A Population of Rat Liver Lysosomes Responsible for the Selective Uptake and Degradation of Cytosolic Proteins*

Two populations of rat liver lysosomes can be distin- guished on the basis of their density. A major difference between these populations is that one contains the heat shock cognate protein of 73 kDa (hsc73) within the lysosomal lumen. The lysosomal fraction containing hsc73 exhibits much higher efficiencies in the in vitro uptake and degradation of glyceraldehyde-3-phosphate dehy- drogenase and ribonuclease A, two well established substrates of the selective lysosomal pathway of intracellu- lar protein degradation. Preloading of the lysosomal population that is devoid of lumenal hsc73 with hsc73 isolated from cytosol activated the selective transport of substrate proteins into these lysosomes. Furthermore, treatment of animals with leupeptin, an inhibitor of lysosomal cathepsins, or 88 h of starvation also increased the amount of hsc73 within their lysosomal lumen, and these in vivo treatments also activated the selective transport of substrate proteins in vitro . Thus, the hsc73 located within lysosomes appears to be required for ef- ficient uptake of cytosolic proteins by these organelles. The difference in hsc73 content between the lysosomal populations appears to be due to differences in their ability to take up hsc73 combined with differences in the intralysosomal degradation rates of hsc73. The increased stability of hsc73 in one population of lysosomes is primarily a consequence of this lysosomal popula-tion’s more acidic pH.

Two populations of rat liver lysosomes can be distinguished on the basis of their density. A major difference between these populations is that one contains the heat shock cognate protein of 73 kDa (hsc73) within the lysosomal lumen. The lysosomal fraction containing hsc73 exhibits much higher efficiencies in the in vitro uptake and degradation of glyceraldehyde-3-phosphate dehydrogenase and ribonuclease A, two well established substrates of the selective lysosomal pathway of intracellular protein degradation. Preloading of the lysosomal population that is devoid of lumenal hsc73 with hsc73 isolated from cytosol activated the selective transport of substrate proteins into these lysosomes. Furthermore, treatment of animals with leupeptin, an inhibitor of lysosomal cathepsins, or 88 h of starvation also increased the amount of hsc73 within their lysosomal lumen, and these in vivo treatments also activated the selective transport of substrate proteins in vitro. Thus, the hsc73 located within lysosomes appears to be required for efficient uptake of cytosolic proteins by these organelles. The difference in hsc73 content between the lysosomal populations appears to be due to differences in their ability to take up hsc73 combined with differences in the intralysosomal degradation rates of hsc73. The increased stability of hsc73 in one population of lysosomes is primarily a consequence of this lysosomal population's more acidic pH.
The selective pathway of lysosomal proteolysis resembles in many respects the import of proteins synthesized on cytosolic ribosomes into the endoplasmic reticulum, mitochondria, peroxisomes, and nucleus. Thus, uptake of cytosolic proteins by lysosomes is saturable and time-and temperature-dependent (9 -11). Substrates for lysosomal import compete with each other (11), and intermediates in the import process can be identified for certain protein substrates (11). In addition, uptake of selective cytosolic proteins by lysosomes is stimulated by ATP/MgCl 2 and a cytosolic heat shock protein, the heat shock cognate protein of 73 kDa (hsc73; Ref. 8), and requires protein-containing components of the lysosomal membrane (9,11). Recently, lgp96 has been identified as the protein responsible for the binding of RNase A and GAPDH to the lysosomal membrane (14). Furthermore, import of proteins into the endoplasmic reticulum (15,16) and mitochondria (17,18) requires an organellar form of a heat shock protein of 70 kDa (hsp70), and some hsc73, most probably a specific isoform, is localized to the lysosomal lumen and is required for the operation of this pathway of proteolysis (19,20), most likely for the import of protein substrates into the lysosomal lumen.
It is well known that lysosomes are morphologically and biochemically heterogeneous (i.e. Refs. 21-24 and references cited therein). In our previous studies of the uptake of GAPDH and RNase A by rat liver lysosomes we noticed, by immunogold procedures, apparent differences among individual lysosomes in their ability to take up selective protein substrates (7,10). We also observed that hsc73 was present within some, but not all, lysosomes, and that it colocalized with RNase A transported into the lysosomes (7). The amount of lysosomal hsc73 increased by 5-10-fold during prolonged starvation, and this increase was due to both an increase in the number of hsc73 molecules per lysosome and also to an increase in the percentage of lysosomes containing hsc73 (7).
Little is known about the mechanism(s) by which hsc73 reaches the lysosomal matrix, but hsc73, like many cytosolic proteins, can presumably enter lysosomes through macroautophagy and microautophagy. In addition, hsc73 may enter lysosomes by the selective import pathway that it stimulates, because two KFERQ motif peptides exist in hsc73 (13). Once in the lysosomal matrix hsc73 is relatively resistant to intralysosomal hydrolysis (19).
In this paper we characterize in rat liver the specific lysosomal population containing hsc73 to gain more insights into the role of intralysosomal hsc73 in the selective uptake of cytosolic proteins. We show that the hsc73 located within lysosomes is of paramount importance for the uptake of cytosolic proteins by these organelles. Lysosomes that contain hsc73 have increased rates of uptake and decreased rates of intralysosomal proteolysis of hsc73, when compared with lysosomes that do not contain hsc73. Finally, the resistance of hsc73 to intralysosomal hydrolysis is explained, at least in part, by the slightly more acidic pH of this population of lysosomes.

EXPERIMENTAL PROCEDURES
Animals-Male Wistar rats weighing 200 -250 g were fasted for 20 h prior to use. In some experiments fasting was prolonged for 88 h. For experiments with leupeptin, rats received the drug (2 mg/100 g of body weight by intraperitoneal injection) 1 h before sacrifice. All rats were fed ad libitum for at least 7 days before the experiments began.
Chemicals-Sources of chemicals and antibodies were as described previously (7, 9 -11) with the following additions: fluorescein isothiocyanate-dextran (FITC-dextran) (average molecular weight, 67 kDa; 0.007 mol fluorescein/mol glucose), glutathione-agarose and bafilomycin A were from Sigma; Trans 35 S-label was from ICN Pharmaceuticals Inc. (Costa Mesa, CA); antibodies against cathepsin L and against the cation-insensitive mannose 6-phosphate receptor were generous gifts from Dr. G. Sahagian (Tufts University, Boston, MA), and the antibody against the lysosomal glycoprotein of 120 kDa (lgp120) was from Dr. I. Mellman (Yale University, New Haven, CT); antibodies against hexokinase, aldolase, 3-phosphoglycerate kinase, and phosphoglycerate mutase were raised in rabbits against the purified proteins following standard procedures (25). Other reagents were of the best analytical quality available. Glutathione transferase (GST)-hsc73, a fusion protein of hsc73 with GST at its amino terminus was obtained from a plasmid (pGX-2T, Pharmacia Fine Chemicals, Uppsala, Sweden) containing a human cDNA encoding hsc73 (26).
Isolation of Lysosomal Fractions-Lysosomes were isolated from a light mitochondrial fraction in a discontinuous metrizamide density gradient (27) by the shorter method previously reported (10). Fractions from the top layer (fraction 1) and the 26.3/19.8% metrizamide interface (fraction 2) were collected separately, diluted five times with 0.3 M sucrose, and sedimented at 37,000 ϫ g for 10 min in a Sorvall centrifuge (rotor SS-34; DuPont Instruments, Herts, UK). Lysosomes from fractions 1 and 2 were resuspended in 0.3 M sucrose and centrifuged again at 10,000 ϫ g for 5 min in a Heraeus Biofuge 28RS (rotor HFA 22.1; Heraeus Sepatech, Osterode, Germany) to separate pellets (P1 and P2, respectively) and supernatants (S1 and S2, respectively).
Lysosomal membranes from P1 and P2 were obtained as described by Ohsumi et al. (28). Briefly, P1 and P2 were subjected to hypotonic shock in 0.025 M sucrose for 30 min at 0°C and centrifuged for 30 min at 105,000 ϫ g in a Beckman L8M ultracentrifuge (rotor 60 Ti) (Beckman Instruments Inc., Palo Alto, CA). The supernatants (lysosomal matrix) were used as such, and the pelleted membranes were washed once with 0.5 M NaCl and then resuspended in 0.3 M sucrose to the original volume.
Proteolysis Measurements-Freshly isolated lysosomes (25 g of protein) were incubated (final volume, 60 l) with protein substrates. The substrates used were: GAPDH (230 nM) radiolabeled by reductive methylation with [ 14 C]formaldehyde (29) to a specific radioactivity of 1.2 ϫ 10 6 dpm/nmol, RNase S-peptide (35 nM) radiolabeled by reductive methylation with NaB 3 H 4 (30) to a specific radioactivity of 1.1 ϫ 10 7 dpm/ nmol or GST-hsc73 (200 nM) metabolically labeled in Escherichia coli with a mixture of [ 35 S]methionine and cysteine (Trans 35 S-label) to a specific radioactivity of 7.3 ϫ 10 6 dpm/nmol. Except where indicated, incubations were carried out at 37°C for 30 min in 0.3 M sucrose, 10 mM MOPS, pH 7.2. They were stopped by the addition of trichloroacetic acid to a final concentration of 10% or, in the RNase S-peptide experiments, phosphotungstate in HCl to a final concentration of 3.25 and 5%, respectively. The acid-soluble material was collected as flow through after filtration in the Millipore Multiscreen Assay System (Millipore, Bedford, MA) using a 0.45-m pore membrane. Radioactivity in the samples was measured in a Beckman LS-7800 liquid scintillation counter (Pharmacia Biotech, Brussels, Belgium), and quenching was corrected by the external standard method. Proteolysis, after subtracting the acid-soluble radioactivity at time zero, was expressed as percentage of the initial acid-insoluble radioactivity converted to acid-soluble radioactivity and normalized to lysosomal protein. ATP and MgCl 2 were used at 10 mM each alone or with 10 g/ml hsc73 in some experiments.
Incubation Conditions to Study the Uptake of Proteins into Lysosomes-Freshly isolated lysosomes (100 g of protein) were incubated in a final volume of 30 l with protein substrates (25 g of RNase A or 50 g of GAPDH) at 37°C for 20 min in 0.3 M sucrose and 10 mM MOPS buffer (pH 7.2). When studying hsc73 uptake, incubations were carried out with a GST-hsc73 fusion protein (2 g) metabolically labeled with a mixture of [ 35 S]methionine and cysteine (see above) and purified over a glutathione-agarose column. Chymostatin and other protease inhibitors, when used, were added to the lysosomes at 0°C for 10 min at three times their final concentrations and then diluted 3-fold into the incubation medium with substrates. Proteinase K treatments were carried out as described previously (10). The integrity of the lysosomal membranes was judged by the latency of the lysosomal enzyme, ␤-hexosaminidase (11), or by the leakage of proteolytic activity into the medium (10).
Electron Microscopy and Morphometry-For conventional electron microscopy, isolated lysosomes were double fixed (glutaraldehyde and OsO 4 ), embedded in Vestopal W, and stained with lead citrate by standard procedures (10). Ultrathin sections were cut with a LKB 4801 A ultramicrotome and observed in a Philips CM-10 electron microscope. Morphometric analysis (7) was performed in randomly selected electron micrographs at a final magnification of 25,000 -35,000ϫ. Average diameters and areas of the lysosomal profiles were measured in the electron micrographs. Form factors (1.00 for a perfect circle and decreasing with increasing deviation from this form) were calculated as described by Lü ers et al. (31).
Intralysosomal pH Measurements-The intralysosomal pH was monitored by measuring FITC-dextran fluorescence as described in Ohkuma et al. (32). Briefly, FITC-dextran was injected intraperitoneally into rats (15 mg/100 g of body weight), and after 20 h lysosomes were isolated. FITC florescence in lysosomes was measured in a spectrofluorometer at 25°C at 495 nm (pH-sensitive fluorescence) and 450 nm (pH-insensitive fluorescence) excitation and 550 nm emission wavelengths. Double standard curves were prepared by measuring the ratio of fluorescence intensities at 495 nm/fluorescence at 450 nm for 2 mg of FITC-dextran at different pHs and in different buffers and by comparing the fluorescence intensities, at 495 nm excitation wavelengths, in intact lysosomes and after disruption of lysosomes with 0.1% Triton X-100. Intralysosomal pHs in the isolated lysosomes were calculated with these curves from their ratio of the 495/450 nm fluorescence intensities, after subtracting the background fluorescence.
General Methods-SDS-polyacrylamide gel electrophoresis (PAGE) (10, 12, or 17% gels for studies with hsc73, GAPDH, or RNase A, respectively) (33), immunoblotting (34), and fluorography (35) were carried out by standard procedures. Hsc73 was purified from bovine brain cytosol by ATP-agarose affinity chromatography (36). Densitometric analysis of the immunoblots was performed with an 2202 LKB Ultroscan laser densitometer (LKB-Pharmacia, Uppsala, Sweden) with a Hewlett-Packard (Palo Alto, CA) 3396 Series II integrator. The linearity of the method was established using different amounts of hsc73, GAPDH, or RNase A. Protein concentrations were measured by a modification of the Lowry et al. (37) method using bovine serum albumin as the standard. Enzymatic activities were measured by the standard procedures reported by Terlecky and Dice (9), Aniento et al. (10), Barrett and Kirschke (38), and Smith and Turk (39). The energy-regenerating system used in some experiments consisted of 10 mM MgCl 2 , 10 mM ATP, 2 mM phosphocreatine, and 50 g/ml creatine phosphokinase. Statistical analyses were performed by the Student's t test.

RESULTS
In previous experiments we found that uptake of RNase A by rat liver lysosomes was heterogeneous (7). In addition, we noticed that the uptake efficiency of GAPDH by lysosomes decreased when these organelles were collected by centrifugation at 10,000 ϫ g for 5 min instead of the usual 37,000 ϫ g for 10 min (10, 11). These observations suggested that different populations of lysosomes with distinct activities in the selective uptake of cytosolic proteins might be separable based on centrifugation. In most of our previous experiments we used a pool of two lysosomal fractions from a discontinuous metrizamide gradient: the top layer (fraction 1) and the 26.3/19.8% metrizamide interface (fraction 2) (10, 11). Therefore, in a first approximation to identify a lysosomal population active in selective protein uptake, the top layer and the 26.3/19.8% metrizamide interface were centrifuged separately and in succession at 10,000 ϫ g for 5 min and at 37,000 ϫ g for 10 min, thus giving a total of four different fractions (two pellets, P1 and P2, and their supernatants, S1 and S2) to analyze. The protein content of these four fractions were as follows per rat liver: P1 ϭ 1.5 Ϯ 0.1 mg, P2 ϭ 1.8 Ϯ 0.2 mg, S1 ϭ 0.14 Ϯ 0.01 mg, and S2 ϭ 0.44 Ϯ 0.05 mg (38.7, 46.3, 3.6, and 11.4% of total lysosomal protein, respectively).
Immunoblot analysis of the levels of hsc73 reveals that one of these fractions, P2, was practically devoid of hsc73, whereas the remaining fractions had quite similar amounts of hsc73 when corrected for the amount of protein analyzed (Fig. 1A). The lysosome fractions referred to as P1 and P2 had similar low levels of broken lysosomes (Ͻ3% as determined by the latency of lysosomal enzymes (11)), but S1 and S2 contained higher levels of broken lysosomes (10%). S1 and S2 also contained more unidentified membrane fragments, as judged by electron microscopy, so the following experiments were carried out only with the P1 and P2 fractions, which represent 85% of the total lysosomal protein and which will be referred to hereafter as HSCϩ and HSCϪ lysosomes, respectively. Fig. 1B shows that hsc73 was present both in the matrix (about 65% of the total by densitometric analyses of similar immunoblots, when corrected for total protein) and in the membranes from HSCϩ lysosomes. The small amount of hsc73 (20% or less of the hsc73 found in HSCϩ lysosomes) occasionally observed in HSCϪ lysosomes (lane 3) was exclusively associated with the lysosomal membrane (lane 7). Proteinase K treatment of both lysosomal fractions confirmed that the major part of the hsc73 that is found associated with HSCϩ lysosomes is in the lysosomal matrix, whereas all hsc73 associated with HSCϪ lysosomes is in the membrane (data not shown).
Other comparisons between the HSCϩ and HSCϪ lysosomal populations revealed many similarities but also some differences. Thus, the study of the specific activities of different lysosomal markers (Table I) showed comparable activities of cathepsin B, ␤-N-acetyl-glucosaminidase, and ␤-hexosaminidase in both fractions. However, lysosomal enzymatic activities in HSCϪ lysosomes were always slightly but consistently higher than in HSCϩ lysosomes. Recoveries of lysosomal enzymes (between 3-5% of total) were consistent with our previous studies working with the complete pool of lysosomes (5%) (10,11). Also, contamination of any of the fractions with mitochondria (based on the activity of ornithine transcarbamoylase and succinate dehydrogenase) or cytosol (based on the activity of lactate dehydrogenase and GAPDH) was negligible. A typical late endosome marker, the cation-independent mannose 6-phosphate receptor, was not detected in either of the two lysosomal populations (data not shown), suggesting that both populations are mainly mature lysosomes. This conclusion is further supported by the high proteolytic activities (see below) and by the high activities of lysosomal enzymes (Table I) in both fractions. The SDS-PAGE pattern of bands was similar for total lysosomes and for lysosomal membrane and matrix fractions ( Fig. 2A). However, some qualitative or quantitative increases in specific protein bands were evident in the membranes and matrix of HSCϩ lysosomes (i.e.: at about 182, 87, 76, and 45 kDa for membranes and 70 and 47 kDa for matrix; marked by arrowheads in Fig. 2A). These differences were evident in each of 10 separate analyses, but the degree of difference varied from experiment to experiment, especially for the matrix proteins. The levels of lysosomal membrane (lgp120, Rat liver lysosomes, isolated from the top layer (fraction 1) and the first interface (fraction 2) of a discontinuous metrizamide gradient (see "Experimental Procedures"), were washed and centrifuged in succession at 4°C for 5 min at 10,000 ϫ g and for 10 min at 37,000 ϫ g to give two pellets (P1 and P2) and their supernatants (S1 and S2). Lanes 1 in A and B are hsc73 (1 g). A, proteins (50 g) from P1 (lane 2), P2 (lane 4), and S2 (lane 5) or 15 g of protein from S1 (lane 3) were subjected to SDS-PAGE and immunoblot analysis with anti-hsc73. B, lysosomes from P1 and P2 were broken by freezing and thawing, and their matrix and membrane fractions were isolated (see "Experimental Procedures"). Proteins (50 g) from each fraction were subjected to SDS-PAGE and immunoblot analysis with anti-hsc73. cytosolic proteins (10) or GAPDH were also indistinguishable (data not shown). The amounts of several cytosolic proteins in HSCϩ and HSCϪ lysosomes were quantitated using specific antibodies. Typical results for hexokinase and aldolase are shown in Fig.  2D. Hexokinase (and a presumed immunoreactive breakdown fragment) is preferentially localized in HSCϪ lysosomes, whereas aldolase is largely confined to HSCϩ lysosomes. Quantitation of this and similar immunoblots indicate that HSCϩ lysosomes are enriched for GAPDH, aldolase, and phosphoglycerate mutase. However, HSCϪ lysosomes were enriched for hexokinase, and both lysosomal populations contained equivalent amounts of phosphoglycerate kinase (Table  II). Enrichment of HSCϩ lysosomes in GAPDH and aldolase does not seem to be related to a lower proteolytical susceptibility of these enzymes once inside the lysosomal matrix, since equivalent rates of degradation were obtained when incubated with broken lysosomes from both populations (data not shown).
The ultrastructural appearance of the HSCϩ and HSCϪ lysosomes is shown in Fig. 3 (A and B). Although there are not large differences, HSCϩ lysosomes are in general smaller and more elongated than HSCϪ lysosomes. Morphometric analysis of 40 randomly selected lysosomal profiles from 10 different electron micrographs per fraction indicates an average area of 0.075 Ϯ 0.017 m 2 for HSCϩ lysosomes and 0.102 Ϯ 0.021 m 2 for HSCϪ lysosomes. This difference is statistically significant (p Ͻ 0.02). Also, the form factor was 0.698 for HSCϩ lysosomes and 0.907 for HSCϪ lysosomes (1.000 is a perfect circle).
We next investigated the uptake of RNase A and GAPDH by the two lysosomal populations. These proteins are two well established substrates of the selective lysosomal pathway of intracellular protein degradation (10, 11). As shown in Fig. 4, HSCϩ lysosomes were much more efficient than the HSCϪ lysosomes in taking up both substrates. We also tested the effect of exogenously added hsc73 plus ATP/MgCl 2 and an energy regenerating system on the proteolysis of GAPDH and RNase S-peptide by the two lysosomal populations (Fig. 5). We found that these compounds were stimulatory with both lysosomal populations as expected (10, 11) but that HSCϩ lysosomes were more active under all conditions. Then we studied whether HSCϩ and HSCϪ lysosomes exhibit differences in their abilities to take up hsc73. Fig. 6A shows that when intralysosomal proteolysis was inhibited and levels of hsc73 in matrix (after proteinase K treatment) were measured, HSCϩ lysosomes were more effective in taking up exogenously added hsc73 (compare lanes 8 and 9), but some hsc73 was also incorporated into HSCϪ lysosomes (Fig. 6A,  lane 9), as was also the case with GAPDH and RNase A (Fig. 4). Although the amount of hsc73 associated to lysosomes after the proteinase K incubation is lower in Fig. 6 than in freshly isolated lysosomes (Fig. 1B), it should be noted that in these experiments lysosomes received additional treatments that may result in some intralysosomal hsc73 becoming accessible to the exogenously added protease. When RNase A was incubated with HSCϪ lysosomes that had incorporated hsc73 in a prior incubation (as in Fig. 6A), a portion of the RNase A was found associated with the lysosomal pellets (Fig. 6B, lane 5), and part of this RNase A was resistant to proteinase K digestion (Fig. 6B, lane 9). The uptake efficiency of these lysosomes were now at least as good as the original HSCϩ lysosomes  figure) were incubated under standard conditions and in the presence of chymostatin (see "Experimental Procedures") with freshly isolated HSCϩ or HSCϪ lysosomes. The lysosomes were incubated with proteinase K to degrade nontranslocated proteins, centrifuged, and subjected to SDS-PAGE and immunoblot analysis with anti-RNase A or anti-GAPDH antibodies. The insets show two representative immunoblots. The lower band in the RNase A immunoblot (left part of the figure) corresponds to an intermediate in the lysosomal transport of that protein (11). Histograms are means Ϯ S.D. of densitometric analyses of similar immunoblots from four different experiments. Differences from HSCϩ significant at p Ͻ 0.01 (*) and p Ͻ 0.001 (**).

TABLE II
Immunolocalization of cytosolic proteins in HSCϩ and HSCϪ rat liver lysosomal fractions Proteins from HSCϩ and HSCϪ lysosomes (100 g) were separated by SDS-PAGE. Immunoblots derived from four independent experiments that varied by less than 10% were prepared as described under "Experimental Procedures." They were developed using a suitable antibody and densitometrically analyzed. The data are presented as an average percentage of the total density (calculated as sum of absorbances of HSCϩ and HSCϪ lysosomes). PGM, phosphoglycerate mutase; ALDO, aldolase; HEXOK, hexokinase; PGK, 3-phosphoglycerate kinase.  8 and 9). The presence of intralysosomal hsc73 was apparently sufficient to make the HSCϪ lysosomes competent for the selective uptake of cytosolic proteins. We also increased the content of hsc73 in HSCϪ lysosomes by in vivo manipulations. In a first approach, rats were treated with leupeptin to inhibit lysosomal proteases prior to the isolation of the two lysosomal fractions. Immunoblot analysis with anti-hsc73 (Fig. 7A, upper panel) reveals an increase in hsc73 (lane 5) in the fraction corresponding to HSCϪ lysosomes. Although after leupeptin treatment the quantity of hsc73 increased in both lysosomal populations, this increase is considerably higher for HSCϪ lysosomes (ten times) than for HSCϩ lysosomes (two times). The uptake efficiency of proteins (RNase A and GAPDH) (Fig. 7A, middle and lower panels) by the two lysosomal populations isolated from rats treated with leupeptin (lanes 4 and 5) was similar to the original HSCϩ lysosomes (lane 2) and much higher than the HSCϪ lysosomes (lane 3).
Other in vivo evidence was provided by prolonged starvation (Fig. 7B). As reported previously (7), during prolonged starvation over the range 0 -88 h, there is a progressive increase in the activity of the hsc73-dependent selective lysosomal proteolytic pathway coincident with an increase in lysosome-associated hsc73. The two lysosomal populations isolated from rats after 88 h of starvation were indistinguishable from those isolated after 20 h based on ultrastructural appearance and SDS-PAGE protein patterns (data not shown). However, after 88 h of starvation there was an increase in hsc73 in both lysosomal populations (Fig. 7B, upper panel), but the increase was most pronounced for the HSCϪ lysosomes. The uptake of two substrate proteins of this pathway, RNase A (Fig. 7B, middle panel) and GAPDH (Fig. 7B, lower panel) by these lysosomal populations showed a significant increase in the uptake efficiency of the original HSCϪ lysosomes after 88 h starvation (compare lanes 3 and 5). From these observations it appears again that the intralysosomal hsc73 is required for an efficient lysosomal uptake of proteins. Hsc73 was taken up most avidly by HSCϩ lysosomes (Fig.  6A, compare lanes 8 and 9), but this difference cannot completely account for the absence of hsc73 in the HSCϪ lysosomes. Because of the dramatic effect of leupeptin on the hsc73 content of HSCϪ lysosomes (Fig. 7A, upper panel), it seemed possible that hsc73 is also degraded more efficiently within the HSCϪ lysosomes than in the HSCϩ lysosomes. In initial attempts to test this idea, we reductively methylated hsc73 but found that these preparations contained mostly inactive hsc73  1 of A and B, respectively. A, freshly isolated HSCϩ and HSCϪ lysosomes (100 g of protein) were incubated, under standard conditions for 20 min at 37°C in the presence of chymostatin, without (lanes 2, 3, 6, and 7) or with hsc73 (8 g) (lanes 4, 5, 8, and 9). Then the lysosomes were treated (lanes 6 -9) or not (lanes 2-5) with proteinase K, centrifuged and subjected to SDS-PAGE and immunoblot analysis with anti-hsc73. B, lysosomes were incubated without or with hsc73 as in A. After incubation and centrifugation, the lysosomes were subjected to a second standard incubation for 20 min at 37°C with 25 g of RNase A in the presence of chymostatin. Samples were then treated (lanes 6 -9) or not (lanes 2-5) with proteinase K, centrifuged, and analyzed by SDS-PAGE and immunoblot with anti-RNase A.
in that the protein no longer could bind to ATP (8). Therefore, we purified a metabolically labeled [ 35 S]GST-hsc73 fusion protein and followed its degradation during incubation with freshly isolated HSCϩ and HSCϪ lysosomes. [ 35 S]GST-hsc73 is almost two times more effectively degraded by HSCϪ lysosomes than by HSCϩ lysosomes (Fig. 8A). Because lower rates of hsc73 uptake were found for HSCϪ lysosomes (Fig. 6A), differences in efficiency on hsc73 degradation between both populations may take place in the lysosomal matrix. To verify this idea, after incubation of intact lysosomes with a large amount (2 M) of [ 35 S]GST-hsc73 so that a signal could be seen in the HSCϪ lysosomes (Fig. 8B, inset, lane 5), we reisolated the lysosomes that had incorporated a portion of the GST-hsc73 and subsequently measured the degradation of the fusion protein. HSCϪ lysosomes were again more effective in degrading the [ 35 S]GST-hsc73 than were HSCϩ lysosomes (Fig. 8B). Similar degradation rates and the same difference between HSCϩ and HSCϪ lysosomes were obtained with the active fraction of hsc73 radiolabeled with 14 C by reductive methylation that retained its ability to bind to ATP (data not shown). This result indicates that the [ 35 S]GST-hsc73 fusion protein was valid for monitoring intralysosomal degradation of hsc73.
The differences between HSCϩ and HSCϪ lysosomes in their degradation efficiency of hsc73 cannot be explained by differences in the proteolytic activity of broken lysosomes (see above). Therefore, we decided to measure the intralysosomal pH of both lysosomal populations using FITC-dextran. We found that the pH in HSCϩ lysosomes was approximately 0.5 pH units lower than that of HSCϪ lysosomes (Fig. 9A). Sup-plying the lysosomal proton pump with ATP/MgCl 2 slightly increased this difference. That the lysosomal pH could be important in the degradation of intralysosomal hsc73 was apparent when HSCϩ lysosomes were incubated for 30 min with or without ATP/MgCl 2 (Fig. 9B). Without ATP/MgCl 2 most intralysosomal hsc73 was degraded after 30 min of incubation of the HSCϩ lysosomes, whereas in the presence of ATP/MgCl 2 when the lysosomes were slightly more acidified (Fig. 9A) the degradation of hsc73 was retarded. The effect of ATP/MgCl 2 was due to intralysosomal acidification and not due to some protective effect of ATP on the hsc73 because when the ATPase inhibitor bafilomycin A was added together with ATP/MgCl 2 , the ATP-stimulated acidification of the HSCϩ lysosomes was blocked (data not shown), and the lysosomal degradation of the endogenous hsc73 was increased (Fig. 9B, compare lanes 3  and 4).
We also investigated the proteolysis of hsc73 by HSCϩ lysosomes treated with increasing amounts of NH 4 Cl to increase their pH (Fig. 9C). Lysosomal pH was modified from 5.4 to 6.4 by NH 4 Cl, and the proteolysis of hsc73 was found to be most effective in the range 5.75-6.20. Therefore, the difference in pH could explain why HSCϪ lysosomes are more effective than HSCϩ lysosomes in degrading their endogenous hsc73. DISCUSSION We have identified two populations of rat liver lysosomes with very different abilities to selectively take up and degrade protein substrates such as GAPDH and RNase A (Figs. 4 and  5). These differences in uptake rates cannot be explained by FIG. 7

. Effect of leupeptin treatment (A) or prolonged starvation (B) on hsc73 content and activities of HSC؉ and HSC؊ lysosomes.
A, rats were treated or not with leupeptin (2 mg/100 g of weight, intraperitoneally, 1 h before sacrifice), and HSCϩ and HSCϪ lysosomes were prepared as described under "Experimental Procedures." B, rats were subjected to 20 or 88 h of starvation before sacrifice and HSCϩ and HSCϪ lysosomes were prepared as above. Upper panels, proteins (100 g) from the various fractions were separated by SDS-PAGE and immunoblotted with anti-hsc73. HSCϩ (lanes 2 and 4)  differences in the amount of lysosomes in both fractions, because only small differences in the activities of lysosomal enzymatic markers were found between both populations (Table  I). In addition, analysis of RNase A uptake into individual lysosomes by immunogold and electron microscopy (7 and data not shown) confirmed that lysosomes with higher content of hsc73 incorporate a significantly larger amount of RNase A.
The two populations of lysosomes are similar in many respects, but they differ markedly in their content of hsc73 (Fig.  1). That the difference in lysosomal hsc73 in the lumen accounts for the different activities of HSCϩ and HSCϪ lysosomes in their ability to selectively take up and degrade proteins could be shown experimentally by "loading" HSCϪ lysosomes with hsc73 in vitro (Fig. 6) and by two different in vivo treatments, leupeptin injection (Fig. 7A) and long term  (Fig. 7B). Importantly, these HSCϪ lysosomes loaded with hsc73 retained all other characteristics of HSCϪ lysosomes such as their more alkaline pH and their more rounded shapes and larger size (data not shown). In addition, membrane stability and proteolytic activity of HSCϪ lysosomes were not modified after long term starvation. These results suggest that the presence of lumenal hsc73 is the direct cause of the activation of the selective degradation pathway.
This interpretation is also consistent with results from cultured human fibroblasts in which intralysosomal hsc73 was neutralized in living cells by the endocytosis of an immunoprecipitating anti-hsc73 antibody (19). This treatment completely blocked the selective lysosomal degradation pathway without affecting other proteolytic pathways. These results together with the results presented here indicate that the intralysosomal hsc73 is required for the selective uptake of protein substrates into lysosomes. Similar roles for other hsp70 familiy members have been documented for the uptake of mitochondrial precursor proteins into mitochondria (15,16) and for the uptake of endoplasmic reticulum or secreted proteins into the endoplasmic reticulum lumen (17,18). Two main hypotheses have been presented about the mechanism of action of those lumenal chaperones (see Ref. 40 for review). In both of these models, a chaperone attached to the trans-side of the membrane interacts with the translocating polypeptide chain and prevents back movement of the protein (molecular ratchet model) or actively, by hydrolyzing ATP, pulls the protein across the membrane (translocation motor model). By analogy with these studies intralyosomal hsc73 could interact with the protein emerging into the lysosomal matrix, providing the driving force to pull the import intermediate into the lysosome and preventing also the back movement of the protein to the cytosol. In this regard, a kinetic intermediate in the transport of RNase A through the lysosomal membrane has been identified (11).
Intralysosomal hsp70 appears to be hsc73 itself because the lysosomal hsp70 is recognized by two antibodies specific for hsc73 among all hsp70s tested, and the intralysosomal hsp70 co-migrates with one of the hsc73 isoforms in high resolution two-dimensional gels (19). 2 We now provide additional support for this conclusion because cytosolic hsc73 can enter lysosomes in vitro, and this hsc73 functions as well as the lysosomal hsp70 in the selective uptake of proteins.
The preferential uptake of hsc73 by HSCϩ lysosomes, evident when proteolysis is previously blocked (Fig. 6A, compare  lanes 8 and 9), suggests that hsc73 can enter lysosomes, at least in part, through the selective pathway that it stimulates. Consistent with this finding is the presence of two KFERQ-like motifs in hsc73 and the ability of hsc73 to bind to other hsc73 molecules (13). Also, we repeatedly find a stimulation of the selective lysosomal uptake of protein substrates by low levels of hsc73 (Fig. 5) but competition by higher levels (data not shown) consistent with hsc73 entering lysosomes by the same pathway.
Combined with the reduced ability of HSCϪ lysosomes to take up hsc73, they also degrade intralysosomal hsc73 more rapidly than do HSCϩ lysosomes (Fig. 8, A and B). This enhanced degradation in HSCϪ lysosomes appeared to be caused by the more alkaline pH in this lysosome population (Fig. 9A). Acidification of HSCϩ lysosomes further reduced the degradation of intralysosomal hsc73 (Fig. 9B), and mild alkalinization using NH 4 Cl increased degradation of intralysosomal [ 35 S]GST-hsc73 (Fig. 9C). In fact, HSCϩ lysosomes degraded [ 35 S]GST-hsc73 twice as rapidly at pH 5.8 as at pH 5.4, which accounts well for the twice as rapid degradation in HSCϪ than in HSCϩ lysosomes.
The significance of other differences between HSCϩ and HSCϪ lysosomes are less clear. The presence of increased levels of certain cytosolic proteins in HSCϩ lysosomes may indicate that they are selective substrates for this proteolytic pathway. This has been shown to be true for GAPDH (10,11) and hsc73 (Fig. 6). Whether aldolase and phosphoglucomutase are also substrates remains to be tested. The relative enrichment of other cytosolic proteins in HSCϪ lysosomes may indicate that they are not good substrates for the selective import pathway but that they can enter lysosomes by alternate means such as microautophagy or macroautophagy and then are relatively resistant to lysosomal proteolysis.
The physiological significance of the different sizes and shapes of HSCϩ and HSCϪ lysosomes is not known. We also do not know the significance of different protein bands in the membranes of HSCϩ and HSCϪ lysosomes. One possibility is that some of the lysosomal membrane proteins that are preferentially found in the HSCϩ lysosomes are components of the selective protein import pathway.
The low pH, the high hydrolytic activities and the absence of typical endosomal markers (41) in HSCϩ and HSCϪ lysosomes suggest that most organelles in these populations are mature lysosomes. It appears also that HSCϪ lysosomes have all the other components needed for the selective transport of cytosolic proteins because this population can be made competent for specific protein uptake by simply preloading it with hsc73. Accordingly, HSCϪ lysosomes, which would be inactive under basal and short starvation times, could be activated and recruited to participate in the selective pathway under certain conditions (e.g. prolonged starvation) by increasing hsc73 in their lumen.