THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 275, No. 18, Issue of May 5, pp. 13171–13174, 2000 © 2000 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
Accelerated Publication Atomic Force Microscopy Reveals Two Conformations of the 20 S Proteasome from Fission Yeast* Received for publication, December 23, 1999, and in revised form, March 3, 2000 Published, JBC Papers in Press, March 9, 2000, DOI 10.1074/ jbc.C901035199 Pawel A. Osmulski and Maria Gaczynska‡ From the University of Texas Health Science Center at San Antonio, Institute of Biotechnology, San Antonio, Texas 78245
The proteasome is a major cytosolic proteolytic complex, indispensable in eukaryotic cells. The barrelshaped core of this enzyme, the 20 S proteasome, is built from 28 subunits forming four stacked rings. The two inner -rings harbor active centers, whereas the two outer ␣-rings play a structural role. Crystal structure of the yeast 20 S particle showed that the entrance to the central channel was sealed. Because of this result, the path of substrates into the catalytic chamber has remained enigmatic. We have used tapping mode atomic force microscopy (AFM) in liquid to address the dynamic aspects of the 20 S proteasomes from fission yeast. We present here evidence that, when observed with AFM, the proteasome particles in top view position have either open or closed entrance to the central channel. The preferred conformation depends on the ligands present. Apparently, the addition of a substrate to the uninhibited proteasome shifts the equilibrium toward the open conformation. These results shed new light on the possible path of the substrate into the proteolytic chamber.
The proteasome, a giant proteolytic complex ubiquitous among Eukaryotes, with archaebacterial and prokaryotic ancestors, is a highly controlled device to keep cell division, transcription, and antigen presentation on track (1). Its structure has been studied extensively and currently is a classical example of a truly beautiful molecular architecture (2). The core 20 S proteasomal particle has a molecular mass of about 700 kDa, and degrades peptides and unfolded proteins. The 20 S proteasome is built from four stacked rings (␣---␣), seven subunits in each ring, with active centers located inside the chamber formed by the two inner rings (-). The 20 S core plus two 19 S regulatory complexes create the eukaryotic 26 S proteasome, which can recognize and degrade ubiquitinated substrates. Superb quality electron micrographs of the 20 and 26 S particles are available, and the crystal structure of archaebacterial and budding yeast 20 S proteasomes have been solved (2–5). The * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ To whom correspondence should be addressed: University of Texas Health Science Center at San Antonio, Institute of Biotechnology, 15355 Lambda Dr., San Antonio, TX 78245. Tel.: 210-567-7262; Fax: 210-567-7247; E-mail:
[email protected]. This paper is available on line at http://www.jbc.org
crystal structure of Thermoplasma acidophilum 20 S proteasome resembles the hollow cylinder, with the 1.3-nm opening in ␣-rings. The opening leads to the internal catalytic compartment (4). Unexpectedly, the budding yeast 20 S particle crystallized with inhibitors showed the ␣-rings sealed by the tightly interwoven polypeptide chains of the ␣-subunits (5). Therefore, the following hypotheses were proposed to explain the path of substrates into the catalytic chamber: (i) the attachment of the 19 S complexes is necessary to open the entrance to the central channel; (ii) a substrate may enter the catalytic chamber by narrow side windows between the rings (5); (iii) a presumed open conformation exists, but was not captured in the crystal form (1). Atomic force microscopy is a widely used technique to supply the dynamic structural data about protein complexes (6 –9). The method does not require protein fixing, can be performed in aqueous buffer, and allows one to scan and analyze repeatedly the same molecules under controlled conditions (6 –10). Taking advantage of these features, we were able for the first time to image the native 20 S proteasome in action. MATERIALS AND METHODS
Purification of 20 S Proteasome—The 20 S proteasomes were isolated from an exponentially growing culture of Schizosaccharomyces pombe strain 972h⫺ and purified by a set of differential centrifugation, gel filtration, and ion exchange chromatography (11, 12). Electrophoretically pure proteasomes (about 50 g of protein/ml) were stored at ⫺20 °C in 50 mM Tris/HCl buffer, pH 7.0, with 20% glycerol. The sample diluted to 2 g/ml with 5 mM Tris/HCl, pH 6.8 (buffer I; no glycerol) displayed proteolytic activity against the commonly used fluorogenic peptide substrates (Bachem): succinyl-Leu-Leu-Val-Tyr-7-amino-4methylcoumarin (SucLLVY-MCA),1 butoxycarbonyl-Leu-Arg-Arg-MCA (BocLRR-MCA), and carbobenzoxy-Leu-Leu-Glu--naphthylamide (ZLLE-NA) (11, 12). Thermally denatured lysozyme was degraded by our proteasome preparations as well, as was assessed by the electrophoretic method (11, 12). No degradation of LVY-MCA and AAF-MCA was detected. Lactacystin (40 M), carbobenzoxy-Leu-Leu-Leu-CHO (ZLLnL; 100 M) and carbobenzoxy-Leu-Leu-Leu-vinyl sulfone (Z-LLLVS; 100 M; all from Calbiochem), inhibited degradation of SucLLVYMCA by at least 98%. Imaging with Atomic Force Microscopy (AFM)—For the AFM imaging 2 l of the sample diluted in buffer I to the final concentrations of 1–10 g/ml were deposited on a freshly cleaved mica surface, after a few minutes overlaid with buffer I, and mounted in the “wet chamber” of the NanoScope IIIa (Digital Imaging). The imaging was performed in a tapping mode, using oxide-sharpened silicon nitride cantilevers with tips of nominal spring constant 0.32 N/m (Nanoprobe). Resonant frequency of the tip was tuned to 8 –10 kHz, with amplitude 200 – 600 mV and a setpoint of about 2 V. Areas ranging from 0.04 to 1 square micrometer were continuously scanned with a rate of 2.1 Hz. The observed lateral drift reached about 10 nm per scan. For the 2 g/ml samples a field of one square micrometer typically contained at least 100 of proteasome molecules, and it was possible to image the same group of molecules for an hour or longer. Peptide substrates (final concentration: 100 M), the protein substrate (final concentration: 0.5 g/ml), or inhibitors (40 or 100 M) were injected directly into the chamber. This enabled imaging the same molecules before (control) and after addition of the ligands. To test the influence of increased ionic strength on the imaging conditions and results (13), the Tris-buffered 1 The abbreviations used are: SucLLVY-MCA, succinyl-Leu-Leu-ValTyr-7-amino-4-methylcoumarin; BocLRR-MCA, butoxycarbonyl-LeuArg-Arg-methylcoumarin; Z-LLE-NA, carbobenzoxy-Leu-Leu-Glu-naphthylamide; LVY-MCA, Leu-Val-Tyr-methylcoumarin; AAF-MCA, Ala-Ala-Phe-methylcoumarin; Z-LLnL, carbobenzoxy-Leu-Leu-LeuCHO; Z-LLL-VS, carbobenzoxy-Leu-Leu-Leu-vinyl sulfone; AFM, atomic force microscopy.
13171
13172
Two Conformations of the 20 S Proteasome
FIG. 1. Molecules of native 20 S proteasome isolated from fission yeast imaged with the tapping mode AFM in liquid. Left, surface plot of a one square micrometer field of control proteasomes. Both side view and top view particles were randomly distributed through the image. A few particles in side view and top view positions are marked with arrows and circles, respectively. To distinguish differently oriented molecules, though, it was necessary to analyze zoomed-in parts of the large field. Right, upper panels: two enlarged fields (100 ⫻ 100 nm) containing mostly top view molecules, zoomed-in from the one square micrometer fields of control proteasomes (left) and proteasomes incubated with a substrate, SucLLVY-MCA (right). Right, lower panels: an array of enlarged images of the top view molecules. The top view images presented 7-fold or imperfect 6-fold symmetry, similar to the early EM images (19). NaCl solution, pH 7.0, was added to the chamber to the final concentration of up to 20 mM Tris and 200 mM NaCl. All experiments were repeated at least twice. Stock solutions of the peptides and lactacystin were made with dimethyl sulfoxide (Me2SO), and final concentration of the solvent reached 2% in the sample. Addition of the solvent alone (169 top view particles were analyzed) did not induce detectable changes in the proteasomes’ conformation, as compared with the samples without Me2SO. Standard plain fit and flattening provided with the NanoScope IIIa software were the only processing applied to the images in the height mode. Measurements of the particles were carried out with the section analysis software after applying the tip locus compensation on the images (Digital Imaging and NIH Image). RESULTS
We established conditions for successful observation of a field of unfixed, active eukaryotic 20 S proteasomes in sequence. The objects detected with AFM were identified as proteasomes on the basis of their distinctive shape and dimensions (Fig. 1). The resolutions of about 1 nm (a field of 200 ⫻ 200 nm) to 4 nm (a field of 1 square micrometer) allowed us to distinguish outlines of single subunits of the complex, in full agreement with the results of Dorn et al. (14), where side view archaebacterial 20 S molecules were imaged with a resolution of about 3 nm. About a third of our molecules were in an “upright”, or top view position, “standing” on their ␣-ring (Fig. 1). Their average diameter was in a range of 10 –13 nm. That agreed with the reported 11–12 nm diameter of 20 S particles (2, 15). The majority of the molecules were in a side view position, “lying” on their side with the longitudinal axis pointing at different directions, although always parallel to the mica surface. They displayed the presence of four rings with outlines of the subunits. The average length of the molecules in the side view position was 15–17 nm and compared perfectly with the 15–17 nm range observed under electron microscope or modeled after crystal structure (2, 15). Closer inspection of the control top view molecules surprisingly revealed that the apical part of their ␣-ring could acquire two distinct conformations. Most of the ␣-rings had a coneshaped, rounded top with no signs of a central opening. Such molecules were classified as “closed” in accord with the crystallographic data. However, about 24% of the molecules showed an arrangement of the ␣-subunits forming a crater-like cavity in the middle. This seemingly “open” top suggested an open
FIG. 2. The top view proteasomes acquire two conformations: closed and open. a, images of the control proteasomes in closed and open conformations. The second and forth image from the left are surface plots of the corresponding molecules. The vertical scale is exaggerated to show the structural difference between the two conformers. The particles were sorted into closed or open classes according to the qualitative feature of the shape of the median sections of their ␣-rings. The sections were executed in four directions, as shown on the diagram. Median sections of more than 90% of all analyzed molecules presented always either a “cone” or a “crater” regardless the direction of a section, a scan direction, the resolution achieved, the density of a sample, or the ionic strength of the buffer. If all four median sections were cone-shaped, the molecule was classified as closed. If all four median sections were crater-shaped, the molecule was classified as open. Among many hundreds of analyzed molecules we never encountered difficulty in univocal classification of particles into open or closed classes according to this qualitative criterion. Only rarely were we able to detect molecules in presumably intermediate conformations, with the shapes of the four median sections strikingly differing between each other. This type of molecules constituted less than 10% of total of our observations and was excluded from further analysis. Diameters of the closed and open molecules measured 1.5 nm from the top, and below that threshold, did not differ significantly, and both were 12 ⫾ 1 nm (n ⫽ 17 closed and n ⫽ 15 open molecules). The diameters of the closed molecules measured at 0.5 nm from the top (6 ⫾ 1 nm; n ⫽ 17) were significantly (p ⬍ 0.001) smaller than diameters of the open particles (7 ⫾ 1 nm; n ⫽ 15). b, median sections of images of a typical control 20 S particle scanned repeatedly. Sections of the open molecules are red. The blue shading in the diagram of top view particle shows the actual part of the molecule represented by the sections. Sections in the direction number 1 (see diagram in Fig. 2a) are presented. The sections executed in all other directions were indistinguishable in shape from the sections pictured.
entrance to the central channel. The closed and open molecules were univocally distinguished by inspection of their median cross-sections (Fig. 2a). The AFM technique gave us the opportunity to image the same molecules a specific number of times (9, 10). We noticed that the same, control 20 S particles scanned repeatedly switched from the open to closed conformation during successive scans (Fig. 2b). Similar conformational changes have also been observed with the AFM for the membrane pore complex (10). We concluded that our observations represented the real movements of the native proteins and that the molecules were not simple “frozen” in just one of the conformations (Fig. 2b). All molecules imaged in sequence were able to change their conformations. However, the long term observations did confirm that all the control molecules spent preferably most of time in the closed conformation (Figs. 2b and 3). Consequently, in any successive scan the closed particles significantly out-
Two Conformations of the 20 S Proteasome
FIG. 3. The relative content of open molecules increases upon addition of the substrate. The conformation of 12 typical single proteasome molecules was monitored in consecutive scans in 4-min intervals (a grid on the horizontal scale represents 4 min). The results came from two independent experiments. Every row of circles represents a single analyzed molecule. The filled circles represent the closed conformation, and the open circles represent the open conformation of the proteasome molecule. The classification into the closed and open conformations was performed as described in the legend to Fig. 2a. The vertical line marks the addition of the proteasome substrate, SucLLVYMCA, into the sample scanned. For n ⫽ 12 molecules, 31% ⫾ 16% were open in control scans, and 74% ⫾ 13% were open after addition of the substrate (mean ⫾ S.D.). Six of the selected scans of control proteasomes (n ⫽ 6) contained at least three particles at given time point and 36% ⫾ 20% of them were open. However, for eight scans (n ⫽ 8) collected after addition of the substrate, the percentage of open molecules increased to 74% ⫾ 13%. Both in the case of counting single molecules and single time points (scans), the difference between control and substrate-treated samples was statistically significant (p ⬍ 0.0001).
numbered the open particles (Fig. 2b and 3). This result was independent of the concentration of proteasomes, the ionic strength of the buffer, the direction of the scan, or the resolution of the imaging (Fig. 3). The closed conformation prevailed in the control, “idle” proteasomes. However, the constant presence of a small population of open molecules stimulated us to test if addition of a substrate would shift the equilibrium between the conformers. For this purpose we added to the scanned sample a commonly used model substrate of the chymotrypsin-like activity of the proteasome, peptide SucLLVY-MCA, and collected data as before. Both the open and closed molecules were still clearly distinguishable after addition of the peptide substrate. However, in sharp contrast to the control sample, now the open molecules constituted a majority (74%) among the top view proteasomes, regardless the conditions of the imaging (Figs. 3 and 4). Similar experiments executed with other model peptide substrates, like BocLRR-MCA for trypsin-like activity or Z-LLE-NA for caspase-like activity of the proteasome, and also with a protein substrate, lysozyme, induced effects comparable with those obtained with SucLLVY-MCA (Fig. 4). We concluded that the presence of the substrate led to a shift in the equilibrium between the closed and open forms, favoring the latter. On the basis of these findings we speculated that the addition of a peptide which is noncleavable by the proteasome would preserve the control-like equilibrium between the forms. We selected two such peptides: LVY-MCA, which closely resembles the good proteasomal substrate SucLLVY-MCA, and peptide AAFMCA used as a substrate for proteolytic complexes like tricorn, multicorn, and tripeptidyl peptidase II (16). In agreement with our prediction, the top view molecules showed partitioning be-
13173
FIG. 4. The relative content of open molecules depends on the type of the ligand bound. The bars represent mean percent of open particles ⫾ S.D. calculated from 5–10 independently scanned fields of proteasomes. The differences between the control and proteasomes incubated with substrates were statistically significant (p ⬍ 0.0001). Proteasomes treated with nondegradable peptides or with inhibitors (lactacystin (LC), Z-LLnL, Z-LLL-VS) or with the substrate following the inhibitors were mostly closed. If the addition of substrate followed nondegradable peptide, the proteasomes became mostly open (p ⬍ 0.0001). The total number of analyzed molecules ranged from 119 (control) to 488 (⫹LC, ⫹SucLLVY-MCA). Forty-two particles were analyzed in the experiment with lysozyme. The distribution of the two conformers was tested in control samples and samples treated with the SucLLVY-MCA substrate under different conditions: different concentration of proteasomes, different scan size (resolution), and different ionic strength of the buffer used (see “Materials and Methods”). In all cases control samples contained 22–27% of the open molecules, and all samples incubated with the substrate contained 70 – 81% of open particles. Between 50 and 100 particles were analyzed for each condition.
tween the open and closed states exactly as the control (Fig. 4). The nondegradable peptide clearly did not block the active centers, since subsequent addition of SucLLVY-MCA resulted in a massive opening of the ␣-rings (Fig. 4). As the next step we tested how addition of the specific proteasome inhibitors would affect conformation of the 20 S complexes. We choose three inhibitors: lactacystin, Z-LLnL, and Z-LLL-VS. Lactacystin binds covalently and irreversibly to the active site threonines inside the -rings. The sites affected are responsible for the most of the chymotrypsin- and trypsin-like activities of the proteasome (2, 17). Z-LLnL and Z-LLL-VS are examples of tripeptide-based inhibitors that bind reversibly (Z-LLnL) or irreversibly (Z-LLL-VS) to all the proteasomal active centers (17). Both Z-LLnL and Z-LLL-VS are able to interact with high affinity with the substrate-binding sites of the proteasome, since their blocked tripeptide chains mimic well the structure of model substrates (17). A preference for the closed conformation was expected upon the treatment with the inhibitors. Indeed, the ratio of the open to closed top view molecules remained similar in the control and the proteasomes treated with the inhibitors (Fig. 4). All individually analyzed molecules could still oscillate between the two conformations. However, the inhibited 20 S particles were no longer able to react on the addition of the SucLLVY-MCA substrate with a massive ␣-ring opening and remained in the control-like equilibrium of the open and closed conformants (Fig. 4). DISCUSSION
The work presented here is to the best of our knowledge the first dynamic atomic force microscopy analysis of the structure of 20 S proteasomes at work. The use of electron microscopy allowed the detection of the rotation of the 19 S caps with respect to the 20 S core as the major movement in the 26 S
13174
Two Conformations of the 20 S Proteasome
proteasome complexes (3). The recently published AFM images of the archaebacterial 20 S proteasomes contained molecules in a side view position only (14). Our purpose was not to confirm the structure of the 20 S complex already known at high resolution. Instead of averaging the collected images, we aimed at obtaining a set of dynamic representations of the topmost parts of the molecules in action. Nevertheless, we could still identify the outlines of single subunits and the expected overall shape and size of the 20 S particles. The imaged shape of the particles was a result of the physical shape of the molecule and its movements and of the dynamic interactions of a tip with the molecule surface. We aimed at comparing molecules imaged under the same physical scanning conditions, with potentially the same tip-molecule interactions, but under different catalytic conditions. A field of the particles imaged during a single scan contained always a randomly oriented population of molecules. The specific distribution of the two conformants in all analyzed fields was always independent of the scan direction, the scan size, and the achievable resolution. Since altering the ionic strength of the aqueous environment alters the tip-sample interactions (13), we performed our scans in buffers of different ionic strength to explore the influence of the tip-sample interactions on the observed results. The distribution and overall appearance of the two conformers remained unchanged in all tested buffers. Therefore, we postulate that the presence of the two conformants and the differences observed in their distribution were caused by the distinct behavior of the molecules and not by tip-induced effects or by an imaging artifact. We detected that the proteasome was switching between forms with an apparently closed and open entrance to the central channel. Both the size and shape of the sensing tip, and repulsion forces acting on the tip, did not allow us to image very narrow and deep openings. Therefore, on the basis of the collected images alone we cannot draw any conclusion as to whether the opening penetrates completely through the ␣-ring into the catalytic chamber. However, since the archaebacterial proteasome possess an all open channel, it seems reasonable to hypothesize that the eukaryotic proteasome can potentially open the channel, too. In accord with crystallographic data we found that the resting state and inhibitor bound proteasomes stay preferably in the closed conformation (5). It is possible that the detection of only this particular form in the crystal structure originated from its abundance and its thermodynamic preference to crystallize. On the other hand, our AFM data provided evidence of continuous conformational changes of native 20 S proteasomes. All the individual top view molecules repeatedly switched between open and closed forms during subsequent scans. The particles remained open for most of the time only when the substrate was added to the catalytically active (uninhibited) enzyme. The presence of a cleavable ligand was clearly not sufficient to shift the equilibrium between the closed and open molecules. Otherwise, the addition of a substrate to the proteasomes blocked with an inhibitor would open most of the molecules. None of the inhibitors tested cause a preference for the open conformation. Therefore, we excluded the possibility that binding any ligand to the catalytic center and/or to the substrate-binding sites caused the shift of equilibrium toward the open conformation. Apparently, both the presence of an unblocked active center and the presence of a cleavable ligand were necessary for the molecules to remain open for most of the time. Both the inhibitors and the substrates acylate the active site threonines. However, in the case of inhibitors there is no release of the leaving group (N-terminal product) and no deacylation of the active site threonine with the release of the re-
maining group (C-terminal product). Therefore, we may hypothesize that the actual release of the leaving products, or the deacylation step in general, promote the open conformation. There is no basis, though, to conclude that any of these steps directly cause opening of the molecule. Our observations stimulated us to search for the possible mechanism driving the nanometer scale structural transition. The open-close transition was not simple harmonic motions, since the proteasomes were captured staying in one of the conformations for a relatively extended period of time. Since a single act of degradation of a polypeptide by the 20 S proteasome may take seconds (17), it seems entirely possible for the molecule to remain in a stable conformation for a relatively long time. A single picture of a field was collected every 4 min, and it took a few seconds to obtain the image of an individual molecule. On this basis the estimated period of the oscillations was in the range of at least several seconds with fast switches between the two stable conformers. The spatial and temporal extent of the protein movement fitted the commonly accepted properties of only one type of structural change, namely the allosteric transition. Since there is strong biochemical evidence for - to -subunit allostery (18) and for the interdependence of the substrate binding and cleaving processes (14), and since the polypeptide chains of ␣- and -subunits are tightly interwoven, it is reasonable to envision a transfer of structural changes not only within the -rings, but also onto the ␣-subunits, to cause the open-close transition. On the other hand, we cannot exclude that interactions of ligands directly with ␣-rings may be, at least in part, responsible for the observed structural changes. Taking into account the complexity of the proteasome molecule we may envision that the described changes in ␣-rings are only a part of the observed allosteric transition. In fact, we already observed ligand-induced conformational changes in the side view molecules.2 Another direction of AFM studies will be to explore the dynamics of the 26 S particle. We already obtained well resolved AFM images of 26 S molecules from fission yeast in our preliminary studies. Capturing the stages during which 26 S proteasome works on the degradation of a ubiquitinated substrate will be a future challenge. REFERENCES 1. Rock, K. L., and Goldberg, A. L. (1999) Annu. Rev. Immunol. 17, 739 –779 2. Bochtler, M., Ditzel, L., Groll, M., Hartmann, C., and Huber, R. (1999) Annu. Rev. Biophys. Biomol. Struct. 28, 295–317 3. Walz, J., Erdmann, A., Kania, M., Typke, D., Koster, A. J., and Baumeister, W. (1998) J. Struct. Biol. 121, 19 –29 4. Lowe, J., Stock, D., Jap, B., Zwickl, P., Baumeister, W., and Huber, R. (1995) Science 268, 533–539 5. Groll, M., Ditzel, L., Lowe, J., Stock, D., Bochtler, M., Bartunik, H. D., et al. (1997) Nature 386, 463– 471 6. Bustamante, C., Erie, D. A., and Keller, D. (1994) Curr. Opin. Struct. Biol. 4, 750 –760 7. Fritz, M., Radmacher, M., Cleveland, J. P., Allersma, M. W., Stewart, R. J., Gieselmann, R., et al. (1995) Langmuir 11, 3529 –3535 8. Hansma, H. G., Kim, K. J., Laney, D. E., Garcia, R. A., Argaman, M., Allen, M. J., et al. (1997) J. Struct. Biol. 119, 99 –108 9. Engel, A., Lyubchenko, Y., and Muller, D. (1999) Trends Cell Biol. 9, 77– 80 10. Muller, D. J., Baumeister, W., and Engel, A. (1996) J. Bacteriol. 178, 3025–3030 11. Gaczynska, M., Rock, K. L., and Goldberg, A. L. (1993) Nature 365, 264 –267 12. Osmulski, P. A., and Gaczynska, M. (1998) Curr. Biol. 8, 1023–1026 13. Muller, D. J., Fotiadis, D., Scheuring, S., Muller, S. A., and Engel, A. (1999) Biophys. J. 76, 1101–1111 14. Dorn, I. T., Eschrich, R., Seemuller, E., Guckenberger, R., and Tampe, R. (1999) J. Mol. Biol. 288, 1027–1036 15. Coux, O., Tanaka, K., and Goldberg, A. L. (1996) Annu. Rev. Biochem. 65, 801– 847 16. Yao, T. T., and Cohen, R. E. (1999) Curr. Biol. 9, R551–R553 17. Bogyo, M., McMaster, J. S., Gaczynska, M., Tortorella, D., Goldberg, A. L., and Ploegh, H. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 6629 – 6634 18. Kisselev, A. F., Akopian, T. N., Castilio, V., and Goldberg, A. L. (1999) Mol. Cell. 4, 395– 402 19. Baumeister, W., Dahlmann, B., Hegerl, R., Kopp, F., Kuehn, L., and Pfeifer, G. (1988) FEBS Lett. 241, 239 –245 2
P. A. Osmulski and M. Gaczynska, manuscript in preparation.
