Abstract
We used quantitative immunogold electron microscopy and biochemical analysis to evaluate the subcellular distribution of Hsp60 in rat tissues. Western blot analysis, employing both monoclonal and polyclonal antibodies raised against mammalian Hsp60, shows that only a single 60-kD protein is reactive with the antibodies in brain, heart, kidney, liver, pancreas, pituitary, spleen, skeletal muscle, and adrenal gland. Immunogold labeling of tissues embedded in the acrylic resin LR Gold shows strong labeling of mitochondria in all tissues. However, in the anterior pitutary and in pancreatic acinar cells, Hsp60 also localizes in secretory granules. The labeled granules in the pituitary and pancreas were determined to be growth hormone granules and zymogen granules, respectively, using antibodies to growth hormone and carboxypeptidase A. Immunogold labeling of Hsp60 in all compartments was prevented by preadsorption of the antibodies with recombinant Hsp60. Biochemically purified zymogen granules free of mitochondrial contamination are shown by Western blot analysis to contain Hsp60, confirming the morphological localization results in pancreatic acinar cells. In kidney distal tubule cells, low Hsp60 reactivity is associated with infoldings of the basal plasma membrane. In comparison, the plasma membrane in kidney proximal tubule cells and in other tissues examined showed only background labeling. These findings raise interesting questions concerning translocation mechanisms and the cellular roles of Hsp60.
Hsp60 (Cpn60) constitutes one of the major and well-characterized molecular chaperone proteins in prokaryotic and eukaryotic organisms, with essential functions in both stressed and nonstressed cells (Ellis and van der Vies 1991; Zeilstra-Ryalls et al. 1991; Craig et al. 1993). In bacteria, Hsp60 (GroEL) is involved in the proper folding and assembly into oligomeric complexes of other proteins, as well as their transport across the plasma membrane (Goloubinoff et al. 1989; Ellis and van der Vies 1991; Zeilstra-Ryalls et al. 1991; Craig et al. 1993). In eukaryotes, Hsp60 has been thought to be present and to function in protein folding only in organelles such as mitochondria and chloroplasts (Cheng et al. 1989; Roy 1989; Ellis and van der Vies 1991; Zeilstra-Ryalls et al. 1991; Craig et al. 1993), which are of endosymbiotic origin (Portier 1918; Wallin 1925; Margulis 1970; Gray 1992). Hsp60 is encoded by nuclear DNA (Jindal et al. 1989; Picketts et al. 1989; Reading et al. 1989) and is synthesized as a larger precursor form containing an N-terminal targeting sequence which is necessary for its mitochondrial import and is cleaved during maturation to the mature form in the mitochondrial matrix (Jindal et al. 1989; Singh et al. 1990). However, a variety of biochemical, immunological, and subcellular localization data have raised the possiblity that Hsp60 as a molecular chaperone may be essential for the biological functioning of other proteins at unexpected extramitochondrial locations (Soltys and Gupta 1999a,b,c). Detailed immunogold labeling studies in a variety of cultured mammalian cell lines have shown that up to 20% of Hsp60 immunoreactivity is present at discrete extramitochondrial sites, including foci on ER, on the cell surface, and in unidentified vesicles and cytoplasmic granules (Soltys and Gupta 1996,1997). One question that needs to be further addressed is whether Hsp60 in animal tissues also localizes at specific extramitochondrial sites. In view of the reported involvement of Hsp60 in autoimmune diseases and other pathological conditions (Kaufmann et al. 1991; van Eden 1991; Kaufmann 1992; Georgopoulos and McFarland 1993; Soltys and Gupta 1999a), detailed study of Hsp60 subcellular distribution in diverse animal tissues is needed. In this study, we examined the subcellular localization of Hsp60 in a variety of rat tissues by immunogold electron microscopy of LR Gold sections using a number of well-characterized monoclonal and polyclonal antibodies specifically raised against mammalian Hsp60 (Jindal et al. 1989; Singh and Gupta 1992). Our results show that although Hsp60 is primarily localized in mitochondria in many tissues, significant amounts of Hsp60 are detected at extramitochondrial sites in pituitary and pancreas, where we found Hsp60 specifically localized in growth hormone granules and zymogen granules, respectively. The presence of Hsp60 in zymogen granules was also confirmed biochemically using Western blot analysis of highly purified zymogen granule preparations. The findings provide evidence in tissues supporting the concept that certain mitochondrial matrix proteins may be shared in other compartments of the cell (Soltys and Gupta 1999b,c). The various possibilities to explain these observations are discussed.
Materials and Methods
Antibodies
The origin of various polyclonal and monoclonal antibodies (MAbs) to Hsp60 is as follows: P1-1 and P1-2 are rabbit polyclonal antibodies that were raised against Hsp60 (eluted from 2-D gel spots) from CHO cells (Gupta and Dudani 1987). The polyclonal antibody P1-3 was raised by immunizing a rabbit with human recombinant Hsp60 protein (PKK13D) (Singh and Gupta 1992) covering the C-terminal 70% (from a.a. 169-546) of the protein. The mouse MAb II-13 was raised against human recombinant Hsp60 protein PKK13A lacking the first 30 amino acids (Singh and Gupta 1992). For the electron microscopic studies, the Hsp60 antibodies were affinity-purified using human recombinant Hsp60 protein (PKK13D) adsorbed to nitrocellulose strips, as described previously (Singh et al. 1997). Other antibodies used include a rabbit polyclonal antibody against sheep growth hormone (ICN; Montreal, Quebec, Canada), a mouse MAb to carboxypeptidase A (Sigma; St Louis, MO), and a mouse MAb to cytochrome oxidase subunit IV (Molecular Probes; Eugene, OR).
Immunoelectron Microscopy
Rats were anesthetized with sodium pentobarbital and perfusion-fixed with freshly dissolved 4% paraformaldehyde in 100 mM sodium phosphate buffer, pH 7.4. Tissues were then excised, cut into small pieces, and postfixed with 0.5% glutaraldehyde in sucrose-cacodylate buffer. Procedures for the embedding and sectioning of cells in LR Gold resin (Polysciences; Warrington, PA) have been described (Soltys and Gupta 1996).
Antibody labeling of LR Gold sections was done using a conventional two-stage procedure with 20-nm gold markers as the secondary reagent. Sections were preabsorbed at room temperature with 20% fetal calf serum in 0.1 M Tris-HCl, pH 7.5 (carrier buffer). Sections were then reacted with affinity-purified polyclonal or monoclonal antibody to Hsp60 in carrier buffer for 1.5 hr at 37C in a humidified incubator. In antibody preadsoption controls, antibody to Hsp60 was reacted with human recombinant Hsp60 protein PKK13A at 30 μg/ ml for 2 hr at 37C before application of the antibody-antigen complex to sections. Washing of sections was for 30 min with 5% bovine serum albumin in 0.1 M Tris-HCl, pH 7.5. Sections were reacted with a 1:5 dilution of goat anti-rabbit or goat anti-mouse IgG-20-nm gold conjugate from British Bio-Cell (Cedarlane Laboratories; Hornby, Ontario, Canada).
For double label immunogold labeling of anterior pituitary growth hormone cells with Hsp60 and growth hormone antibodies, sections were labeled with the different antibodies on opposite sides (Bendayan 1982), as follows. One face of the tissue section was labeled with a 1:1000 dilution of anti-growth hormone antibody in carrier buffer followed by a 1:5 dilution of goat anti-rabbit IgG-20-nm gold conjugate (British BioCell) in carrier buffer for 4 hr at 37C. The other face of the tissue section was then labeled with affinity-purified polyclonal antibody to Hsp60 as above, then reacted with a 1:40 dilution of goat anti-rabbit IgG (Jackson Immunoresearch; West Grove, PA) in carrier buffer for 1 hr at 37C, washed again, then reacted with a 1:5 dilution of rabbit anti-goat IgG-10-nm gold conjugate in carrier buffer for 4 hr at 37C (Soltys and Gupta 1996).
After washing, including a high-salt wash with 0.5 M KCl in carrier buffer followed by washes with H2O, LR Gold sections were stained with 2% osmium tetroxide (15 min) followed by 2% uranyl acetate in 0.1 M maleate buffer, pH 6.0 (5 min). Sections were examined at 80 kV with a JEOL 1200 EX transmission electron microscope.
Quantitative Analysis of Immunogold Labeling
Immunogold labeling intensities in different subcellular compartments were determined by direct planimetry and counting of gold particles per μm2 using a Kontron MOP Video-plan (Carl Zeiss; Toronto, Ontario, Canada) as described previously (Bendayan 1982). A total of 10-15 different cells were evaluated for each tissue and the value given for each subcellular compartment is the mean ± SEM (n = total number of organelles). Labeling was considered significant at p< 0.05. In the analysis of plasma membrane-associated labeling, all gold particles found within 50 nm of either side of the membrane were scored. Thus, values represent the number of particles found within an area measuring 0.1 μm × the linear length of the membrane evaluated.
Purification of Zymogen Granules
Zymogen granules were purified according to described procedures (Cronshagen et al. 1994). Briefly, fresh bovine pancreas was separated from connective tissue and minced with a scalpel blade. The tissue was then homogenized in buffer containing 0.25 M sucrose, 5 mM 2-N-morpholino-ethanesulfonic acid (MES), 0.1 mM MgSO4, 1 mM phenylmethylsulfonyl fluoride (PMSF) with a glass homogenizer. The homogenate was then centrifuged at 500 × g for 10 min at 4C. The supernatant was then centrifuged at 2000 × g for 10 min at 4C, using the same centrifuge and rotor. The resulting supernatant was removed by suction, leaving a brownish layer of mitochondria on top of the white zymogen granule pellet. This mitochondrial layer was partially removed by gentle swirling twice with 0.3 M sucrose and was then discarded. The crude zymogen granule pellet was resuspended in 25 ml of ice-cold sucrose/Percoll buffer (250 mM sucrose, 40% Percoll, 50 mM MES, 100 mM MgSO4, 100 mM ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraaccetic acid (EGTA), 1 mM PMSF, 1 mM dithiothreitol, pH 6.5) and centrifuged at 100,000 × g for 25 min at 4C. The white band of zymogen granules was removed, diluted 10-fold in homogenization buffer, and centrifuged at 2000 × g for 10 min at 4C, yielding a pure zymogen granule pellet.
Gel Electrophoresis and Western Blots
Samples were electrophoresed in 10% SDS-PAGE as described previously (Soltys and Gupta 1996). Proteins were transferred electrophoretically from polyacrylamide gels to nitrocellulose sheets. Blots were blocked with 3% BSA in 0.9% NaCl, 10 mM Tris-HCl, pH 7.4, and this was also the carrier buffer for all antibodies. Blots were reacted with the following antibodies: 1:1000 P1-2 rabbit polyclonal antibody; 1:10,000 MAb to carboxypeptidase A; 1:10,000 MAb to cytochrome oxidase subunit IV. Visualization of polyclonal antibodies was with horseradish peroxidase-conjugated secondary antibody directed against rabbit IgG (BioRad Lab; Mississauga, Ontario, Canada) and color development with 4-chloro-1-naphthol (BioRad). Visualization of MAbs was with alkaline phosphatase-conjugated secondary antibody directed against mouse IgG (BioRad) and color development with bromochloroindoyl phosphate/nitroblue tetrazolium (BCIP/NBT) (BioRad).
Results
Detection of Hsp60 in Tissue Extracts by Western Immunoblotting
The monoclonal and polyclonal antibodies against Hsp60 used in this study were raised against Hsp60 purified from Chinese hamster ovary cells or against recombinant human Hsp60, and their exclusive specificity for Hsp60 has been previously established biochemically (Gupta et al. 1985; Gupta and Dudani 1987; Jindal et al. 1989; Singh and Gupta 1992). Western blot analysis of rat tissues (Figure 1) shows that only a single 60-kD protein is reactive with the antibodies in all tissues examined in this study, including pancreas, pituitary, kidney, cerebellum, heart, adrenal gland, liver, skeletal muscle, and spleen. Hsp60 protein levels show variation from tissue to tissue, with highest reactivity in liver. However, similar results were obtained for the mitochondrial marker cytochrome oxidase (subunit IV) (not shown), suggesting that the observed variation in Hsp60 protein levels from one tissue to another likely reflects known differences in mitochondrial content in different tissues.
Hsp60 is synthesized as a higher molecular weight precursor form containing an N-terminal targeting sequence that is cleaved after import into the mitochondrial matrix (Jindal et al. 1989; Picketts et al. 1989; Reading et al. 1989). The precursor form in cultured cell lines can be readily resolved on polyacrylamide gels from the mature form of Hsp60 as a higher molecular weight band (Soltys and Gupta 1996). The results in Figure 1, which shows only a single band corresponding to the mature protein in all tissues, indicate that the concentration of the precursor form of Hsp60 in rat tissues is too low to be detected by Western blot analysis.
Western immunblot detection of Hsp60 in rat tissues. (
Localization of Hsp60 in Pancreatic Zymogen Granules and Western Immunoblotting of Purified Granules
Rat tissues embedded in the acrylic resin LR Gold were sectioned and then probed with Hsp60 antibodies followed by immunogold markers. Similar results were obtained using both polyclonal and monoclonal antibodies against mammalian Hsp60. In pancreatic acinar cells, shown in Figure 2, Hsp60 was localized in both mitochondria and zymogen granules. Zymogen granules have a distinctive size and morphology and are easily distinguished from mitochondria. In separate experiments, these zymogen granules were also shown to be labeled with antibody against carboxypeptidase A (not shown). The results of quantitative immunocytochemical analysis of acinar sections (Figure 3) show that the Hsp60 labeling intensity in mitochondria is approximately sixfold higher in mitochondria than in zymogen granules (mean = 241 vs 15.5 gold particles/μm2, respectively). The low labeling observed served in other compartments, including the ER, cytosol, and nucleus (~10 particles/μm2), and in the Golgi apparatus, on the cell surface, and in the acinar lumen (not shown), was similar to the labeling of the plastic sections in regions without cells. Similar background labeling was also obtained in control experiments using (a) preadsorption of the primary Hsp60 antibody with recombinant Hsp60 before application to sections or (b) omission of the primary antibody (not shown). Thus, intense and highly specific Hsp60 immunogold labeling in pancreatic acinar cells is observed in both mitochondria and zymogen granules but not in other subcellular compartments.
Hsp60 subcellular localization in rat pancreatic acinar cells using immunogold electron microscopy. LR Gold sections were labeled with polyclonal antibody to Hsp60, followed by secondary antibody bound to 20-nm gold particles. ZG, zymogen granules; M, mitochondria; N, nucleus. Bar = 0.5 μm.
Quantitative analysis of Hsp60 immunogold labeling in pancreatic acinar cells. Mitochondria (M), 241.3 ± 15.5 gold particles/μm2 (n = 16); zymogen granules (ZG), 37.9 ± 3.1 gold particles/μm2 (n = 13); cytosol plus endoplasmic reticulum (C+ER), 9.0 ± 2.3 gold particles/μm2 (n = 10); nucleus (N), 11.1 ± 1.5 gold particles/μm2 (n = 2). p<0.05.
To obtain independent biochemical evidence that Hsp60 is present in pancreatic zymogen granules, zymogen granules were purified and analyzed by Western immunoblotting (Figure 4). Bovine rather than rat pancreas was used for these studies because tissue from only a single animal was required for large-scale purification of zymogen granules. In Figure 4A, Lane 2, anti-Hsp60 antibody specifically identifies a 60-kD protein in the purified zymogen granule fraction. This protein co-migrates with mature Hsp60 (not shown). A 60-kD protein is not detected in the Coomassie-stained gel in Figure 4, Lane 1, indicating that this protein is in low abundance and below the detection sensitivity of the Coomassie dye procedure. Figure 4, Lane 3 shows control labeling of the zymogen granule fraction with antibody against carboxypeptidase A, a zymogen granule marker. The purity of the zymogen granule fraction used in Figure 4A is shown by the Western blotting results in Figure 4B, which shows double-label antibody labeling of fractions at different steps in the zymogen granule purification using a combination of antibody against Hsp60 and antibody against mitochondrial cytochrome oxidase (subunit IV) (COX). COX, which serves as a mitochondrial marker, was present in a crude fraction known to contain both mitochondria and zymogen granules (Figure 4B, Lane 2) but no COX reactivity was present in the purified zymogen granule fraction (Figure 4B, Lane 1). Therefore, purified zymogen granules are free of mitochondrial contaminants but contain an Hsp60-related protein, which confirms the EM localization results showing Hsp60 reactivity in zymogen granules.
Western immunoblot analysis of purified zymogen granules. (
Localization of Hsp60 in Secretory Granules in the Anterior Pituitary
In the pituitary, EM localization (Figures 5, 6A, and 6B) revealed the presence of Hsp60 reactivity in secretory granules of growth hormone secretory cells of the anterior pitutary. Figure 5 is a low-magnification overview of a typical growth hormone secretory cell, showing Hsp60 labeling in both mitochondria (M) and growth hormone granules (GH). The identity of the labeled granules was determined by double immunogold labeling studies using antibody to growth hormone as the second label. Figure 6A shows that both the Hsp60 and growth hormone antibodies label the same granules. A mitochondrion is also noted in this field, which labels only with antibody to Hsp60. As in acinar cells, Hsp60 reactivity in growth hormone secretory cells is not found in ER, cytosol, or other compartments, in which reactivity is at background levels. Quantitative analysis of labeling in different compartments, as shown in the histogram in Figure 7, shows that the Hsp60 labeling intensity in growth hormone granules is approximately twofold higher than in mitochondria of the same cells (mean = 142 vs 60 particles/μm2, respectively). It should also be noted, however, that mitochondrial labeling intensity is approximately fourfold lower in growth hormone secretory cells than in the Figure 3 values for mitochondria in acinar cells (mean = 60 vs 241 particles/μm2, respectively), which may represent real differences in Hsp60 content or differences in the accessibility of Hsp60 epitopes in different structures.
Hsp60 subcellular localization in rat anterior pituitary using immunogold electron microscopy. Rabbit polyclonal antibody to Hsp 60 and secondary antibody bound to 20-nm gold particles were used. GH, growth hormone granules; M, mitochondria; N, nucleus. Bar = 0.5 μm.
That labeling of secretory granules in the pituitary is limited to growth hormone granules was suggested by inspection of different neighboring cells. In Figure 6B a growth hormone secretory cell (cell 1) is present in the upper half of the micrograph and all secretory granules in this cell are labeled. In the bottom half of the same micrograph is a second unidentified cell (cell 2) containing primarily unlabeled secretory granules. Note that the labeling intensity in mitochondria in cell 2 is comparable to that in mitochondria in the growth hormone secretory cell. There is some indication of differential Hsp60 labeling in different secretory granules in cell 2, which suggests that Hsp60 is restricted to a subpopulation of secretory granules in other secretory cell types. However, when both labeled and unlabeled granules in cell 2 are treated as the same compartment in quantitative analysis, the labeling of secretory granules in cell 2 is at background levels (not shown).
Hsp60 localization in growth hormone granules of the anterior pituitary gland. (
Quantitative analysis of Hsp60 immunogold labeling in the anterior pituitary. Twenty-nm gold markers used. Mitochondria (M), 60.3 ± 9.5 gold particles/μm2 (n = 15); growth hormone granules (GH), 142.2 ± 13.4 gold particles/μm2 (n = 2); cytosol and endoplasmic reticulum (C+ER), 4.0 ± 0.4 gold particles/μm2 (n = 9); nuclei, 3.0 ± 0.7 gold particles/μm2 (n = 6). p<0.05.
Plasma Membrane Localization of Hsp60 in Kidney Distal Convoluted Tubule Cells
Figure 8 shows the basal region of a kidney distal convoluted tubule cell, a cell abundant in mitochondria. Note the membrane profiles interdigitating between the mitochondria. These are invaginations of the plasma membrane. Mitochondria are believed to be involved with these lateral cell membranes in active transport of ions from the renal tubule fluid. The observed Hsp60 immunogold labeling is primarily in mitochondria. There is also a low level of Hsp60 reactivity at foci on or in proximity to infoldings of the plasma membrane (examples indicated by arrows). Quantitative measurements (Figure 9) show that anti-Hsp60 labeling of mitochondria in kidney distal tubule cells is similar to that in the anterior pituitary (Figure 7), with mitochondrial labeling intensity in both tissues ~60 gold particles/μm2. Therefore, Hsp60 expression at the level of individual mitochondria in these tissues appears comparable and approximately fourfold lower than in pancreatic acinar cells. Hsp60 labeling on or in close proximity to the plasma membrane in distal tubule cells (defined as labeling within 50 nm of the membrane on either side) was approximately fourfold lower than in mitochondria and fourfold higher than background labeling. As in other tissues, labeling observed in the ER, cytosolic, and nuclear compartments was at backgound levels. In comparison, in kidney proximal tubule cells there was similar intense labeling of mitochondria but no reactivity was found associated with the plasma membrane (not shown). In addition, no cell surface reactivity was found in pituitary, cerebellum, heart, adrenal gland, liver, skeletal muscle, and spleen (not shown).
Hsp60 localization in the rat kidney distal convoluted tubule. Rabbit polyclonal antibody to Hsp60 and secondary antibody bound to 20-nm gold particles were used. Hsp60 reactivity is found in mitochondria (M) and along infoldings of the plasma membrane (PM). N, nuclei. Bar = 0.5 μm.
Discussion
The results of our biochemical and immunogold EM labeling studies show that although Hsp60 in different rat tissues is primarily localized in the mitochondrial matrix compartment, in pituitary and pancreas high levels of Hsp60 reactivity are also found in growth hormone granules and zymogen granules, respectively. Purified zymogen granules free of mitochondrial contamination also were shown to contain a Hsp60-related protein by Western blot analysis, confirming the EM localization results. Procedures for the specific purification of growth hormone granules from the pituitary are not presently available. Previous studies in cultured cell lines indicated that Hsp60 is generally expressed at low levels on the cell surface and in endoplasmic reticulum (Soltys and Gupta 1996,1997). The present study of adult rat tissues found Hsp60 reactivity associated with the plasma membrane only in kidney distal convoluted tubule cells, and definitive localization in ER was not obtained in any tissue.
Quantitative analysis of Hsp60 immunogold labeling in kidney distal convoluted tubule. Mitochondria (M), 55.5 ± 8.5 gold particles/μm2 (n = 20); plasma membrane-associated (PM) (defined as labeling within 50 nm of either side of the membrane, an area measuring O.1 μm × the linear length of membrane), 14.5 ± 6.5 gold particles/μm2 (n = 3); cytosol (C), 4.0 ± 0.4 gold particles/μm2 (n = 2); nucleus (N), 4.0 ± 2.0 gold particles/μm2 (n = 3). p<0.05.
Evidence for an extramitochondrial localization of Hsp60 has been reported in previous studies of mammalian tissues. In pancreatic β-cells, Hsp60 reactivity has been observed in mature insulin secretory granules in addition to mitochondria (Brudzynski et al. 1992a,b). Hsp60 antibodies were found to specifically label the central core of mature insulin secretory granules, but no significant labeling was observed in immature secretory granules. A previous study of pancreatic acinar cells using polyclonal antibodies raised against homologues of Hsp60, cpn10, and mHsp70 from the photosynthetic bacterium Chromatium vinosum reported that all three Hsps are present all along the secretory pathway (Velez-Granell et al. 1994; Le Gall and Bendayan 1996). Although Hsp60 and GroEL show significant sequence homology (Gupta 1995,1996), the potential importance of these findings to secretory mechanisms, and to disease states in which Hsp60 has been implicated (Soltys and Gupta 1999a), has necessitated that the presence of Hsp60 at extramitochondrial sites in acinar cells be reexamined using antibodies specifically raised against mammalian Hsp60. Our present results show intense and highly specific Hsp60 labeling only in zymogen granules and in growth hormone granules, and not along the entire secretory pathway. The latter result is not a consequence of low labeling efficiency, as indicated by the following. The Hsp60 labeling intensities we obtain in mitochondria may well be the highest immunogold labeling intensities thus far reported for any mitochondrial protein. This includes the 200-500 gold particles per mitochondrion in cross-section we previously reported for cultured cells using 10-nm gold markers (Soltys and Gupta 1996) and the quantitative determinations reported here for various tissues using 20-nm gold markers. For pancreatic acinar cells, the 241 particles/μm2 in mitochondria in the present study represents ~25-fold higher Hsp60 labeling intensity in mitochondria than in a previous study using 10-nm gold-protein A markers by Velez-Granell et al. (1994), and our value of 40 particles/μm2 for zymogen granules represents ~1.5-fold higher labeling intensity than in the aforementioned study. Therefore, our lack of ER and Golgi labeling is not due to a low labeling efficiency technique. Other technical reasons may be contributing to the lack of ER and Golgi labeling in our studies of pancreatic and pituitary cells. Epitopes on the Hsp60 molecule while in transit through the ER and Golgi may be inaccessible, perhaps due to complex formation with other proteins, or are destroyed by fixative. Our use of perfusion-fixation, widely considered the method of choice, may interfere with the detection of Hsp60 in ER and Golgi if there is enhanced fixation in these compartments, compared with the results previously obtained by fixing sliced pancreas only after removal of the tissue from the animal (Velez-Granell et al. 1994). In addition, the levels of Hsp60 in transit through these compartments may indeed be exceedingly low and below our detection ability. An alternative explanation for our results, however, would be that Hsp60 bypasses the ER and Golgi and is directly imported into secretory vesicles. This latter possibility was previously proposed to explain Hsp60 localization specifically in mature insulin secretory granules (Brudzynski et al. 1992a,b). It is noteworthy that the antibody used by Velez-Granell et al. (1994) was raised against bacterial Hsp60 (GroEL) and reacted with more than one protein band in Western immunoblots of pancreatic tissue, which raises concerns about antibody specificity and makes it less certain that the low reactivity they detected in ER and the Golgi apparatus is due to Hsp60. How Hsp60 reaches secretory granules remains an open issue (Soltys and Gupta 1999b,c).
In other mammalian tissues examined to date, Hsp60 was present in rat liver in both mitochondria and peroxisomes using a variety of polyclonal and monoclonal antibodies (Velez-Granell et al. 1995; Soltys and Gupta 1996). In the case of cell surface expression of Hsp60 in tissues, it has been found that stressed aortic endothelial cells exposed to cytokines or high heat, but not control cells, express Hsp60 on their cell surface, as detected by fluorescence imaging, and are susceptible to complement-dependent lysis by Hsp60-specific antibodies (Xu et al. 1993a,b,1994; Schett et al. 1995,1997). Hsp60 may also be expressed on the cell surface of certain blood cells, including T-cells undergoing apoptosis (Poccia et al. 1996), on the surface of stressed macrophages (Koga et al. 1989), and in certain tumor cells (Fisch et al. 1990; Selin et al. 1992; Kaur et al. 1993; Fitzgerald and Keast 1994; Khan et al. 1998).
Results of the present study and a previous study from this laboratory (Soltys and Gupta 1996) have raised and dismissed the possibilities that extramitochondrial Hsp60 reactivity might be an artifact resulting from either (a) adventitious crossreaction of anti-bodies with unrelated proteins or (b) reaction with precursor Hsp60 that failed to enter mitochondria, rather than a reaction with mature protein. The reasons these possibilities are unlikely are summarized as follows. The possibility that the extramitochondrial labeling could be nonspecific is unlikely because several different monoclonal and polyclonal antibodies all gave similar results. These antibodies, which were raised against mammalian Hsp60, have been shown previously to react specifically with this protein in 1- and 2-D gel blots (Gupta et al. 1985; Gupta and Dudani 1987). They show no crossreactivity with the cytosolic TCP-1 protein (unpublished results), which is distantly related to the Hsp60 family of proteins (Gupta 1995). In immunoprecipitation experiments with different cells and tissues, these antibodies immunoprecipitate only Hsp60. Furthermore, as observed previously and in the present study, preadsorption of the antibodies with the purified recombinant human Hsp60 abolishs both mitochondrial and extramitochondrial labeling, indicating the specificity of the observed labeling. Therefore, the possibility that the extramitochondrial labeling could be due to crossreactivity with some other antigen is unlikely. With respect to whether the extramitochondrial Hsp60 is the precursor and not the mature form of Hsp60, previous experiments in CHO cells in which maturation of Hsp60 was inhibited using the ionophore nonactin, which blocks mitochondrial import and consequently prevents processing of the precursor to the mature form, the larger precursor form of Hsp60 was the only protein that was immunoprecipitated (Soltys and Gupta 1996). Together, these results strongly suggest that the only protein with which the antibodies react is mitochondrial Hsp60. The second possibility, that extramitochondrial Hsp60 labeling may result from a reaction with precursor Hsp60, is also excluded by the above-cited nonactin experiments and by the fact that under normal growth conditions precursor Hsp60 is not detected in cultured mammalian cells (Soltys and Gupta 1996). Western immunoblot results in different tissues (see Figure 1) also show only a single reactive band in all tissues, suggesting that significant levels of precursor Hsp60 are also not present in tissues and cannot account for the extramitochondrial labeling observed. The most likely possibility to explain the results of this investigation is that although most of the Hsp60 is localized in mitochondria, smaller amounts of this protein are also present at other cellular sites, including zymogen granules, growth hormone granules, and the plasma membrane of kidney distal tubule cells.
The finding of Hsp60 in secretory granules, including growth hormone granules, zymogen granules, and insulin secretory granules, suggests that certain cell types would secrete Hsp60. Hsp60 has been detected in exocrine pancreatic juice (Velez-Granell et al. 1994). Recently, it was found that a Hsp60-like protein is secreted by cultured neuroglial cells and by a neuroblastoma cell line (Bassan et al. 1998). Secretion of the Hsp60-like protein in neuroblastoma cells was increased in the presence of vasoactive intestinal peptide (VIP), a neurotransmitter and neuromodulator that induces cytokine release. The secreted Hsp60 could possibly have neuroprotective effects (Bassan et al. 1998). If that is correct, target cells may bind and internalize the secreted Hsp60.
Does extramitochondrial Hsp60 have physiological functions? By definition, molecular chaperones themselves do not have a direct function in cellular phenomena but rather facilitate the functions of other proteins by their effects on folding, transport, and insertion or translocation across membranes (Ellis 1987; Ellis and van der Vies 1991; Zeilstra-Ryalls et al. 1991; Ryan et al. 1997). Zymogen granules, growth hormone granules, and the condensed insulin core in insulin secretory granules all represent highly organized supramolecular structures that secrete functional hormones, enzymes, or insulin. Hsp60 as a chaperone might have a role in the condensation or packaging of secretory granule contents. The established role of Hsp60 in the formation of oligomeric protein complexes and in bacterial protein secretion (Ellis 1987; Zeilstra-Ryalls et al. 1991) suggests that the Hsp60 in these granules is involved in similar functions. In addition, Hsp60 may be involved in transporting secreted macromolecules to target tissues.
The finding of Hsp60 at extramitochondrial locations remains unexplained in terms of protein targeting and translocation mechanisms. There is no evidence for the existence of more than one Hsp60 gene or for alternate splicing of the mRNA for this gene product. These possibilities are also not supported by the results of the nonactin experiment in CHO cells, which clearly showed that on abolishment of the mitochondrial membrane potential only the precursor form of Hsp60 accumulates in cells (Soltys and Gupta 1996). The failure to see any other form of Hsp60 in this experiment indicates that processing of the precursor form occurs only in mitochondria and not in any other subcellular compartment, and suggests that the mature extramitochondrial Hsp60 that is detected in localization studies is derived from the same precursor protein which is initially imported and processed in mitochondria. After export from mitochondria, Hsp60 may then enter the normal secretory pathway or may be directly imported into secretory vesicles. However, there are presently no known mechanisms for the export of proteins from the mitochondrial matrix compartment to other cellular destinations. Various hypothetical mechanisms for the possible export of mitochondrial matrix proteins from mitochondria have been discussed in detail elsewhere (Soltys and Gupta 1999a,b,c).
The findings reported here support and extend to tissues a variety of recent evidence indicating that certain mitochondrial matrix proteins may also be shared in other compartments of the cell (Soltys and Gupta 1999b,c). A central implication of the subcellular localization studies and the cumulative evidence pointing to an involvement of Hsp60 in diverse cellular processes is that specific mechanisms must exist for the export of Hsp60 from mitochondria and that this protein also probably has important functions at specific extramitochondrial sites. Characterization of the trafficking pathways and the role of Hsp60 at extramitochondrial sites will be of great interest.
Footnotes
Acknowledgements
Supported by a grant from the Medical Research Council of Canada.