Evidence for a Particulate Location of Ubiquitin Conjugates and Ubiquitin-conjugating Enzymes in Rabbit Brain*

Conjugate ubiquitin was previously found in the nucleus, cytoplasm, and membranes of eukaryotic cells while the enzymes of the ubiquitin-conjugating system appear to be cytoplasmic. We have prepared the mito- chondrial fraction from rabbit brain by discontinuous density gradient ultracentrifugation and by Western blotting, using a specific antibody against conjugate ubiquitin, showing that it contains ubiquitin conjugates in a very wide molecular weight range. Electron microscopy and measurement of specific enzyme markers show that this fraction not only contains mitochondria but also some endoplasmic reticulum vesicles. Immuno- staining with anti-ubiquitin IgG followed by immunodecoration with colloidal gold particles provides evi- dence for the presence of conjugate ubiquitin both in mitochondria and in the endoplasmic reticulum. Fur- thermore, this “mitochondrial fraction” shows a pro-nounced ATP-dependent ability to conjugate ‘“I-ubiquitin into a number of endogenous proteins as evi- denced by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. Addition of El, Ez, and Ea, the enzymes of the ubiquitin conjugating system purified from rabbit reticulocytes, does not further increase this ubiquitination nor

Conjugate ubiquitin was previously found in the nucleus, cytoplasm, and membranes of eukaryotic cells while the enzymes of the ubiquitin-conjugating system appear to be cytoplasmic. We have prepared the mitochondrial fraction from rabbit brain by discontinuous density gradient ultracentrifugation and by Western blotting, using a specific antibody against conjugate ubiquitin, showing that it contains ubiquitin conjugates in a very wide molecular weight range. Electron microscopy and measurement of specific enzyme markers show that this fraction not only contains mitochondria but also some endoplasmic reticulum vesicles. Immunostaining with anti-ubiquitin IgG followed by immunodecoration with colloidal gold particles provides evidence for the presence of conjugate ubiquitin both in mitochondria and in the endoplasmic reticulum. Furthermore, this "mitochondrial fraction" shows a pronounced ATP-dependent ability to conjugate '"Iubiquitin into a number of endogenous proteins as evidenced by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. Addition of E l , Ez, and E a , the enzymes of the ubiquitin conjugating system purified from rabbit reticulocytes, does not further increase this ubiquitination nor incorporate ubiquitin into additional protein bands. The same mitochondrial fraction is not able to carry out any ATPdependent degradation of 'z61-albumin; however, it contains an isopeptidase activity able to release the covalently incorporated 'z61-ubiquitin and is also able to conjugate '261-ubiquitin to exogenous proteins as oxidized RNase. By affinity chromatography on ubiquitin-agarose of fraction I1 of a crude Triton X-100 extract of the mitochondrial fraction, several proteins corresponding in M, to the E, and Ez. enzymes were obtained. These proteins were also able to form specific ubiquitin-thiol ester bounds on sodium dodecyl sulfate-polyacrylamide gels and to support 'zsI-ubiquitin conjugation to oxidized RNase. Detergent fractionation of the mitochondrial fraction provided evidence for a possible localization of the ubiquitin conjugating activity in the mitochondrial external membrane and endoplasmic reticulum. The presence of an active ubiquitin protein conjugating system in mitochondria and endoplasmic reticulum may be related to the turnover of organelle proteins as well as to specific cell functions such as import of proteins into mitochondria * This work was supported by Consiglio Nazionale delle Ricerche Target Projects "Biotechnology and Bioinstrumentation" and "Aging." 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.
Ubiquitin is a small polypeptide of M, 8565 with an amino acid sequence extremely conserved and involved in several basic cellular functions (for a review, see Ref. 1).
In 1977 (2, 3) ubiquitin was found in the nucleus, bound to histone HZA, and a few years later it was found to mediate the cytosolic protein breakdown (4, 5) and also bound to cell surface proteins (6-8). More recently ubiquitin was also identified as a component of neurofibrillary tangles in Alzheimer's disease and other neurodegenerative diseases (9-11). In other words, ubiquitin was found to mediate gene transcription (12), DNA repair (13), and cell cycle progression (14,15) as well as selective protein degradation (16), stress responses (17), and modulation of immune response (6). All these different processes involve the conjugation of ubiquitin to specific target proteins through the action of an ATP-dependent ligation system (16). This conjugation is mediated by an ubiquitin activating enzyme El (18-20), a family of ubiquitin carrier proteins E2* (21"23), and a ubiquitin-protein ligase E3 with several different substrate specificities (24, 25). Although there are ubiquitin conjugates in the nucleus, at the cell surface, and in the cytoplasm, the enzymes of the ubiquitin conjugating system appear to be cytoplasmic (26), with few exceptions (27).
In this paper we present evidence for the presence of ubiquitin conjugates in a particulate fraction of rabbit brain that contains mitochondria and endoplasmic reticulum. Furthermore, we also surprisingly found that this fraction, although not able to carry out protein degradation, is able to covalently conjugate ubiquitin in an ATP-dependent manner to endogenous and exogenous substrate proteins and also possesses an isopeptidase activity responsible for ubiquitin release from the conjugates. These data seem to be of interest in understanding the intracellular localization of the pathway(s) responsible in eukaryotic cells for ubiquitin conjugation, and to our knowledge they represent the first evidence for an active ubiquitin-protein conjugating system in a particulate fraction of the brain.

EXPERIMENTAL PROCEDURES
Materiak-Ubiquitin, chloramine T, Protein A, anti-ubiquitin antiserum, and anti-rabbit IgG gold conjugate (10 nm) were obtained from Sigma. Ubiquitin and Protein A were iodinated by Iodo-Beads (N-chlorobenzenesulfonamide-derivatized polystirene beads from Pierce Chemical Co.) as suggested by the manufacturer. The specific activity obtained was 2-2.5 lo6 cpmlpg. Ficoll was from Pharmacia LKB Biotechnology Inc.
Preparation of the "Mitochondrial Fraction"-The mitochondrial fraction from rabbit brain was prepared by a modification of the procedure described in Ref. 28. Briefly, two rabbit brains were

Intracellular Localization
chopped with scissors and homogenized in a Potter-Elvehjem homogenizer in 10 volumes of isolation medium (0.32 M sucrose, 0.5 mM EDTA, 10 mM Tris-HC1, pH 7.4) with a Teflon pestle (clearance 0.1 mm). The homogenate was centrifuged for 3 min at 1,500 X g, the supernatant carefully decanted, and then centrifuged at 12,500 X g for 10 min to obtain the crude mitochondrial pellet. This pellet was resuspended in 40 ml of isolation medium and centrifuged again as above. The pellet of this last centrifugation was resuspended in 6 ml of isolation medium and layered onto a discontinuous Ficoll density gradient. The gradient consisted of 4 ml each of 15, 12, 9, and 7.5% Ficoll (w/w) in isolation medium. Each tube received 1.5 ml of sample and was centrifuged at 100,000 X g for 1 h in a SW28 swingout rotor in a Beckman L8 70 ultracentrifuge. At the end of the centrifugation, the myelin was at the interface between the isolation medium and the 7.5% Ficoll solution, while the mitochondria were collected at the interface between the 9 and 12% Ficoll solution. The mitochondria were then resuspended in 40 ml of isolation medium and pelleted at 12,500 X g for 10 min. This procedure was repeated twice to wash out the Ficollpresent. The final pellet was resuspended in 2 ml of isolation medium and used in further studies as outlined below. Enzymes Assay and Protein Determination-The activities of hexokinase (EC 2.7.1.1), lactate dehydrogenase (EC 1.1.1.27), glutamate dehydrogenase (EC 1.4.1.3), and NADPH-cytocrome c reductase, rotenone insensitive, (EC 1.6.2.3) were determined as described in Refs. 29-32, respectively. Protein was determined by the Bradford method (33) using bovine serum albumine as a standard.
Purification of E,, E2, and E3 from Rabbit Reticulocytes-Reticulocyte fractions were prepared as previously reported (34). The lysate was fractionated on a DE52 column. The unadsorbed material was designated fraction I and contained ubiquitin, whereas proteins adsorbed into the resin and eluted by 0.5 M KC1 were fraction I1 (35).
The three enzymes (E,, E,, and E3) required for the conjugation of ubiquitin with proteins were purified from fraction 11. Briefly, the 0.5 M KC1 eluate was adjusted to 0.1 M Tris-HC1, pH 7.5, and 5 mM EDTA and then resolved into 0-30% (crude E3) and 30-85% (crude E, and E,) precipitating fractions by addition of solid ammonium sulfate (23). After centrifugation the resulting pellets were each suspended in one volume of 50 mM Tris-HC1, pH 7.5, containing 1 mM DTT' and then extensively dialyzed against the same buffer.
Fractions were stabilized by addition of 0.5 mM ATP and stored at -80 "C until use (7-10 days). Isolation of E1 and the family of E, proteins was obtained by covalent affinity on an Affi-gel-ubiquitin column as described by Haas and Bright (23). Ubiquitin affinity column was synthesized using N-hydroxysuccinamide-activated Bio-Gel A from Bio-Rad to a final concentration of 4 mg of ubiquitin/ml of bed volume. E, + E2 were eluted simultaneously by 0.1 M Tris-HCl, pH 9.0, containing 10 mM DTT and then adjusted to pH 7.5 with HCl and concentrated to 6 ml by ultrafiltration using YM-5 membrane. The final E, + E, preparation contained 16 pg/ml of protein while the final crude E3 preparation contained 1.3 mg/ml of protein.
Assay of Ubiquitin Conjugation-The conjugation of ubiquitin to the mitochondrial fraction was assayed by incubation of rabbit brain mitochondria with 1251-ubiquitin. The reaction mixture contained, in a final volume of 250 pl, 50 mM Tris-HC1, pH 7.5, 2 mM MgCl,, 3 mM DTT, 2 mM ATP, 10 mM creatine phosphate, 10 pg of creatine phosphokinase, lo7 cpm of 1251-ubiquitin, 200 pg of mitochondrial protein and where indicated 25 p1 of the (E, + E2) and 25 pl of the E3 solutions. At time 0,30, and 60 min of incubation at 37 "C, aliquots of the reaction mixture (70 pl) were layered onto 1 ml of 0.35 M sucrose solution and centrifuged for 15 min in an Eppendorf microcentrifuge to separate the mitochondria from the bulk of unreacted 1251-ubiquitin and the other soluble components of the reaction mixture. The mitochondrial pellet was then solubilized with 100 p1 of solubilization buffer (50 mM Tris-HC1, pH 8.6, containing 0.2% (v/ v) Triton X-loo), centrifuged again at 14,000 rpm for 15 min, and then boiled for 2 min with the Laemmli electrophoresis sample buffer (36). All samples were then electrophoresed in SDS-polyacrylamide gels, fixed, dried, and autoradiographed. Quantitation of 1251-ubiquitin conjugation was also performed by counting the solubilized mitochondrial pellet after different incubation times (from 5 to 60 min). Since this radioactivity can either be due to the covalent binding of lz5Iubiquitin to mitochondrial protein as well as to "'1-ubiquitin thiol ester formation with E, and E, and to unreacted 1251-ubiquitin, 50 pl of solubilized mitochondria (incubated and processed as above) were The abbreviations used are: DTT, dithiothreitol; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. first added to 10 p1 of 2-mercaptoethanol, heated at 90 "C for 5 min, cooled down to 4 "C, precipitated by 50 p1 of 5% (w/v) trichloroacetic acid, and the protein precipitate washed with 5% trichloroacetic acid on a glass fiber filter (Millipore) that was finally counted (ubiquitin conjugate through isopeptide linkage). An identical amount of solubilized mitochondria was precipitated with cold 5% trichloroacetic acid, washed on glass fiber filters, and counted (total ubiquitin conjugate). The difference between the total conjugate ubiquitin and the amount of conjugate through isopeptide linkage was taken as the amount of ubiquitin bound through thiol ester linkage. When the maximum capability of the mitochondrial fraction to conjugate ubiquitin was tested oxidized RNase was added to the incubation mixture at a final concentration of 100 pg/ml. This was found to be a saturating concentration under the assay conditions employed.
Western Blotting-Electrophoresis of the mitochondrial fraction was performed on 10% polyacrylamide slab gels essentially according to Laemmli (36). Gels were electroblotted according to Towbin et al. (37). The buffer was 25 mM Tris, 192 mM glicine, 20% (v/v) methanol, 0.02% (w/v) sodium dodecyl sulfate. Blotting was performed at 0-4 "C for 20 h at 40 V. After blotting the nitrocellulose sheets were washed for 10 min in 20 mM Tris, 500 mM NaCl, pH 7.5 (TBS) and then agitated for 1 h in a 3% (w/v) gelatine solution in TBS. After washing in 20 mM Tris, 500 mM NaCI, 0.05% (v/v) Tween 20, pH 7.5 (TTBS) the blots were then incubated for 5 h with 1500 rabbit antiubiquitin IgGs (kindly provided by A. Haas, Milwaukee, WI) and diluted in 1% (w/v) gelatin in TTBS, which detect ubiquitin-protein conjugates or a Sigma anti-ubiquitin antiserum which detect preferentially soluble ubiquitin. The membranes were then washed twice in TTBS and incubated for 1 h in 50 ml of 2.106 cpm/ml of lz5Iprotein A. After washing first in TTBS and then in TBS, the membranes were dried and autoradiographed.
Electron Microscopy-The mitochondrial fraction was fixed in 2% (w/v) paraformaldeide containing 0.5% (v/v) glutaraldehyde for 1.5 h at 4 "C. The fixative solution was chosen to block the natural degeneration of cellular component but not to interfere with the immunolabeling procedure. The fixed samples were then rinsed with Sorenson phosphate buffer, pH 7.4, and kept in the same buffer overnight at 4 "C. The samples were then postfixed at 4 "C in 1% osmium tetroxide for 45 min, rinsed in phosphate buffered saline, pH 7.4, at room temperature, gradually dehydrated in mixtures of bidistilled water and acetone in a series of steps of increasing acetone concentration, and finally embedded in Epon resin (the ultrastructural preservation is allowed by this resin and the protein antigenicity is retained). After polymerization, the samples were cut with a diamond knife on an ultramicrotome and the ultrathin sections (about 80 nm in thickness) were collected on nickel grids. Section staining was performed using floating grids, sections down, on droplets of the immunolabeling and washing solutions placed on strips of Parafilm, taking care not to wet the reverse side of the grid or to let the sections dry. After each incubation, grids were washed by sucessive floatations on fresh droplets of buffer, and the excess reagent or buffer was removed with pieces of filter paper. All steps of the immunolabeling procedure were carried out at room temperature as follows. The sections were hydrated in a 60% acetone solution, then transfered in TBS for 10 min. Blocking was performed in 3% (w/v) gelatine in TTBS for 30 min followed by two washing steps of 5 min each in TTBS. The sections were then incubated with 150 rabbit antiubiquitin IgG (A. Haas) for 3 h at room temperature in the antibody buffer (1% gelatine in TTBS). The samples were then washed three times in TTBS and then icubated for 1.5 h with 1:lO dilution of antirabbit IgG gold conjugate in antibody buffer. The final immunolabeling step was followed by two washes of 5 min each of TTBS followed by two further washes in TBS. At the end, grids were washed in the E. M. Philips CM 10. distilled water, air dried, stained with lead citrate, and observed under Treatment with Digitonin-The mitochondrial fraction was incubated with increasing concentrations of digitonin in isotonic media at 1.5 mg of protein/ml. Incubations were at 4 "C for 30 min with constant stirring. The suspensions were immediately centrifuged at 14,000 rpm in an Eppendorf microcentrifuge for 10 min at 4 "C and the supernatant collected for the measurements of specific enzyme markers as outlined under "Results."

RESULTS
Preparation of Mitochondria-Preparation of mitochondria from brain free from non-mitochondrial contaminants has always been recognized as a problem (28). The reasons for 21020 Intracellular Localization of Ubiquitin Conjugates these difficulties are the heterogeneity of the nervous tissue, the large amounts of lipid and membranous materials, and the heterogeneity of mitochondria in their contents. Nevertheless, appropriate methods have been devised and relatively pure mitochondrial fractions may be isolated (28). The procedure we employed to isolate brain mitochondria involves the use of density gradient centrifugation and provides a particulate fraction which contains significant amounts of hexokinase. In fact the specific activity of this enzyme (810 & 50 milliunits/mg protein) is higher than those previously reported for other mitochondria preparations as was the specific activity of another mitochondrial enzyme (glutamate dehydrogenase 913 & 40 milliunits/mg) (38,39). Furthermore, the contamination by cytoplasmic proteins evaluated by measuring the lactate dehydrogenase activity of this fraction was less than 3-4% of the total activity of the homogenate. In contrast, the activity of an endoplasmic reticulum marker (NADPH cytocrome c reductase) was 25% of the total activity. Several different conditions (i.e. homogenizing conditions and solutions, density gradients, temperature, etc.) were modified t o further improve the purification of this mitochondrial fraction but without significant differences in the results obtained. This contamination of brain mitochondria by endoplasmic reticulum has also been observed by others (39). The particulate association of a ubiquitin conjugating activity within the mitochondrial fraction was investigated in the fractions obtained by discontinuous density gradient ultracentrifugation of the crude mitochondrial pellet (see "Experimental Procedures" for details). As shown in Table I, the lactate dehydrogenase activity (a cytoplasmic marker) is highest in the fraction containing mainly myelin while the l2'I-uhiquitin conjugating activity is mainly present in the mitochondrial fraction. The increased ''7-ubiquitin conjugating activity to lactate dehydrogenase ratio in the mitochondrial fraction suggests that the former could not have arisen from contamination with cytosol.
Evidence for the Presence of Bound Ubiquitin-The mitochondrial fraction prepared as above was solubilized with 0.25% (v/v) Triton X-100 in 0.1 M Tris-HCI, pH 7.5, for 30 min, submitted to SDS-polyacrylamide gel electrophoresis and Western blotting. Immunostaining of bound ubiquitin with an affinity purified rahbit-anti ubiquitin antibody provided evidence for the presence of ubiquitin conjugates to a number of proteins with different M , (Fig. 1). This result was confirmed using several different preparations of the mitochondrial fraction and its specificity was proved by the absence of detectable protein bands when using a normal rabbit

TABLE I
Distribution of "'I-ubiquitin conjugating activity and selected enzyme markers in the fractionated mitochondral pellet The crude mitochondrial pellet was fractionated by discontinuous density gradient ultracentrifugation on four layers of 15, 12, 9, and 7.5% Ficoll (w/w) as descrihed under "Experimental Procedures." After centrifugation a fraction was collected at the interface between the isolation medium and the 7.5% Ficoll. This is the myelin fraction. The fraction collected at the interface between 9 and 12% Ficoll solution is the mitochondrial fraction. Fraction  IgG instead of the anti-ubiquitin IgG, or an antiserum prepared against non-denatured uhiquitin that is much more specific for the unconjugate polypeptide (Sigma U-5379). Further evidence for the presence of conjugate uhiquitin in this mitochondrial fraction was obtained hy electron microscopy and immunodecoration with anti-uhiquitin IgG followed by a goat anti-rabhit IgG colloidal gold conjugate. As shown in Fig.  2 this mitochondrial fraction, as expected from the assay of enzyme markers of the different cellular organelles, not only contains mitochondria hut also a numher of endoplasmic reticulum vesicles.

Myelin
Furthermore, hoth mitochondria and endoplasmic reticulum contain ubiquitin conjugates in similar proportions and specifically bound to the membranes of these organelles. The addition of only the second antihody or of normal rahhit I& does not show any specific gold head conjugation (not shown).
Ubiquitin Conjugation-The presence of conjugate uhiquitin in mitochondria and endoplasmic reticulum prompted US to investigate the capability of the rahhit hrain mitochondrial fraction to conjugate '"T-uhiquitin to endogenous protein substrates. The mitochondrial fraction was incuhated with 12"I-ubiquitin, in the presence and ahsence of ATP and an ATP-generating system. Samples were removed at different time intervals from zero to 60 min, loaded onto 0.3 M sucrose solution, and centrifuged for 15 min in an Eppendorf microcentrifuge to separate the protein pellet from unconjugate ' :' Iubiquitin. The mitochondrial pellet was then soluhilized with 0.25% Triton-X 100 in 0.1 M Tris-HCI, pH 7.5, and the solubilized protein separated by SDS-polyacrylamide gel electrophoresis followed by autoradiography. Fig. 3 shows that the rabbit brain mitochondrial fraction contains hoth the enzymes of the ubiquitin-conjugating system as well as proteins that can be uhiquitinated. The most prominent ""Tubiquitin conjugates in t.he autoradiograms show molecular mass of 91, 87, 67, 50, 34, and X? kDa. In addition a hand of >200 kDa is also evident. Uhiquitination of proteins present in the mitochondrial fraction seems to he limited hy substrate availability since addition of the exogenous (from rahhit re-

FIG. 2. Immunolocalization
of ubiquitin-protein conjugates in the rabbit brain mitochondrial fraction. Ultrathin sections of the mitochondrial fraction were hydrated and stained with rabbit anti-ubiquitin IgG (specific for ubiquitin-protein conjugates) followed by goat anti-rabbit IgG colloidal gold conjugate as a second antibody. Left and right panels show the results of two different experiments. from rabbit brain was incubated at 37 "C with IY5I-ubiquitin in the presence ( A and C) or absence (R) of ATP and an ATP-generating system as described under "Experimental Procedures." At time 0,30and 60-min aliquots of the reaction mixtures were layered onto 0.35 M sucrose solutions and centrifuged at 14,000 rpm in an Eppendorf microcentrifuge. The pellets were then solubilized with 0.25% Triton X-100, centrifuged again as above, and the soluble fraction separated by SDS-PAGE after denaturation by boiling in the Laemmli sample buffer (36). In C, El, E?, and E2 purified from rabbit reticulocytes were added to the reaction mixture as specified in the text. The "' Icontaining conjugates were revealed by autoradiography. ticulocytes) conjugating enzymes E l , E?, and En only slightly increases the incorporation of "'I-ubiquitin and does not show additional protein bands (Fig. 3). The presence in the autoradiograms of protein bands with molecular mass >ZOO kDa, even in the absence of added ATP, is probably due to endogenous ATP in the samples. Of interest, this labeled protein band decreases with time in the absence of ATP suggesting that an isopeptidase activity able to release ubiquitin from the conjugates is likely to be present in these samples. The time zero of these experiments corresponds to the time needed t o centrifuge the reaction mixture to separate the unreacted '"I-ubiquitin from the mitochondrial fraction. This time is of a few minutes. The differences in conjugate pattern between Figs. 1 and 3 are probably due to the fact that while in Fig. 1 the endogenous ubiquitin conjugates are shown, while in Fig.  3 only protein able to accept additional ":'I-ubiquitin molecules becomes evident. In other words, it is possible that proteins already ubiquitinated are not able to accept further ubiquitin molecules.
Estimation of Conjugation Activity and Conjugate Turnover-In order to have a careful estimate of both the rates of ubiquitin conjugation to and ubiquitin release from the mitochondrial fraction the amount of '"I-ubiquitin that become conjugated through isopeptide and thiol ester linkages were determined. '"'I-Ubiquitin conjugation assays were found to be linear for at least 30 min a t 37 "C, to be strictly ATPdependent, and to proceed a t a rate of 0.67 & 0.1 nmol/min/ 100 pg protein (not shown). After 15 min of incubation, the '""I-ubiquitin bound through thiol esters was 40 ? 5% of the total '"'I-ubiquitin incorporate (as determined by '"I-ubiquitin released by boiling in the presence of an excess of 2mercaptoethanol). This percentage decrases while increasing the incubation time. After 30 min of incubation, a typical figure for the '"I-ubiquitin-thiol esters was 5 & 1% of total covalently bound "'I-ubiquitin. In order to estimate the rate of ubiquitin release from the mitochondrial fraction, unreacted ""I-ubiquitin and the soluble components of the ubiquitin conjugating system (ATP, creatine phosphokinase, creatine phosphate, etc.) were separated from the mitochondrial fraction by centrifugation a t 14,000 X g over a 0.35 M sucrose cushion. The mitochondrial pellet was then resuspended in a similar volume of isolation medium containing unlabeled ubiquitin and the incubation continued for 30 min a t 37 "C. Evaluation of ".'I-ubiquitin release was linear with time and with the amount of protein and provided values of 0.15 k 0.03 nmol/min/100 pg protein.
Evidence for the Enzymes of the Ubiquitin Conjugating System-Although from the data reported above it is evident that the mitochondrial fraction contains the enzymes of the ubiquitin conjugating system, we performed more accurate analyses to identify the possible isozymic species involved. In fact, eukaryotic cells are known to contain five low M , proteins (E2%) able to accept ubiquitin from the El-ubiquitin thiol ester and able to function in ubiquitin transfer to proteins. However, these isozymes have different functions and only some of these function in E3-dependent conjugation of ubiquitin to proteins (1). Several attempts performed on the crude Triton X-100 extract of the mitochondrial fraction to estimate El and E? contents were without success. Finally, we decided to prepare the fraction I1 of the soluble Triton X-100 extract and to purify these enzymes by ubiquitin-affinity chromatography according to previously published procedures (21). SDS-polyacrylamide gel electrophoresis and silver staining of the protein eluted from the affinity column with 2 mM AMP and 0.04 mM sodium pyrophosphate provided evidence for the presence of a protein with a molecular mass of 105 kDa, while the subsequent elution of the same column with 50 mM Tris-HCI, pH 9.0, and 2 mM dithiothreitol eluted several proteins with the prominent components showing molecular mass of 105 kDa, and 32, 24, and 17 kDa (Fig. 4) 4. A , SDS-PAGE of the ubiquitin-affinity chromatography fractions from the rabbit brain mitochondrial fraction. The rabbit brain mitochondrial fraction was solubilized by 0.25% Triton X-I00 and separated by DEAE chromatography into fraction I (unretained) and fraction I1 (eluted by 0.5 M KCI). Fraction I1 was dialyzed in 50 mM Tris-HCI, pH 7.5, containing 1 mM DTT and chromatographed in the presence of ATP on an ubiquitin-affinity column as in Ref. 23. After extensive washing the column was eluted with 2 mM ATP, 0.04 mM pyrophosphate ( A , I ) and then by 50 mM Tris-HCI, pH 9.0, containing 2 mM DTT (A, 2). These fractions were concentrated by ultrafiltration, separated in a 4-20% SDS-PAGE mini-gel (Bio-Rad), and stained with silver (Gelcode from Pierce Chemical Co.). 1 yg of total protein/lane were loaded. B, '*'I-ubiquitin thiolester formation by protein retained in the ubiquitin-affinity column. The fraction I1 was chromatographed on an ubiquitin-affinity column as in A except that the column was eluted in a single fraction by 50 mM Tris-HCI, pH 9.0, containing 10 mM DTT. This protein fraction was concentrated by ultrafiltration and incubated for thiolester formation as in Ref. 40. Half of the sample was quenched by dilution in an identical volume of electrophoresis sample buffer and boiled for 2 min ( B , 1 ) while the remaining was quenched by dilution in a similar buffer without mercaptoethanol and without boiling ( B , 2). Electrophoresis was in a 10% SDS-PAGE at 5 "C and detection was by autoradiography. that similar M, values have been previously reported for the El and Ez enzymes respectively (22,40). In similar experiments the enzymes of the ubiquitin conjugating system were eluted together from the affinity column in a single step by 50 mM Tris-HC1, pH 9.0, and 10 mM dithiothreitol and their ability to form l2'1-ubiquitin adducts determined in the absence of 2-mercaptoethanol. These results are shown in Fig.  4 and provide evidence for the presence of active ubiquitin conjugating enzymes in the mitochondrial fraction. Addition of El purified as above does not increase the number of E2s detectable by this system (not shown).
" ' Z -Ubiquitin Conjugation to Exogenous Proteins-The evidence for an active ubiquitin conjugating system in the rabbit brain mitochondrial fraction prompted us to investigate the ability of this system to conjugate ubiquitin also to exogenous proteins. As shown in Fig. 5, the ubiquitin conjugating en- '251-Ubiquitin was incubated a t 37 "C with oxidized RNase A (100 pg/ ml), the mitochondrial fraction (2 mg/ml), and ATP plus an ATPgenerating system as described under "Experimental Procedures." At time 0, 30-and 60-min aliquots of the incubation mixture were removed and loaded onto 5% (w/v) Ficoll solution in an Eppendorf tube and centrifuged a t 14,000 rpm for 10 min to separate the mitochondrial fraction pellet from oxidized RNase A. The pellets were discarded while supernatants were boiled in electrophoresis sample buffer (36) and submitted to SDS-PAGE in a 10% polyacrylamide gel. Detection was by autoradiography. zymes of the mitochondrial fraction can form multiple lZ5Iubiquitin conjugates with oxidized RNase. These results were obtained without any addition of detergents and under isotonic conditions thus indicating that the ubiquitin conjugating enzymes of this fraction can have access to soluble proteins as well. Finally, addition of low concentrations of detergents as 0.2 mg of digitonine/mg of proteins (Fig. 6) or 0.05% Triton X-100 (not shown) increase the conjugation of '"I-ubiquitin to the endogenous protein substrates. Since this conjugation involves additional protein bands it can be concluded that ubiquitin conjugation in the mitochondrial fraction is limited by substrate availability and that the addition of low concentrations of detergents or the addition of exogenous protein substrates increase the amount of ubiquitin-protein conjugates. Conjugation of "'I-ubiquitin to endogenous substrates decrease at high digitonine concentration due to a partial solubilization of the '2'I-ubiquitin conjugating activity (see below) and of protein substrates.
Digitonin Fractionation of the Mitochondrial Fraction-The results reported above show that the ubiquitin conjugating activity does not distribute exclusively with mitochondria markers. To assess to what extent mitochondria serve as a source for the enzymes of this conjugating system, the mitochondrial fraction was fractionated with digitonin. The results obtained (Fig. 7) show that the release of lactate dehydrogenase (a cytoplasmic marker enzyme that is probably entrapped in the few synaptosomes present in the mitochondrial fraction) is not paralleled by loss of conjugating activity, thus excluding the cyplasmic origin for the latter activity of this particulate fraction. Furthermore, the response to digitonine of the "'I-ubiquitin conjugating activity appears to correlate with the release of NADH-cytocrome c reductase, an enzyme marker for the mitochondrial external membrane (39). Assay of Protein Breakdown-The mitochondrial fraction was also used to determine whether or not it contained an ATP-dependent proteolytic activity in addition to the ubiquitin conjugating activity. Assay of '251-labeled bovine serum albumin breakdown performed as previously described (29) for 2 h at 37 "C does not show any ATPand/or ubiquitindependent degradation (not shown).

DISCUSSION
Ubiquitin has been implicated in several basic cellular functions in the nucleuos, cytosol, and membranes in a number of cell types (reviewed in Ref. 1). All these functions require the conjugation of ubiquitin to specific target proteins (through its carboxyl terminus) catalyzed by E,, E*%, and E3, commonly known as the enzymes of the ubiquitin conjugating system. In some cases it has been shown that E3 is not necessary for ubiquitin protein conjugation (22, 40, 41). The results reported in this paper show that a particulate fraction of the brain, enriched in mitochondria and with some endoplasmic reticulum, contains ubiquitin conjugated to a number of proteins as well as the enzymes of the ubiquitin conjugating system. This is somewhat unexpected since many reports have provided evidence for an intracellular distribution of ubiquitin that involves the nucleus and cytoplasm. Hovever, specific roles for ubiquitin in mitochondria and endoplasmic reticulum have been suggested. In mitochondria ubiquitin has been implicated in the insertion of newly synthesized enzymes into the outer membrane (42) and in labeling this organelle for proteolytic degradation (43, 44). It has been suggested, though not demonstrated, that the endoplasmic reticulum may represent a place for ubiquitination of extracellular domains of cell surface proteins (45). Furthermore, it has been also shown that newly synthesized proteins in the endoplasmic reticulum that fail to fold correctly or assemble into appropriate oligomeric complex are retained in the endoplasmic reticulum and eventually degraded in an ATP-dependent manner, distinct from lysosomal degradation (46). To what extent our observation, namely that a particulate fraction of the brain is able to covalently conjugate ubiquitin to both endogenous and exogenous substrates, can contribute to clarify the points mentioned above is difficult to say. In fact, the mitochondrial fraction we have prepared contains both mitochondria and endoplasmic reticulum so that one or both cellular fractions can contribute to the activity observed. However some evidence suggests that the second hypothesis is likely to be correct. In fact, immunoelectron microscopy showed the presence of ubiquitin conjugates in both mitochondria and endoplasmic reticulum. Since these compartments would be normally inacessible to cytoplasmic ubiquitin conjugating enzymes we may conclude that these conjugates are formed by local ubiquitin conjugating enzymes. As an alternative explanation these conjugates would have been formed in the cytosol and imported as ubiquitin conjugates in the compartments considered. However, this seems to be unlikely for the reasons discussed in Ref. 45. Furthermore, preliminary results from our laboratory show that both the endoplasmic reticulum and mitochondria prepared from rabbit liver (that is much more easy to fractionate than the brain) are able, although to a different extent, to conjugate ubiquitin? Finally, we have observed (not shown) that several mitochondrial fractions with different contents of endoplasmic reticulum contain almost similar specific activity of the ubiquitin conjugating system suggesting that this system is not a peculiar property of one or the other of these cell compartments.
The characterization of the ubiquitin conjugating system of the mitochondrial fraction reported in this paper provides further support to the idea of compartimentalization of the system. The evidence that low concentrations of detergents