presumably through formation of disulfide bridges with the single cysteine (Cys37 ) present in the POMP sequence (Fig. 3C). However, since the environment is reducing in the cytosol, it seems unlikely that this type of interaction is essential for POMP multimerisation. Whole cell lysates run on nonreducing gels showed a single band at the predicted molecular weight of approximately 16 kDa (not shown). 3.3. Cellular distribution of POMP First, specificity of the polyclonal anti-POMP antibody was tested by comparing the staining properties of the polyclonal serum with the pre-immune serum (Fig. 4A). In addition to

264

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

Fig. 3. Gel-filtration chromatography of POMP. (A) Purified POMP was applied on a Sephadex 75 pg chromatography column under native conditions (50 mM Tris–HCl, 150 mM NaCl, pH 8.0). Ovalbumin (43 kDa, Ve = 65.5 ml), ribonuclease A (13.5 kDa, Ve = 87 ml) and CD21cyto (4.4 kDa, Ve = 94 ml) served as calibration proteins. (B) Western blot analysis of the POMP-containing fractions; L = filtered sample (load), fractions 1 + 2, and 5–9. Asterisks indicate POMP dimers and tetramers. (C) SDS-PAGE under non-reducing and reducing conditions and immunoblot analysis of 6×His-POMP with purified POMP-specific antibody.

monomeric POMP, the polyclonal serum also recognised bands, which could correspond to POMP dimers and tetramers (indicated by asterisks), while the pre-immune serum did not show a specific staining of the purified POMP, GST or 6×His-POMP protein. The cellular localisation of POMP was determined by using purified anti-POMP antibody for immunoprecipitations from cytosolic- and membrane-enriched lysates of the two Blymphoid cell lines, Raji and Daudi. The majority of the protein was not membrane-associated, but cytosolic (Fig. 4B). To further investigate the cellular localisation of POMP, the Daudi Blymphoid cell line (Fig. 5A–C) and the non-lymphoid HEK293 cells (Fig. 5D–F) were immunostained with purified POMP antibody. In both cell types, it was evident that POMP localises to the cytosol as well as to the nucleus. The control staining with unspecific rabbit antibody and the secondary antibody did not show the specific cytosolic and nuclear staining (not shown). Since the nuclear POMP staining showed discrete spots where the staining was more intense, we further processed z-stack images of HEK293 cells by deconvolution and 3D reconstructions (Fig. 5G–I), confirming that the spots were localised within the nucleus. In Daudi B-cells this strong staining in dots was not present (Fig. 5A–C). Immunofluorescence staining of proteins within PML-NBs is dependent on the use of Triton X-100 for permeabilisation [22], which could only be used for HEK293

cells. B-cells (Daudi) are very fragile and required a more gentle treatment with saponin, which probably explains why POMP staining in those dots, that might represent PML-NBs, was not observed in these cells. 4. Discussion In the process of preparing recombinant POMP for antibody production we found that the protein is only weakly stained with the Coomassie and Ponceau dyes. Coomassie-staining largely depends on basic amino acids [28] and it has been reported, that acidic proteins have a low affinity to the Coomassie dye [24]. With a theoretical pI of 5.01 and about 9% basic amino acids POMP does not appear to be unusual when compared with the other proteins listed in Table 1. Thus, the different staining properties cannot be explained by these two parameters. The theoretical sensitivity for Coomassie blue was reported to be 0.5–0.05 ␮g [24,26] and 0.25 ␮g for Ponceau red [26]. Although sufficient amounts of POMP were loaded in our experiments, the protein could not be visualised adequately with these stainings. It can be excluded that POMP diffuses from the gel during the Coomassie-staining and destaining process because a zincstaining on top of the Coomassie-staining disclosed the protein (Fig. 2D). Therefore, the protein reveals weak staining prop-

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

Fig. 4. Localisation of POMP. (A) Specificity of the generated rabbit polyclonal POMP antibody: purified POMP, GST and 6×His-POMP were separated on a 12% SDS-PAGE, blotted onto a membrane and the blots developed with rabbit serum collected prior to immunisation, and with serum collected after 3 injections of purified POMP. Asterisks indicate POMP dimers and tetramers. (B) Immunoprecipitation of POMP from cytosolic- (C) and membrane-enriched (M) lysates of (107 ) Raji and (107 ) Daudi cells and immunoblot analysis with purified POMP-specific antibody.

erties, staining only faintly with both dyes. In contrast, POMP was easily detectable with reverse zinc-stainings. Here, 13 ␮g of POMP (13.6 ␮g of 6×His-POMP) were visible as 1–2 ␮g (1–5 ␮g for 6×His-POMP) in zinc-stained gels when compared with BSA (and CD21cyto) staining. We therefore assume – if the zinc-stainings can be considered quantitative – that in addition to the weak staining properties of POMP in Coomassie and Ponceau, the A280 -calculations revealed misleading results. The molar absorption coefficient ε of proteins depends on the aromatic amino acids tryptophan and tyrosine and on the presence of cystine, i.e. disulfide bonds [25,27] and it is a reliable method to calculate protein concentrations by the Lambert–Beer law using ε together with absorption measurement at 280 nm and the molecular weight. The extremely low ε values for POMP can be explained by the absence of tryptophans and the presence of only one tyrosine and one cysteine residue. Pace et al. [25] admitted a reduced reliability for the ε calculation of proteins when no tryptophans are present. However, the average deviation for proteins without tryptophans in aqueous solutions was given as only 6.5%. In our experiments, POMP protein elutes with an apparent molecular weight of 64 kDa when gel-filtrated. This value may be explained as a homo-tetramerisation of the protein which, as a monomer, has a calculated molecular weight of 16 kDa and which is also apparent in SDS-PAGE under reducing conditions. The relatively small range of the column (3–70 kDa) allows to draw the conclusion, that POMP likely does not elute

265

as pentamer (80 kDa), as a protein of this molecular weight would not be retained by the column. POMP was not found in the flowthrough, but eluted with an elution volume (Ve ) of ∼50 ml. If POMP would elute as trimer (48 kDa), it would have been expected to elute in close vicinity of the 43 kDa ovalbumin (Ve = 65.5 ml), with a calculated Ve of ∼62 ml. To confirm our findings, we attempted to run a native gel with subsequent western blotting (not shown), however, under the conditions tested, POMP did not adhere to the membrane, while the used reference protein (BSA) did. Secondary structure predictions of POMP showed that the protein is composed of long stretches of ␣-helices, and no further characterised coils in between ([29]; Supplementary Fig. S1). According to the Globe program, the protein is most probably not folded in a globular shape [30]. Due to the high percentage of hydrophobic amino acids (42%) (Table 1), especially of leucines (17%) in the POMP sequence one could speculate that some hydrophobic interactions would lead to tetramerisation, perhaps in a variation of the four-helix bundle motif [31]. Hydrophobic interactions are known to be the most common motif for multimerisation of proteins. GST itself is known to stabilise its dimeric form through a hydrophobic lock-and-key motif [32]. Furthermore, several chaperones act as dimers, tetramers or oligomers, e.g. protein disulfide isomerase (PDI) [33] and the E. coli chaperone, SecB [34]. The addition of the thrombin cleavage site adjacent to POMP and the removal of the GST-tag using thrombin added the three amino acids glycine, serine, and arginine N-terminal to the protein. These residues are known to be rather destabilising in proteins [35], and thus probably do not contribute to the capability of POMP to tetramerise. That 6×His-POMP did not form visible dimers and tetramers may be explained by the fact that the protein was completely denatured through the purification procedure, as it was dissolved in 8 M urea, pH 5.9, and therefore was not expected to fold in its native form. Through yeast-two hybrid and co-immunoprecipitation analysis with His-tagged proteasome proteins, the interaction of the pro-forms as well as the mature forms of ␤5 and ␤5i/LMP7 was shown with POMP [13]. Assembly of the two 13-16S precursor proteasome complexes which both already contain POMP could suggest that at least a dimerisation of POMP is required for 20S proteasome maturation. We here report about the possibility that POMP may act as tetramer, and speculate that POMP dimerises within the 13-16S precursor complexes and then further associates to the tetrameric form when the two precursor half proteasomes are assembled to the mature 20S proteasome. However, crystal structure analysis of the protein should provide more information. Our immunoprecipitation and confocal data showed that a large amount of POMP is present in the cytoplasm. Localisation studies in mammalian cells have shown that 20S proteasomes are found throughout the cell. In this work, we could now show specific POMP-staining in the cytosol as well as in the nucleus. These data are in good agreement with the data from studies in yeast, where the majority of the proteasomes reside throughout the nucleus [18] and GFP-tagged Ump1 protein has been shown to be present in the nucleus as well [19].

266

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

Fig. 5. Immunofluorescence stainings of POMP. Immunofluorescence stainings of POMP (Alexa Fluor 488, green) and nuclei (DAPI, blue) in Daudi (A–C) and HEK293 cells (D–F). Deconvolved pictures of POMP and DAPI stained HEK293 cells from different perspectives. (G) XY-view from the top of one HEK293 cell, (H) 3D-reconstruction with isosurface around the nucleus (DAPI) and clipping of the nucleus, (I) XZ-view from the side of the cell shown in (G). Original magnification was 630× for all images. Scale bars: 20 ␮m.

M.M. Hoefer et al. / International Journal of Biological Macromolecules 38 (2006) 259–267

Additionally, in HEK293 cells, but not in Daudi B-cells, we found an accumulation of POMP in characteristic ‘dots’ in the nucleus. The intensely POMP-labelled spots may correspond to discrete subnuclear domains termed PML-NBs [21,22]. Under normal growth conditions, almost every mammalian cell usually harbours 10–30 PML-NBs, doughnut-like shaped structures of 0.2–1 ␮m in diameter. More than 40 different proteins have been described to be associated with them. Although only PML and Sp100 proteins are constitutive residents in PML-NBs, other proteins have been described to be recruited under specific conditions such as genotoxic stress or oncogenic transformation [36]. As the HEK293 cells were not specifically stimulated, we assume that POMP may be associated to PML-NBs due to the oncogenic transformation of the cells. Moreover, proteasomes have been localised in PML-NBs [20], and upon IFN-␥ treatment an increase of PML-NB size and numbers as well as an accumulation of the IFN-␥-inducible proteasome activator PA28 was observed [22]. In summary, we describe the purification of POMP and its peculiar behaviour using common staining methods. Furthermore, our gel-filtration analysis suggests that POMP can form tetramers. We also found that POMP is located in the cytosol as well as in the nucleus and accumulates in discrete subnuclear spots. Acknowledgements We would like to thank Elisabeth von Seydlitz for excellent technical help, Dr. Samuel Solomon for peptide mass fingerprinting and Drs. Simon Hellstern, Daniel Lusche and Miguel Cabrita for suggestions and critical comments on the manuscript. This work was supported by the Thurgauische Stiftung fuer Wissenschaft und Forschung, the Hans-Hench-Stiftung and the European Union through grant LSHM-CT-2004-005264 and the Deutsche Stiftung Sklerodermie to HI. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ijbiomac.2006.03.015. References [1] [2] [3] [4] [5]

A. Hershko, A. Ciechanover, Annu. Rev. Biochem. 67 (1998) 425. P.M. Kloetzel, F. Ossendorp, Curr. Opin. Immunol. 16 (2004) 76. M.H. Glickman, D. Raveh, FEBS Lett. 579 (2005) 3214. M.H. Glickman, A. Ciechanover, Physiol. Rev. 82 (2002) 373. E. Kruger, P.M. Kloetzel, C. Enenkel, Biochimie 83 (2001) 289.

267

[6] E. Kruger, U. Kuckelkorn, A. Sijts, P.M. Kloetzel, Rev. Physiol. Biochem. Pharmacol. 148 (2003) 81. [7] M. Groettrup, S. Khan, K. Schwarz, G. Schmidtke, Biochimie 83 (2001) 367. [8] L. Burri, J. Hockendorff, U. Boehm, T. Klamp, R.J. Dohmen, F. Levy, Proc. Natl. Acad. Sci. U.S.A. 97 (2000) 10348. [9] T.A. Griffin, J.P. Slack, T.S. McCluskey, J.J. Monaco, R.A. Colbert, Mol. Cell. Biol. Res. Commun. 3 (2000) 212. [10] E. Witt, D. Zantopf, M. Schmidt, R. Kraft, P.M. Kloetzel, E. Kruger, J. Mol. Biol. 301 (2000) 1. [11] P.C. Ramos, J. Hockendorff, E.S. Johnson, A. Varshavsky, R.J. Dohmen, Cell 92 (1998) 489. [12] Q. Chen, Q. Ding, J. Thorpe, R.J. Dohmen, J.N. Keller, Free Radic. Biol. Med. 38 (2005) 226. [13] S. Heink, D. Ludwig, P.M. Kloetzel, E. Kruger, Proc. Natl. Acad. Sci. U.S.A. (2005). [14] D. Schenkein, Clin. Lymphoma 3 (2002) 49. [15] P.M. Voorhees, E.C. Dees, B. O’Neil, R.Z. Orlowski, Clin. Cancer Res. 9 (2003) 6316. [16] A. Goy, A. Younes, P. McLaughlin, B. Pro, J.E. Romaguera, F. Hagemeister, L. Fayad, N.H. Dang, F. Samaniego, M. Wang, K. Broglio, B. Samuels, F. Gilles, A.H. Sarris, S. Hart, E. Trehu, D. Schenkein, F. Cabanillas, A.M. Rodriguez, J. Clin. Oncol. 23 (2005) 667. [17] S.V. Rajkumar, P.G. Richardson, T. Hideshima, K.C. Anderson, J. Clin. Oncol. 23 (2005) 630. [18] S.J. Russell, K.A. Steger, S.A. Johnston, J. Biol. Chem. 274 (1999) 21943. [19] A. Lehmann, K. Janek, B. Braun, P.M. Kloetzel, C. Enenkel, J. Mol. Biol. 317 (2002) 401. [20] C. Wojcik, G.N. DeMartino, Int. J. Biochem. Cell. Biol. 35 (2003) 579. [21] K. Mattsson, K. Pokrovskaja, C. Kiss, G. Klein, L. Szekely, Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 1012. [22] R.P. Fabunmi, W.C. Wigley, P.J. Thomas, G.N. DeMartino, J. Cell. Sci. 114 (2001) 29. [23] S. Tian, W. Liu, Y. Wu, H. Rafi, A.S. Segal, G.V. Desir, Am. J. Physiol. Renal. Physiol. 283 (2002) F142. [24] C. Fernandez-Patron, E. Hardy, A. Sosa, J. Seoane, L. Castellanos, Anal. Biochem. 224 (1995) 263. [25] C.N. Pace, F. Vajdos, L. Fee, G. Grimsley, T. Gray, Protein Sci. 4 (1995) 2411. [26] O. Salinovich, R.C. Montelaro, Anal. Biochem. 156 (1986) 341. [27] S.C. Gill, P.H. von Hippel, Anal. Biochem. 182 (1989) 319. [28] M. Tal, A. Silberstein, E. Nusser, J. Biol. Chem. 260 (1985) 9976. [29] L.J. McGuffin, K. Bryson, D.T. Jones, Bioinformatics 16 (2000) 404. [30] B. Rost, J. Liu, Nucleic Acids Res. 31 (2003) 3300. [31] S. Kamtekar, M.H. Hecht, FASEB J. 9 (1995) 1013. [32] C.S. Alves, D.C. Kuhnert, Y. Sayed, H.W. Dirr, Biochem. J. (2005). [33] T.J. Zhao, W.B. Ou, Q. Xie, Y. Liu, Y.B. Yan, H.M. Zhou, J. Biol. Chem. 280 (2005) 13470. [34] L.L. Randall, J.M. Crane, G. Liu, S.J. Hardy, Protein Sci. 13 (2004) 1124. [35] S.C. Kwok, R.S. Hodges, J. Biol. Chem. 279 (2004) 21576. [36] T.G. Hofmann, H. Will, Cell Death Differ 10 (2003) 1290. [37] E. Gasteiger, C. Hoogland, A. Gattiker, S. Duvaud, M.R. Wilkins, R.D. Appel, A. Bairoch, in: J.M. Walker (Ed.), The Proteomics Protocols Handbook, Humana Press, 2005, p. 571.