Histochem Cell Biol (2005) 124: 409–421 DOI 10.1007/s00418-005-0056-2
O R I GI N A L P A P E R
Skanda K. Sadacharan Æ Bhag Singh Timothy Bowes Æ Radhey S. Gupta
Localization of mitochondrial DNA encoded cytochrome c oxidase subunits I and II in rat pancreatic zymogen granules and pituitary growth hormone granules Accepted: 29 July 2005 / Published online: 27 August 2005 Springer-Verlag 2005
Abstract Cytochrome c oxidase (COX) complex is an integral part of the electron transport chain. Three subunits of this complex (COX I, COX II and COX III) are encoded by mitochondrial (mit-) DNA. High-resolution immunogold electron microscopy has been used to study the subcellular localization of COX I and COX II in rat tissue sections, embedded in LR Gold resin, using monoclonal antibodies for these proteins. Immunofluorescence labeling of BS-C-1 monkey kidney cells with these antibodies showed characteristic mitochondrial labeling. In immunogold labeling studies, the COX I and COX II antibodies showed strong and specific mitochondrial labeling in the liver, kidney, heart and pancreas. However, in rat pancreatic acinar tissue, in addition to mitochondrial labeling, strong and specific labeling was also observed in the zymogen granules (ZGs). In the anterior pituitary, strong labeling with these antibodies was seen in the growth hormone secretory granules. In contrast to these compartments, the COX I or COX II antibodies showed only minimal labeling (five- to tenfold lower) of the cytoplasm, endoplasmic reticulum and the nucleus. Strong labeling with the COX I or COX II antibodies was also observed in highly purified ZGs from bovine pancreas. The observed labeling, in all cases, was completely abolished upon omission of the primary antibodies. These results provide evidence that, similar to a number of other recently studied mit-proteins, COX I and COX II are also present outside the mitochondria. The presence of mit-DNA encoded COX I and COX II in extramitochondrial compartments, provides strong evidence that
S. K. Sadacharan Æ B. Singh T. Bowes Æ R. S. Gupta (&) Department of Biochemistry and Biomedical Sciences, McMaster University, Hamilton, L8N 3Z5, Canada E-mail:
[email protected] Tel.: +1-905-525-9140 Fax: +1-905-522-9033
proteins can exit, or are exported, from the mitochondria. Although the mechanisms responsible for protein exit/export remain to be elucidated, these results raise fundamental questions concerning the roles of mitochondria and mitochondrial proteins in diverse cellular processes in different compartments. Keywords Immunogold labeling Æ Electron microscopy Æ Mitochondrial proteins Æ Zymogen granules Æ Growth hormone granules Æ Subcellular localization Æ Protein export
Introduction Cytochrome c oxidase (COX) or complex IV is present in the inner mitochondrial membrane and carries out the terminal step in the electron transport chain, i.e. transferring electrons from cytochrome c to oxygen (Scheffler 2001). The transfer of electrons is accompanied by the vectorial pumping of protons from the matrix to the intermembrane space. Complex IV is composed of 13 protein subunits that contain heme and copper, ten of which are encoded by nuclear DNA. The three largest subunits of the COX complex viz., COX I, COX II and COX III, are transcribed and translated within the mitochondria and form the catalytic core of complex IV (Schon 2000; Scheffler 2001). Mutations in COX I, COX II and COX III give rise to a variety of unrelated disorders, such as sporadic anemia, sporadic myopathy and encephalomyopathy (Schon 2000). Mutations in nuclear encoded proteins involved in the assembly of complex IV are also known to occur (Wallace 1999; Shoubridge 2001). Mutation in SURF1, which is involved in the assembly of COX I, COX II and COX III subunits, results in the Leigh syndrome, which is characterized by a degeneration of the brain stem and basal ganglia, with elevated lactic acid levels (Shoubridge 2001). Mutation in SCO2, which is involved in the insertion of copper centers into subunits, results in cardioencephalomyopathy (Shoubridge 2001). Mutation in COX 10, which is
410
involved in heme formation, results in leukodystrophy and tubulopathy (Schon 2000; Shoubridge 2001). In recent years, our lab, as well as those of others, have shown that a large number of proteins, which were previously believed to be present solely in the mitochondria (e.g., 60 kDa heat shock chaperone protein (Hsp60), Hsp10 or chaperonin 10, mit-Hsp70 or mortalin, P32 protein, cytochrome c, aspartatae aminotransferase (mAspAT), tumor necrosis factor receptor associated protein 1 (TRAP-1), adenylate kinase, etc.), are also present at discrete extramitochondrial locations such as the plasma membrane and other types of membrane enclosed secretory granules (Brudzynski et al. 1992; Velez-Granell et al. 1994; Soltys and Gupta 1996; Singh et al. 1997; Cechetto and Gupta 2000; Cechetto et al. 2000, 2002; Felts et al. 2000; Ran et al. 2000; Soltys et al. 2000, 2001; Brokstad et al. 2001; Sadacharan et al. 2001). Several of these mitochondrial proteins (viz., Hsp60, Hsp10, mAspAT, P32 and TRAP1) also carry out novel functions, which are unrelated to their known mitochondrial functions, at these locations (Morton et al. 1992; Wadhwa et al. 1993; Jones et al. 1994; Soltys and Gupta 1999; Bradbury and Berk 2000; Felts et al. 2000; Ghebrehiwet et al. 2001). Most of these proteins are encoded by a single copy nuclear gene and, for many of them, other possibilities such as alternate transcription, alternate mRNA translation or splicing, have been ruled out (Soltys and Gupta 1996, 1999; Singh et al. 1997; Bradbury and Berk 2000; Sass et al. 2001; Mueller et al. 2004). It has also been established that, for many of these proteins, the proteins found at extramitochondrial sites are the mature form of those proteins that lack the mitochondrial targeting sequence (MTS). The cleavage of the MTS is specifically carried out by a mitochondrial matrix resident protease (MMRP) (Soltys and Gupta 1999; Mueller et al. 2004). To account for the presence of these proteins at extramitochondrial sites, two possible mechanisms have been suggested: (1) the partial import of the protein into the mitochondria, so that their MTS is cleaved by the MMRP, followed by retrograde translocation of the protein into the cytosol (Stein et al. 1994; Sass et al. 2001) and (2) the full import of the protein into the mitochondria, followed by their export to other locations by, as of yet, unknown mechanisms (Poyton et al. 1992; Soltys and Gupta 1999, 2000). Since all the nuclear encoded proteins are translated in the cytosol before they are imported into the mitochondria, it has proven difficult to exclude the possibility of retrograde translocation in their extramitochondrial localizations. Mitochondrial (mit-) DNA encodes 13 proteins, which are transcribed and translated within the mitochondria (Scheffler 2001). If these proteins are found to be present at specific extramitochondrial sites, this will exclude the possibility of retrograde translocation as well as other non-specific possibilities and also provide compelling evidence for the existence of specific mechanisms by which proteins can exit from mitochondria. Therefore, the present study has used
the high resolution immunogold electron microscopy technique to determine the subcellular localization of COX I and COX II, which are encoded by mit-DNA, in a variety of normal rat tissue sections, embedded in LR Gold resin, using specific antibodies to these proteins. Our results indicate that in many tissues such as the liver, kidney and heart, COX I and COX II are mainly localized within mitochondria. However, in rat pancreatic acinar cells and the anterior pituitary, strong and specific labeling due to these antibodies was also observed in zymogen granules (ZGs) and the growth hormone secretory granules. In a recent study, the mit-DNA encoded NADH dehydrogenase subunit 2 (ND2) was also shown to be present in postsynaptic densities in the rat brain, where it acted as an adaptor for the Src protein (Gingrich et al. 2004). The presence of these mit-DNA encoded proteins at specific extramitochondrial sites provides strong evidence that these proteins can exist out of the mitochondria and also perform functions at other subcellular locations. These results raise important questions regarding the cellular functions of mitochondria and mitochondrial proteins, which have important implications for mitochondrial diseases (Wallace 1999; Schon 2000; Shoubridge 2001).
Materials and methods Antibodies Mouse monoclonal antibodies, against the human COX subunit I (Cat. No. A6403, Clone number 1 D6-E1-A8) and COX II (Cat. No. A6404, Clone 12C4-F2), were purchased from Molecular Probes (Eugene, OR). Mouse monoclonal and rabbit polyclonal antibodies to the human Hsp60 protein were raised in our earlier work (Soltys and Gupta 1996). The sources of other antibodies used in this work are as follows: rabbit polyclonal antibody against human a-amylase (Sigma; St. Louis, MO); rabbit polyclonal antibody against sheep growth hormone (ICN; Montreal, Quebec, Canada). For immunoelectron microscopy, secondary goat anti-rabbit and goat anti-mouse IgG 20 nm gold conjugate antibodies were purchased from British BioCell (Hornby, Ontario). For immunofluorescence microscopy, secondary fluorescein-conjugated goat anti-mouse or goat anti-rabbit IgG was purchased from Jackson ImmunoResearch Labs (West Grove, PA). Immunofluorescence labeling of cells was carried out in a manner similar to what has been described earlier (Soltys and Gupta 1992). Briefly, African green monkey kidney BS-C-1 cells (ATCC CCL-26) were grown on sterile coverslips for 48 h, so that they attained a flattened morphology. The coverslips were briefly washed in Tris buffered saline (TBS) (0.9% NaCl, 10 mM Tris– HCl and pH 7.5) and the cells were fixed in ice-cold methanol at 20C for 20 min and then briefly washed in TBS. Following fixation, the cells were blocked with
411
50% normal goat serum in 0.1 M Tris–HCl (pH 7.5) and then incubated with 1:100 dilution (10 lg/ml) of COX 1 and COX II antibodies, or 1:40 dilution of rabbit polyclonal Hsp60 antibody, for 45 min at 37C in a humidified chamber. After this primary incubation was over, the cover slips were washed with three changes of TBS for 15 min. The cells were then labeled with the secondary fluorescein conjugated goat anti-mouse (or anti-rabbit) IgG diluted 1:100 for 45 min at 37C in a humidified chamber. After washing the coverslips with TBS and then rinsing them briefly with water, they were mounted on a slide containing a small drop of mounting medium (1 mg/ml paraphenyldiamine, in 90% glycerol and 100 mM Tris, pH 8.0). The labeled cells were examined as described earlier.
Immunoelectron microscopy Spraque–Dawley rats (Charles River Labs, Wilmington, MA) were anesthetized with sodium pentobarbital and perfusion fixed with 4% paraformaldehyde in 100 mM sodium phosphate buffer (pH 7.4) that was freshly made before use. Tissues were excised, kept on ice, cut into 1 mm cubes, washed three times with 0.1 M sucrose, 0.1 M maleate buffer (pH 6.0), postfixed for 30 min in 1% uranyl acetate in the same buffer, and then followed by 4% uranyl acetate in 50% ethanol for 30 min. Serial dehydration was carried out with ethanol in the order of 70, 90, 95 and 100%, with each step lasting 10 min. The rest of the procedure for embedding and sectioning of cells in LR Gold resin was done as described earlier (Soltys and Gupta 1996; Sadacharan et al. 2001; Cechetto et al. 2002). For labeling of rat tissue sections embedded in LR Gold resin, the sections were first wetted in a carrier buffer of 0.1 M Tris–HCl (pH 7.5) for 10 min, and then preabsorbed at room temperature with 20% fetal calf serum in the carrier buffer. the sections were then incubated with 1:40 (25 lg/ml) diluted COX I or COX II antibody in the carrier buffer for 2 h at 37C, in a humidified chamber. The antibody to human amylase was used at a dilution of 1:500. Following this, the sections were washed with three changes of 1% BSA for 10 min. The sections were then incubated with the secondary goat anti-mouse (or antirabbit) IgG 20 nm gold conjugate, at a 1:10 dilution in the carrier buffer for 4 h at 37C, in a humidified chamber. The control labeling was carried out in an identical manner, except that the step involving incubation with the primary antibodies was omitted. Following this, the sections were washed with 0.5 M KCl in the carrier buffer and quickly rinsed in water. The sections were then stained with 4% uranyl acetate in 25% ethanol for 20 min, rinsed briefly with 25% ethanol and then airdried. The sections were examined at 80 kV with a JEOL 1200 EX transmission electron microscope. Quantitation of the gold labeling, in different subcellular compartments, was carried out by direct planimetry and counting
the number of gold particles per square micrometer in 5– 10 sections as described in earlier works (Cechetto and Gupta 2000; Cechetto et al. 2000).
Purification of ZGs from bovine pancreas ZGs were isolated from bovine pancreas, with modifications to the previously published protocol (Thevenod et al. 1990). Fifty grams of bovine pancreatic tissue was immersed in an ice cold homogenization buffer containing 250 mM sucrose, 50 mM MOPS (pH 7.0), 0.1 mM MgSO4, 0.1 mM EGTA, 1 mg/ml defatted BSA, 0.2 mM PMSF (added just before use from a 100 mM stock solution in ethanol), 1 mM benzamidine and 0.01% soybean trypsin inhibitor. The pancreas was cut into small pieces with a scissor, discarding any fat and connective tissue. The small pieces of pancreas were suspended in 15% (wt/vol) of an ice-cold homogenization buffer and homogenized using a Tissuemizer for 10 s at 1,900 rpm. The tissue was subsequently homogenized using a motorized Teflon glass homogenizer (10 strokes at about 500 rpm). The homogenate was then centrifuged for 10 min at 500· g at 4C, to remove unbroken cells, debris and nuclei. The resulting supernatant was further centrifuged for 15 min at 1,500· g at 4C. The resulting brownish layer of mitochondria, on top of the white zymogen granule pellet, was gently removed by swirling two times with 1 ml of the homogenization buffer and saved for western blotting. The crude zymogen granule pellet was resuspended in the Percoll–sucrose–MES gradient (40% Percoll, 250 mM sucrose, 50 mM MES, pH 6.2, 2 mM EGTA, 0.2 mM MgSO4, 1 mg/ml defatted BSA, 0.2 mM PMSF, 1 mM benzamidine and 0.01% soybean trypsin inhibitor), at 20 ml/g of tissue, and centrifuged at 20,000· g for 20 min at 4C. The ZGs were obtained as a white band at the bottom of the tube. The ZGs were diluted tenfold in a wash buffer (250 mM sucrose, 50 mM MES, pH 6.2, 2 mM EGTA, 0.2 mM MgSO4, 1 mg/ ml defatted BSA, 0.2 mM PMSF, 1 mM benzamidine and 0.01% soybean trypsin inhibitor), washed twice, and loaded again onto another Percoll–sucrose–MES gradient, centrifuged, washed twice and carried through one more cycle of Percoll–sucrose–MES gradient centrifugation. The ZGs pellet from the final run was suspended in 125 mM Tris–HCl (pH 6.8) and employed for further analysis. Immediately after purification, the ZGs were fixed in 4% paraformaldehyde in a 125 mM MES buffer (pH 6.5) overnight, at 4C. The granules were then postfixed in uranyl acetate, dehydrated in ethanol, embedded in LR Gold resin and sectioned as described before (Soltys and Gupta 1996). The purity of the ZGs was determined by electron microscopy and the sections from these granules were then used for labeling with COX I and COX II antibodies.
412
Results The COX I and COX II antibodies used were purified mouse monoclonal IgGs antibody against human antigens. The specificities of these antibodies have been previously established by western immunoblot analysis, immunofluorescence and immunohistochemistry (Taanman et al. 1993; Capaldi et al. 1995). In western blots of human cell lines, COX I and COX II react specifically, with one band corresponding to 40 and the other to 25 kDa, respectively. The subcellular localization of cross-reactive proteins to these antibodies was initially examined by the immunofluorescence labeling of cultured African green monkey kidney BS-C-1 cells. Both COX I and COX II antibodies showed a punctate, beadstring shaped, labeling pattern, which is typical of mitochondrial labeling (Fig. 1). Similar immunofluorescence labeling patterns with these antibodies have been previously reported with human cell lines (Taanman et al. 1996). Based on immunofluorescence studies, there was no indication of significant extramitochondrial labeling with these antibodies. Subcellular localization of COX I and COX II in normal rat tissues Subcellular localization of COX I and COX II, in rat tissue sections, was studied in detail using immunogold electron microscopy. The advantage this method has is that it examines, at a very high resolution, the subcellular location of the proteins directly, as it exists in a normal cell or tissue. For this purpose, rat tissue sections embedded in LR Gold resin were probed with mono-
Fig. 1 Immunofluorescence labeling of cultured BS-C-1 cells with (a) mouse monoclonal antibody to COX I, (b) mouse monoclonal antibody to COX II and (c) rabbit polyclonal antibody to Hsp60
clonal antibodies to COX I and COX II, followed by detection with secondary antibodies conjugated to 20 nm gold particles. The results of these studies, with various rat tissue sections, are as follows. Liver In the liver, both COX I and COX II antibodies were found to strongly label mitochondria (Fig. 2). In comparison to mitochondria, the labeling intensity in the nucleus and the ER/cytoplasm was found to be four- to eightfold lower (Table 1).
Kidney Figure 3 shows the observed labeling with COX I and COX II antibodies in rat kidney sections. Similar to the liver, in tissue sections from the distal convoluted tubule region of the rat kidney, a strong labeling of elongated mitochondria, which is typical in this region of the kidney (Bulger 1988), was seen. In addition to mitochondria, strong labeling with COX I antibody was also observed in dense structures that correspond to the condensing vacuoles (Fig. 3a). The condensing vacuoles arise from the endocytotic apparatus and their contents eventually end up in lysosomes (Bulger 1988). In our earlier work, strong labeling of the condensing vacuoles was also observed with antibodies to the mAspAT. In contrast to mitochondria and the condensing vacuoles, the labeling densities of COX I or COX II antibodies in the nucleus and ER/cytoplasm were found to be five- to sevenfold lower (Table 1) providing evidence of their specificity.
413 Fig. 2 Subcellular localization of COX I (a) and COX II (b) in rat liver sections. Labeling is seen mainly in mitochondria, with labeling in all other compartments at near background levels. Immunogold markers conjugated to 20 nm gold particles were used. Bars in all electron micrographs are 500 nm. M mitochondria, N nucleus, ER endoplasmic reticulum
Table 1 Quantification of immunogold labelings with COX I and COX II antibodies
COX I cgrid COX II
a b
Liver Kidney Pancreas Liver Kidney Pancreas
Mitochondria
Nucleus
ER/cytoplasm
Zymogen granules
25.0a ±0.767b 15.4±2.41 36.1±4.93 72.6±4.53 59.4±7.72 62.0±1.89
3.16±0.601 2.33±0.581 2.79±0.365 4.58±0.221 3.41±0.875 3.31±0.596
7.07±1.20 3.27±0.627 7.66±0.767 14.9±0.418 17.6±1.22 16.1±2.23
N/A N/A 63.4±9.07 N/A N/A 88.2±11.6
Normalized mean gold particle density (particles/lm2). Standard error
Heart
Pancreas
In the heart sections, the labeling with COX I and II antibodies was, again, mainly limited to the mitochondria. The results of these antibodies are presented in Fig. 4. In contrast to the mitochondria, only minimal labeling was observed in the rest of the cardiac sarcomeres, nucleus and cytoplasm. A few gold particles were seen on the Z-lines separating the sarcomeres, but their significance is unclear.
In rat pancreatic acinar cells, labeling with COX I and COX II antibodies was not limited to the mitochondria (Fig. 5). In this case, in addition to the mitochondria, strong labeling was also observed in the ZGs. Additionally, these antibodies also labeled the condensing vacuoles, which are precursors to the ZGs. Quantitation of the observed labeling indicated that the labeling intensity for both COX I and COX II antibodies was
414 Fig. 3 Immunogold localization of COX I (a) and COX II (b) in rat distal convoluted tubule region of the kidney. In addition to mitochondria, dense labeling of condensing vacuoles (CV) is also seen
significantly higher in the ZGs when compared to the mitochondria (Table 1). In contrast to mitochondria and the ZGs, very little labeling was seen in the nucleus and the labeling density in the ER/cytoplasm was also several fold lower. It is unclear whether any of the labeling, in the latter case, was specifically present in the ER. All the known proteins in the ZGs are derived through the secretory pathway, which involve their passage through ER and Golgi compartments (Jamieson and Palade 1967). To determine that the observed lack of labeling of ER, with COX I and COX II antibodies, was not caused by any experimental conditions, labeling of the pancreatic sections, with a known ZGs resident protein, amylase, was carried out (Jamieson and Palade 1967; Palade 1975). The results of these studies are presented in Fig. 6b. As expected, in this case, strong labeling of the entire secretory pathway, as well as dense labeling of the ZGs, was observed. In this case, the cytoplasmic labeling was primarily localized in the ER and very little labeling was seen in the mitochondria, nucleus or the rest of the cytoplasm. These results provide strong evidence that the observed labeling
patterns, with the various antibodies, are specific. In all the experiments described above, parallel controls, where the primary antibody was omitted during labeling, were included. The omission of the primary antibodies in all instances led to the complete abolishment of labeling in all compartments (see Fig. 6a), indicating that the observed labeling was specific and not caused by the reactivity of the immunogold markers with the acrylic resin in which the tissue was embedded. Pituitary In the rat anterior pituitary tissue sections, strong labeling with both COX I and COX II antibodies was observed in the growth hormone granules, with only background labeling seen in the nucleus and the cytoplasm (Fig. 7). These labeled granules have been identified as growth hormone granules, based on their reactivity with an antibody to the growth hormone in our earlier work (Cechetto et al. 2000). The labeling pattern within the granules was heterogeneous, with
415 Fig. 4 Immunogold localization of COX I and COX II antibodies in rat heart sections. Labeling is mainly limited to mitochondria, with little labeling seen in nucleus or sarcomeres (S)
certain regions being more intensely labeled than others. In contrast to the strong labeling of growth hormone granules with the COX I and COX II antibodies, mitochondria in the pituitary sections were very poorly labeled with these antibodies. These results are similar to those seen previously with antibodies to Hsp60, Hsp10 and cytochrome c, where the labeling intensity of the growth hormone granules was more than twofold higher than that observed for mitochondria (Cechetto et al. 2000; Sadacharan et al. 2001; Soltys et al. 2001). Immunogold labeling studies with purified ZGs In view of the strong labeling of the ZGs with the COX I and COX II antibodies, it was of interest to determine whether these antibodies also reacted with the purified ZGs. To investigate this, ZGs were purified from bovine pancreas, using three cycles of Percoll–sucrose–MES gradient centrifugation (Thevenod et al. 1990). Due to their higher density, in comparison with the mitochondria, the ZGs are clearly separated from the mitochondria by this method. The purity of the ZGs preparation thus obtained was assessed by means of electron microscopy and it was found to contain
