J Biol Chem, Vol. 274, Issue 34, 24014-24022, August 20, 1999

Dual Targeting Property of the N-terminal Signal Sequence of P4501A1
TARGETING OF HETEROLOGOUS PROTEINS TO ENDOPLASMIC RETICULUM AND MITOCHONDRIA^*

Shripad V. Bhagwat^§, Gopa Biswas, Hindupur K. Anandatheerthavarada, Sankar Addya, William Pandak^¶, and Narayan G. Avadhani

From the  Department of Animal Biology and the Mari Lowe Center for Comparative Oncology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6047 and ^¶ Section of Gastroenterology, McGuire Veterans Administration Medical Center, and Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298-0711

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Recent studies from our laboratory showed that the -naphthoflavone-inducible cytochrome P4501A1 is targeted to both the endoplasmic reticulum (ER) and mitochondria. In the present study, we have further investigated the ability of the N-terminal signal sequence (residues 1-44) of P4501A1 to target heterologous proteins, dihydrofolate reductase, and the mature portion of the rat P450c27 to the two subcellular compartments. In vitro transport and in vivo expression experiments show that N-terminally fused 1-44 signal sequence of P4501A1 targets heterologous proteins to both the ER and mitochondria, whereas the 33-44 sequence strictly functions as a mitochondrial targeting signal. Site-specific mutations show that positively charged residues at the 34th and 39th positions are critical for mitochondrial targeting. Cholesterol 27-hydroxylase activity of the ER-associated 1-44/1A1-CYP27 fusion protein can be reconstituted with cytochrome P450 reductase, but the mitochondrial associated fusion protein is functional with adrenodoxin + adrenodoxin reductase. Consistent with these differences, the fusion protein in the two organelle compartments exhibited distinctly different membrane topology. The results on the chimeric nature of the N-terminal signal of P4501A1 coupled with interaction with different electron transport proteins suggest a co-evolutionary nature of some of the xenobiotic inducible microsomal and mitochondrial P450s.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cytochrome P450 monooxygenases (P450s)¹ are widely distributed in the biological kingdom and constitute a superfamily of heme proteins, which catalyze the metabolism of drugs, toxic chemicals, and an array of physiological substrates (1-3). In metazoan organisms, P450s involved in the physiological pathways, and also the inducible P450s, implicated in the xenobiotic metabolism are predominantly localized in the ER (routinely referred to as microsomal P450), as well as mitochondria of liver and a number of extrahepatic tissues (1-7). Models on the topological organization of microsomal P450s suggest that the heme protein is anchored on the ER membrane through a single transmembrane domain, with most part of the protein facing the cytosolic side (8-11). It is believed that only a short N-terminal segment (5-10 amino acid residues) of the P450 faces the ER lumen. Some studies also suggest additional extrinsic interaction of cytoplasmic exposed P450 domains with the lipid bilayer (12). The N-terminal 35-40 residues of different microsomal P450s are thought to provide the transmembrane anchor, in addition to stop transfer signal for membrane insertion (9, 13, 14). Membrane topology of the mitochondrial P450s remains relatively less clear. Based on the apparent lack of an N-terminal transmembrane helical domain in the constitutively expressed P450Scc and P450c27 (15, 16), and solubility of the protein in alkaline Na₂CO₃ (17, 18), the mitochondrial P450s are thought to be organized in a membrane extrinsic topology.

Proteins targeted to the ER and mitochondria contain distinct targeting signals, differ with respect to translational sites, and follow different targeting pathways (19-24). Proteins targeted to mitochondria are translated on cytoplasmic free ribosomes and post-translationally transported to mitochondria. The N-terminal cleavable or uncleaved amphipathic signal sequence binds to the mitochondrial outer and inner membrane translocase complexes (TOMs and TIMs, respectively) through a pathway involving the ATP-dependent Hsp70 family chaperones (19, 23). A recent study also suggested that members of the TOM complex provide the aqueous channel for protein transport across the mitochondrial outer membrane (25). The ER-targeted proteins, on the other hand, contain distinct N-terminal hydrophobic signals for binding to signal recognition particle, which in turn targets the emerging nascent chains to the ER (19, 20, 26, 27). With few exceptions, protein targeting to the ER is thought to be co-translational (reviewed in Ref. 20). Thus, the prevailing view is that the mitochondrial or ER destination of a protein is determined at the pre-translation stage by virtue of its signal sequence property. In some cases, where the same gene products are targeted to the ER and mitochondria, the 5' end of the mRNA sequence is altered by differential transcription using alternate transcription start sites or by differential splicing to introduce mitochondrial specific targeting signals (20, 28-31).

In contrast to the prevailing dogma, recent results in our laboratory showed that the BNF-inducible hepatic P4501A1 contains a chimeric N-terminal signal, which facilitates its targeting to both the ER and mitochondria (32, 33). The mitochondrial targeted P450, exhibiting high erythromycin N-demethylase activity, was shown to render protection against erythromycin-mediated inhibition of translation in transfected COS cells (34). Our results therefore suggest the existence of a novel pathway by which a primary translation product is targeted to two different cytoplasmic organelles. The mitochondrial targeting of P4501A1 involves an N-terminal endoprotease cleavage of the protein to activate a cryptic mitochondrial targeting signal. In the present study, the chimeric nature of the N-terminal signal sequence of P4501A1 (residues 1-44) was further investigated by testing the ability of this motif to target heterologous proteins, such as DHFR and mature portion of the mitochondrial P450c27, lacking the mitochondrial targeting signal, to ER and mitochondria. In addition to providing insights on the nature of the chimeric signal, our results also demonstrate that the microsomal targeted fusion protein functionally interacts with the microsome-specific electron transport protein, P450 reductase, whereas the mitochondrial targeted fusion protein interacts with Adx + Adr. These latter results shed new insights on the co-evolution of enzyme systems for mitochondrial and microsomal drug metabolism.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Construction of Expression Plasmids-- Various deletion and point mutations and also N-terminal or C-terminal fusions were generated by overlap polymerase chain reaction of parent full-length P450c27 cDNA (35) and P4501A1 cDNA (36) using primers containing 5' HindIII and 3' XbaI sites. Resultant cDNAs were cloned in the mammalian expression vector pCMV4 (37) and also pGEM7zf plasmids. In all cases the cDNA constructs were engineered to contain a Kozak consensus sequence, immediately upstream of the initiator ATG codon, for efficient translation (38). The two point mutations, namely R34D and K39I, were also generated by overlap polymerase chain reaction as described before (32). Sequence properties of all the plasmid constructs were ascertained by DNA sequencing. PET-21b-porin cDNA clone encoding yeast porin was provided by Dr. D. Pain.

In Vitro Transport of Labeled Proteins into Rat Liver Mitochondria-- Mitochondria were isolated from rat liver as described (39) and used for in vitro protein transport using a system modified from Gasser et al. (40). Various cDNA constructs in pGEM7zf plasmid vector (1-2 µg of circular DNA) were used as templates for generating ³⁵S-labeled translation products in TNT reticulocyte lysate system (Promega Corp., Madison, WI). Translation reactions were carried out in the presence of [³⁵S]Met (40 µCi/50-µl reaction, 1,000 Ci/mmol; NEN Life Science Products) using the protocol recommended by the manufacturer. Protein import was carried out essentially as described (32, 41). The reaction mixture (final volume of 200 µl) contained 4 µl of ³⁵S-labeled translation products (10⁵ cpm), 500 µg of freshly isolated mitochondria from a 10 mg/ml suspension in sucrose/mannitol buffer, 60 µl of energy mix (10 mM ADP, 10 mM GTP, 2.5 mM CDP, 2.5 mM UDP, 50 mM malate, 20 mM isocitrate), and 70 µl of transport buffer (0.6 M mannitol, 20 mM Hepes, pH 7.4, 1 mM MgCl₂, 2.5 mg/ml bovine serum albumin, with or without added inhibitors). Following incubation at 28 °C for 60 min, the reaction mixtures were cooled on ice for 5 min and divided into three equal portions. One portion was mixed with 10 volumes of sucrose/mannitol buffer containing protease inhibitor mix from Roche Molecular Biochemicals, Mannheim, Germany. Unless otherwise stated, the other two portions were incubated with 100 and 200 µg of Pronase/mg of mitochondrial protein, respectively, for 30 min on ice. The protease- treated samples were diluted 10-fold with sucrose/mannitol buffer containing protease inhibitor mix as described above. Mitochondria were reisolated from both the protease-treated and untreated samples by sedimentation through 1.35 M sucrose and were washed twice with sucrose/mannitol buffer containing protease inhibitors (32). The mitochondrial proteins were dissociated in Laemmli's sample buffer at 95 °C for 5 min and were analyzed by SDS-PAGE (42). The gels were either subjected to fluorography or imaged through a Bio-Rad GS-525 Molecular Imager.

Transient Transfection of cDNAs in COS Cells-- COS M6 cells were used for transient transfection with various cDNA constructs as described before (32, 43). Cells were cultured in Dulbecco's modified Eagle's medium containing 10% (v/v) fetal bovine serum and gentamycin (50 µg/ml) and transfected using Superfectamine, a lipophilic transfection reagent (Quiagen Inc., Germany) using the manufacturer's recommended protocol. Cesium chloride-banded plasmid DNAs (5-10 µg/plate) were used for transfection. 60-65 h after transfection, cells from 10 plates (100 mm) were pooled, homogenized in a Teflon-fitted glass homogenizer (10 strokes at 5,000 revolutions), and used for the isolation of mitochondrial and microsomal organelles by differential centrifugation methods (6). The mitochondrial fraction was resuspended in sucrose/mannitol buffer by gentle homogenization and further purified by banding through a discontinuous sucrose gradient (32). The mitochondrial fraction banding at the interface of 1.35 and 1.6 M sucrose was recovered, washed twice with the sucrose/mannitol buffer, and used for the Western blot analysis. The microsomes were recovered from the 10,000 × g supernatant by centrifugation at 120,000 × g for 1 h in a Sorvall RC 120-EX micro-ultracentrifuge. In all protein expression experiments, the cells were also co-transfected with 2 µg each of rat P450 reductase expression cDNA and bovine adrenodoxin cDNA (see Addya et al. (32)). The efficiency of cDNA transfection was monitored by assaying for NADPH P450 reductase activity, and plates showing 75-100% activities were chosen for the isolation of subcellular fractions. The extent of cross-contamination of mitochondrial and microsomal membrane fractions was assessed by Western blot analysis of fractions using antibodies specific for mitochondria (Adx antibody) and microsome (P450 reductase antibody). The whole cell extracts were prepared by ultrasonication method (32), and the protein concentration was determined by a dye-binding method (44).

Sterol 27-Hydroxylase Assay-- The 27-OH activity in isolated mitochondria and microsomes was determined by a modification of the method of Petrak and Latario (45) as described by Stravitz et al. (46). Membrane fractions (3 mg) or purified enzyme (100-150 pmol) in 0.5-ml reaction volumes of 0.1 M potassium phosphate buffer, pH 7.5, were incubated at 37 °C for 15 min, in the presence of energy-regeneration system (0.2 units of isocitrate dehydrogenase, 5 mM sodium isocitrate, 1.2 mM NADPH). 78 µM cholesterol in 1% 2-hydroxypropyl -cyclodextrin was added as the substrate (46). Enzyme reconstitution was carried out either in the presence of 0.2 nmol of Adx plus 0.02 nmol of Adr or 0.1 nmol of P450 reductase. Unless otherwise stated, microsomal P450 was reconstituted in lipid vesicles (dilauryl-phosphatidylcholine) as described (17, 45). Steroid products were oxidized with cholesterol oxidase (20 min at 37 °C), extracted twice with 3 ml of N-hexane, evaporated under nitrogen gas, and resolved by reverse-phase high pressure liquid chromatography (Beckman Instruments, Fullerton, CA; Ultrasphere silica C-18 column, 4.6 × 25 cm), at a flow rate of 0.8 ml/min in 30% methanol and 70% acetonitrile. Testosterone propionate was used as an internal recovery standard. The products were detected and quantitated by measuring the absorbance at 240 nm. All samples were run in duplicate, using appropriate enzyme blanks.

Western Blot Analysis of Proteins-- Mitochondrial and microsomal proteins and whole cell extracts were subjected to SDS-PAGE (42) on 10% polyacrylamide gels, and the separated proteins were electroblotted onto nitrocellulose membranes (48). The membranes were immunostained with monoclonal antibody to rat liver mitochondrial P450c27 (35, 49). Polyclonal antibody to porin was provided by Dr. D. Pain. Immunoblots were developed using alkaline phosphatase-conjugated anti-mouse IgG as the secondary antibody by the nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate color development method, using a reagent kit from Bio-Rad.

Confocal Immunofluorescence Microscopy-- COS cells were grown on coverslips to 80% confluency and were transfected with three different cDNA constructs. Fixing of cells and immunostaining were carried out as described (50) with some modifications (32). Paraformaldehyde-fixed (2% paraformaldehyde for 30 min) cells were permeabilized by treatment with 0.1% Triton X-100 and blocked with 5% goat serum for 1 h at 37 °C. Cells were double immunostained with 1:10 dilution of mouse antibody to P450c27 and 1:50 dilution of mouse monoclonal antibody for the mitochondrial genome-encoded COX subunit I (Molecular Probes, Inc., Eugene, OR) as a positive control. Cells were washed repeatedly with phosphate-buffered saline and were incubated with FITC-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) for the detection of P450c27 and Alexa 594-conjugated anti-mouse IgG (Molecule Probes, Inc. Eugene, OR) for the detection of COX subunit I protein. Incubation with secondary antibodies (1:100 dilution) was carried out for 1 h at 37 °C. Unbound secondary antibodies were removed by repeated washing with phosphate-buffered saline. Fluorescence microscopy was carried out under a TCS laser scanning microscope (Leica Inc., Deerfield, IL). 0.5-µm optical sections were scanned at the z axis with both FITC and Texas red channels fully open to prevent any shifting or distortion of the images.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

In Vitro Transport of P4501A1-DHFR Fusion Proteins into Mitochondria-- Mitochondrial targeting property of the N-terminal signal of P4501A1 was further investigated using chimeric fusion of 1A1 and DHFR protein sequences as shown in Fig. 1A. Specifically, we used the wild type 1-44, 33-44, and also R34D- and K39I-substituted 1-44 sequence of 1A1 as possible signals for targeting DHFR into isolated mitochondria. Fig. 1A also shows the amino acid sequence of the putative chimeric signal (N-terminal 1-44 sequence) of P4501A1. Fig. 1, B and C, shows the results of in vitro mitochondrial transport with various 1A1 and DHFR fusion proteins. In support of our previous results (32, 33), the gel pattern in Fig. 1B (left panel) shows that intact 1-44 sequence is a less efficient mitochondrial targeting signal for the transport of native 1A1 protein as compared with the N-terminal truncated 33-44 sequence. Results in Fig. 1B (right panel) also show that unmodified DHFR is not imported into mitochondria significantly, whereas the mouse COX Vb protein (43) used as an internal control is imported efficiently into a protease-resistant compartment.

View larger version (33K): [in this window] [in a new window]   Fig. 1.   In vitro transport of 1A1-DHFR fusion proteins into isolated rat liver mitochondria. Various 1A1 and DHFR fusion proteins as indicated in A were generated by transcription-linked translation of cDNA constructs in TNT reticulocyte lysate system in the presence of [35S]Met-labeled translation products and were used for in vitro transport in isolated mitochondria as described under "Experimental Procedures." The shaded area in A reflects the position of the 1A1 sequence and open area reflects the position of DHFR sequence of the fusion proteins. The Mut1A1 construct carries R34D and K39I substitutions in the 1A1 signal sequence. B and C, 200 µg of mitochondrial proteins each from in vitro transport reactions were resolved by SDS-PAGE, before () or after (+) treatment with 100 µg/mg Pronase. ++ indicates mitochondria treated with 200 µg/mg Pronase. Radiolabeled proteins were detected by fluorography. Translation products used for in vitro transport are indicated at the bottom of lanes in B, and fusion proteins used for the transport assays in C are shown at the bottom of the gel pattern. C, lanes marked TP (translation product) 4000-8000 cpm of respective translation products, and in lanes marked IP (input) reaction mixtures without Pronase treatment (about 40,000 cpm) were loaded.

Results in Fig. 1C show that the N-terminally fused 1-44 sequence of P4501A1 can function as a signal for targeting an otherwise cytosolic protein, DHFR, into mitochondria. Additionally, as shown for P4501A1 and +5/1A1 proteins (32), the putative signal sequence functions as an uncleaved signal since no significant processing of the fusion protein into a product, co-migrating with DHFR, is observed. As expected from previous studies using DHFR fusion constructs, methotrexate, a competitive inhibitor of DHFR, also inhibited mitochondrial import confirming that fully folded proteins with bound cofactors or ligands are import-incompetent (23, 31, 47, 51). The labeled protein protected against Pronase digestion indeed appears to reside inside the mitochondrial membrane, since treatment of in vitro incubated mitochondria with 0.1% Triton X-100 makes it sensitive to proteolytic attack. Results also show that 1-44/Mut1A1-DHFR protein is not imported significantly. Although not shown, 1-44/Mut1A1-DHFR protein resembled the wild type 1-44/1A1-DHFR protein in size, and also the input protein amount was similar to the wild type protein. Finally, as with the import of 1A1 protein, the 33-44 sequence was more efficient as mitochondrial targeting signal as compared with 1-44 sequence. The import of 33-44/1A1-DHFR fusion protein was also inhibited by methotrexate, and also imported protein became sensitive to proteolysis following addition of 0.1% Triton X-100. These results demonstrate that 1-44 and 33-44 sequences can function as mitochondrial transport signals, although the latter motif is relatively more active.

In Vitro Transport of P4501A1-P450c27 Fusion Proteins into Mitochondria-- Fig. 2A (a) shows the map of the rat P450c27 reading frame (35, 52), including the N-terminal cleaved mitochondrial targeting signal (marked as Sig Seq) and mature portion of the protein. The first 20 amino acid residues constitute the mitochondrial targeting signal, which includes the three positively charged residues (Arg) at 6th, 8th, and 10th positions that are critical for mitochondrial targeting. The mc27 (+21P450c27) construct in Fig. 2A (b), lacking the mitochondrial targeting signal, was derived by polymerase chain reaction amplification, and c-f (Fig. 2A) represent the various fusion constructs of 1A1 and mc27 (c27 portion shaded) as indicated. The latter series also consisted of mutant constructs (1-44/1A1Mut-mc27 and 33-44/1A1Mut-mc27) with K34D and R39I substitutions targeted to the putative signal sequence region of 1A1. The in vitro transport of ³⁵S-labeled proteins (Fig. 2B) encoded by the various cDNA constructs into isolated mitochondria was tested by protection against increasing concentrations of Pronase (100 and 200 µg/mg protein). The quantitation of the gel in Fig. 2C shows the extent of protection relative to the input radioactive protein at each of the Pronase levels used for digestion. It is seen from Fig. 2, B and C, that the wild type P450c27 with intact targeting signal was imported efficiently (30-40% of input) into mitochondria. The mc27 protein lacking mitochondrial targeting sequence, on the other hand, showed vastly reduced import (5-10% of input). It is also seen that significantly high levels (20-25% of input) of the 1-44/1A1-mc27 fusion protein was imported. The overall level of import of the fusion protein was about 50-60% of the rate of import of intact P450c27 protein. Interestingly, the fusion protein containing more truncated 1A1 signal (33-44/1A1-mc27) was imported at a higher efficiency (60-70% of input protein) than that of P450c27 protein with a cleavable N-terminal signal sequence. Furthermore, both 1-44/mut1A1-mc27 and 33-44/Mut1A1-mc27 proteins carrying R34D and K39I mutations were imported at an extremely low level (<5% of input), perhaps at the background level, further confirming the role of positive residues in mitochondrial targeting. In the case of P450c27, 1-44/1A1-mc27, and 33-44/Mut1A1-mc27 proteins, we also carried out control experiments to ensure the energy requirements for the transport and also intramitochondrial location of the protein that was resistant to proteolysis. It is seen that both carbonyl cyanide chlorophenylhydrazine, a mitochondrial specific ionophore which disrupts mitochondrial membrane potential, and oligomycin, which depletes the mitochondrial ATP pool, inhibit the transport. Additionally in all three cases, addition of 0.1% Triton X-100 to in vitro incubated mitochondria makes the labeled protein sensitive to Pronase digestion confirming that it is compartmentalized within the membrane compartment.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   In vitro transport of 1A1-P450c27 fusion proteins into isolated mitochondria. Full-length P450c27, +21P450c27, without N-terminal signal sequence (mc27) and various 1A1-mc27 fusion proteins shown in A were generated by transcription-coupled translation of cDNA constructs in the presence of 35S-labeled Met and used for in vitro transport into isolated mitochondria (B) as described under "Experimental Procedures." In panel A, open areas represent 1A1 signal components. Mitochondria were re-isolated from reaction mixture, and 100 µg of protein in each case was subjected to SDS-PAGE without () or with treatment with 100 µg/mg (+) and 200 µg/mg (++) Pronase, as indicated. The gels were imaged by scanning through a Bio-Rad GS-525 Molecular Imager, and values for no protease (IP, input) and protease-treated samples were plotted in C. Lanes marked TP (translation products) were loaded with 20,000-25,000 cpm protein, and lanes marked IP contained reaction mixture without Pronase treatment (40,000-50,000 cpm). Treatment with carbonyl cyanide chlorophenylhydrazine and oligomycin were at 50 µM level for 10 min at room temperature before adding the translation products. Triton X-100 (0.1% v/v) was added following import assay but before the addition of Pronase.

In Vivo Targeting of Fusion Proteins to Mitochondria and ER in COS Cells-- The extent of in vivo targeting of various fusion proteins to mitochondria and microsomes was tested by transient transfection of cDNA constructs in COS cells. Whole cell homogenates or isolated mitochondria and microsomes from transfected cells were analyzed by Western immunoblot using monoclonal antibody against rat P450c27. In Fig. 3, the in vivo translational efficiency of each cDNA construct was tested by Western blot analysis of whole cell homogenates. As seen form Fig. 3A, expression of full-length P450c27 protein yielded two closely migrating components consistent with the precursor and processed P450c27. Expression of 1-44/1A1-mc27 and +33-44/1A1-mc27 each yielded single antibody-reactive protein bands. In view of our published results on the N-terminal processing of full-length 1A1 protein in COS cells, and C6 glioma cells (32, 34), detection of a single band with 1-44/1A1 fusion protein was surprising. Although not shown, partially purified rat liver cytosolic protease was unable to process the 1-44/1A1-mc27 fusion protein suggesting that internal sites of 1A1 may be needed for the endoprotease action. Results in Fig. 3B show that N-terminal truncated mc27 protein was expressed efficiently, whereas the 1-44/Mut1A1-mc27 and 33-44/Mut1A1-mc27 proteins were expressed at lower efficiencies.

View larger version (44K): [in this window] [in a new window]   Fig. 3.   Expression of 1A1-P450c27 fusion proteins by transient transfection in COS cells. COS cells were transfected with 10 µg/plate of various wild type (A) or mutant cDNA constructs (B), and 150 µg of protein of total cell extract from each transfection were subjected to SDS-PAGE and Western blot analysis using monoclonal antibody to P450c27 as described under "Experimental Procedures." In both A and B, 1.5 µg of purified P450c27 was loaded as a standard.

Fig. 4A shows the relative distribution of various fusion proteins in the mitochondrial and microsomal fractions under transient transfection conditions. Quantitation of immunoreactive protein in different cellular fractions is presented below the gel pattern. As expected, intact P450c27 protein is exclusively targeted to the mitochondrial compartment with a very low level of antibody-reactive component in the microsomal fraction. Results also show that the 1-44/1A1-mc27 fusion protein is targeted to both the mitochondria and microsomes, and the antibody-reactive protein contents in the two organelles were nearly similar. Furthermore, we were unable to see any significant size difference between the mitochondrial and ER-associated protein suggesting no significant protein processing. This is in contrast to the size difference of the mitochondrial and microsomal P450s when intact P4501A1 protein is expressed by transient transfection (32, 34). Results in Fig. 4B show that 1-44/Mut1A1-mc27 protein is efficiently targeted to the microsomal membrane, whereas no significant mitochondrial targeting is observed. Furthermore, the 33-44/Mut1A1-mc27 protein is not targeted to either of the two membrane compartments, further confirming the critical role of the positively charged residues at 34 and 39 positions for mitochondrial targeting. A control experiment with the mc27 (+21P450c27) construct, lacking the mitochondrial targeting signal, shows no significant mitochondrial or microsomal targeting (Fig. 4C), although a significant accumulation of the protein in the cytosolic fraction is observed.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   In vivo targeting of 1A1-P450c27 fusion proteins to mitochondria and ER. COS cells were transfected with 10 µg/100-mm plate of type 1A1-mc27 (A), 1A1 Mut-mC27 cDNA constructs (B), or mature P450c27 cDNA constructs (C) and co-transfected with 2 µg/plate each of the rat P450 reductase cDNA and bovine Adx cDNA constructs. Mitochondria and microsomes were isolated from transfected cells as described under "Experimental Procedures," and 100 µg of protein from each subcellular fraction was subjected to SDS-PAGE and Western blot analysis using monoclonal antibody against P450c27. The gels were scanned through a Bio-Rad FluorS Imager, and the relative intensities of antibody reactive bands in A are presented. C, 100 µg of cytosolic protein (100,000 × g supernatant) was loaded. D, mitochondrial and microsomal fractions from wild type 1-44/1A1-mc27 cDNA transfected cells from A were subjected to Western blot analysis using antibody against Adx (left panel) and antibody against P450 reductase (right panel).

In all the experiments on the subcellular distribution of fusion proteins (Figs. 4,A-C), cells were co-transfected with cDNAs for P450 reductase, which is known to be exclusively targeted to the ER, and Adx, which is exclusively targeted to mitochondria. As a routine practice, mitochondrial and microsomal isolates from transfected cells were probed with P450 reductase and Adx antibodies to assess the level of cross-contamination. Representative immunoblots with mitochondrial and microsoma1 proteins from 1-44/1A1-mc27 cDNA-transfected cells (from Fig. 4A) are shown in Fig. 4D. It is seen that the 12-kDa Adx antibody-reactive protein is detected only in the mitochondrial fraction with no detectable band in the microsomal fraction (left panel). Conversely, the P450 reductase antibody cross-reactive 78-kDa protein is detected only in the microsomal fraction with no detectable antibody-reactive protein in the mitochondrial fraction. Mitochondrial and microsomal isolates from other transfected cells from this series of experiments (Fig. 4, A-C) showed a similar degree of purity (results not shown). Additionally, although results are not presented, mitochondrial isolates contained less than 1% of microsome-specific marker enzyme NADPH cytochrome c reductase (rotenone sensitive), and >90% mitochondrial specific marker enzymes cytochrome c oxidase and isocitrate dehydrogenase.

Immunohistochemical Analysis of P450c27 Fusion Proteins Targeted to ER and Mitochondria-- The dual targeting property of the N-terminal signal sequence of P4501A1 was further verified by immunofluorescence microscopy of COS cells transfected with different 1A1-mc27 fusion constructs. Immunohistograms in Fig. 5, ac, represent cells transfected with intact P450c27 cDNA construct. As expected, P450c27 antibody stained mostly particulate and punctate structures resembling mitochondrial membranes. As seen from Fig. 5b, and the overlay in c, the staining pattern with P450c27 antibody is nearly identical to that with antibody to mitochondrial gene-coded COX I protein. In Fig. 5, d and e, cells transfected with 1-44/1A1-mc27 construct were stained with antibody to P450c27 and mitochondrial encoded COX I protein, respectively. The P450c27 antibody staining (Fig. 5d) is more diffuse and includes some punctate particulate structures in addition to lighter membrane structures, possibly representing the ER. Additionally, consistent with recent reports on the detection of P4502E1 and 2B1 antibody-reactive proteins on the hepatocyte plasma membrane (53, 54), the histogram in Fig. 5d also shows low level of targeting of the fusion protein to the plasma membrane fraction of transfected cells. The staining pattern with COX I antibody in Fig. 5e and the overlay in f indicate that some of the particulate structures stained with P450c27 antibody may represent mitochondrial membrane particles. Expression of more truncated 33-44/1A1-mc27 fusion protein, on the other hand, yielded an antibody staining pattern nearly similar to that of intact P450c27 expression, in that the antibody-reactive protein is predominantly associated with punctate structures. Furthermore, the P450c27 antibody-stained structures in Fig. 5g are nearly quantitatively superimposed with the COX I antibody-stained structures (see h and i), suggesting mostly mitochondrial localization. Although not shown, staining patterns in Fig. 5, a and d, did not resemble the staining patterns with antibody against Golgi-specific marker protein -COP. These results together suggest that 1-44/1A1 signal functions as a chimeric signal capable of targeting proteins to both ER and mitochondria, and the sequence 33-44 of P4501A1 functions predominantly as a mitochondrial targeting signal.

View larger version (79K): [in this window] [in a new window]   Fig. 5.   Subcellular localization of fusion proteins expressed in COS cells by immunohistochemical analysis. COS cells were transfected with full-length P450c27 cDNA (a), 1-44/1A1-mc27 cDNA (d), 33-44/1A1-mc27 cDNA (g), and serially immunostained as follows: 1) stained with monoclonal antibody to P450c27 and FITC-conjugated anti-mouse IgG (a, d, and g) and 2) stained with monoclonal antibody against mitochondrial genome-encoded COX I protein and Alexa 594-conjugated anti-mouse IgG (b, e, and h). In c, f, and i, the anti-P450c27 and anti-COX I antibody staining patterns were superimposed. Details of staining, washing, and confocal microscopy were as described under "Experimental Procedures."

Distinct Membrane Topologies of 1-44/1A1-mc27 Protein Targeted to Mitochondria and ER-- Previous results showed that both nearly intact P4501A1 (+5/1A1) and more truncated +33/1A1 targeted to liver, brain, and lung mitochondria occur as membrane extrinsic proteins that are extractable in alkaline Na₂CO₃ buffer (17, 18, 34). In the present study, since the 1-44/1A1 signal appears to function as an uncleaved signal for targeting P450c27 protein to mitochondria, we decided to determine its membrane topology in the mitochondrial and microsomal membranes. COS cells transfected with the 1-44/1A1-mc27 cDNA construct were fractionated, and the mitochondrial and microsomal fractions were extracted with alkaline Na₂CO₃ buffer as described before (17, 55). P450c27 targeted to COS cell mitochondria by cDNA transfection and endogenously expressed porin were used as internal controls for mitochondrial membrane extrinsic (matrix) and transmembrane (outer mitochondria) proteins, respectively. As shown in Fig. 6A (left panel), over 90% of the antibody-reactive 1-44/1A1-mc27 protein targeted to mitochondria partitioned in the Na₂CO₃ soluble fraction (S), whereas more than 80% of the protein targeted to the ER was insoluble (P) under identical extraction conditions. As expected, full-length P450c27 expressed under similar conditions partitioned exclusively in the alkaline-soluble fraction (Fig. 6A, left panel). As shown in the right panel (Fig. 6A), about 85% of the endogenously expressed porin is found in the alkaline insoluble (P) fraction indicative of its transmembrane orientation.

View larger version (36K): [in this window] [in a new window]   Fig. 6.   Distinct membrane topology of 1-44/1A1-mc27 fusion protein targeted to the ER and mitochondria. A, mitochondria and microsomes from 1-44/1A1-mc27 cDNA or full-length P450c27 cDNA transfected COS cells were extracted with 0.1 M Na2CO3, pH 11.0. The total soluble protein recovered by trichloroacetic acid precipitation (S) and proteins from the insoluble membrane fraction (P) were resolved by SDS-PAGE and subjected to Western blot analysis using monoclonal antibody to P450c27 (left panel). The right panel represents Western blot of alkaline-soluble (S) and -insoluble (P) protein fractions probed with antibody to porin. B, 35S-labeled translation products as indicated at the top of lanes were used for in vitro transport assays, and resultant mitochondria were subjected to extraction with 0.1 M Na2CO3 extraction following a milder protease digestion (100 µg/mg Pronase for 10 min on ice). Alkaline-soluble (S) and -insoluble (P) proteins were resolved by SDS-PAGE, and the gel was imaged through the Bio-Rad GS-525 molecular imager. Details of alkaline extraction and recovery of protein fractions were as described before (17).

The membrane extrinsic organization of newly imported 1-44/1A1-mc27 fusion protein in mitochondria was further ascertained using the in vitro transport system. In vitro imported porin was used as an internal control. Since transmembrane-associated porin is relatively more sensitive to protease digestion, in this series of experiments (Fig. 6, A and B), mitochondria sedimented through a discontinuous sucrose gradient were subjected to milder protease digestion (100 µg/mg Pronase for 10 min on ice), which did not affect the integrity of the membrane-associated porin. In keeping with the in vivo expression data in Fig. 6A, over 90% of the in vitro imported P450c27 as well as 1-44/1A1-mc27 fusion proteins were extractable with alkaline Na₂CO₃. The case with the N-terminal truncated 33-44/1A1-mc27 protein was similar. However, more that 95% of the in vitro imported porin was resistant to extraction with Na₂CO₃ suggesting its transmembrane orientation. These results therefore provide evidence that the proteins targeted to two different membrane compartments using the 1-44 signal motif of 1A1 assume different membrane topologies.

Catalytic Properties of Mitochondrial and ER-targeted Fusion Proteins-- The 27-OH activity of the 1A1-mc27 fusion proteins with the P450 reductase and mitochondrial Adx + Adr electron transfer proteins was assayed to gain insight on the functional properties of the fusion proteins, which assume different membrane orientations in the two membrane compartments. It is seen (Table I) that 27-OH activity of the purified rat liver mitochondrial P450c27 is fully supported by Adx + Adr electron transfer proteins to yield a specific activity of 2.9 nmol (Table I). The microsomal P450 reductase was, however, unable to support the 27-OH activity of the purified enzyme under the assay conditions (Table I). Results also show that uninduced rat liver microsome did not show any detectable 27-OH activity. Although not shown, reconstitution in dilaurylphosphatidylcholine vesicles and addition of P450 reductase did not yield any activity with the microsomal preparation. As expected, isolated rat liver mitochondria showed about 5.8 nmol activity even in the absence of added electron transfer proteins. In support of previous observations, cholesterol appears to yield significantly lower 27-OH activity as compared with vitamin D₃ for 25-OH activity and 5-cholestene derivatives for 27-OH activity (4, 52, 56).

The P450 contents and 27-OH activities of mitochondrial and microsomal isolates from COS cells transfected with intact P450c27 or 1A1-c27 fusion proteins are presented in Table II. Because of the generally low endogenous microsomal P450 reductase and mitochondrial Adx in these cells, membrane fractions from transfected cells were reconstituted with exogenous electron transport proteins. Additionally, reconstitution of microsomal membranes was carried out with added liposomes to obtain optimal activity. Results show that both mitochondria and microsomes from mock-transfected cells contained no detectable P450, whereas both membrane fractions from 1-44/1A1-mc27-expressing cells contained 80-120 pmol/mg P450. Microsomes from cells expressing truncated 33-44/1A1-mc27 fusion protein contained very low (<8 pmol/mg) P450, whereas mitochondria from these cells contained 65 pmol/mg P450. The observed P450 distribution in transfected COS cells is consistent with the antibody-reactive protein distribution (Fig. 4) in similarly transfected cells. Consistent with its known targeting properties of P450c27, microsomes from COS cells expressing the full-length protein showed no detectable activity, whereas mitochondria from these cells, when reconstituted with Adx + Adr, yielded 2.5 nmol activity. Both mitochondria and microsomes from cells expressing truncated P450c27 (+21P450c27) did not show any activity. Interestingly, both the microsomes and mitochondria from cells transfected with mc27 fused to the intact 1A1 chimeric signal (1-44/1A1-mc27) yielded activity in the range of 1.4 to 1.9 nmol, whereas only the mitochondrial fraction of cells transfected with fusion protein containing the N-terminal truncated signal sequence (33-44/1A1-mc27) yielded 2.2 nmol activity. For unknown reasons, mature P450c27 or its fusions expressed in COS cell mitochondria yielded only about half of the activity obtained with intact liver mitochondria suggesting that a yet uncharacterized liver mitochondrial component may be needed for the optimal activity of the P450. The results on the P450 contents and enzyme activity patterns provide direct support for the chimeric nature of the 1-44/1A1 signal. In support of a recent study (70), these results also show that mitochondrial P450c27 in its transmembrane orientation might functionally interact with microsomal P450 reductase.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

A majority of the mitochondrial matrix targeted proteins contain N-terminal helical amphipathic sequences that are cleaved off immediately after import by a mitochondrial metalloendoprotease (57). However, mitochondrial targeting signals ranging from non-helical to random secondary structure or mostly polar sequences have been reported (20, 58-60). A number of proteins also contain either N-terminal, C-terminal, or internal uncleaved signals (51, 61-63) for targeting proteins to mitochondrial inner or outer membrane, as well as the matrix compartments. Results of this study therefore show that the N-terminal 1-44 sequence of P4501A1 possesses dual membrane targeting property. In support of previous results with intact P4501A1 apoprotein (32), the N-terminal signal motif can target heterologous proteins, DHFR, and mature portion of P450c27 to both the ER and mitochondria under both in vivo and in vitro conditions. The N-terminal signal of P4501A1 (1-44/1A1) is chimeric in that the N-terminal most 32 residues are highly essential for ER targeting, whereas the next 12 amino acid residues (residues 33-44), containing three positively charged residues, function as a cryptic mitochondrial targeting signal.

It was recently shown that BCS1 protein, the Reiske FeS protein of the bc₁ complex, which is organized in a transmembrane (N_out  C_in) topology on the inner mitochondrial membrane, contains an internal targeting and a membrane anchor signal (51). The putative mitochondrial targeting signal, rich in basic amino acid residues (residues 69-83), immediately follows the putative transmembrane domain at residues 45-68. In this respect, N-terminal signal sequence of P4501A1 resembles that of the internal signal of BCS1. In contrast to the integral membrane orientation of the BCS1 protein, however, both 1-44/1A1-mc27 and +33-44/1A1-mc27 fusion proteins targeted to mitochondria assume a membrane extrinsic topology (Fig. 6). Another notable difference is that the removal of the N-terminal transmembrane domain activates the putative mitochondrial targeting signal of the P4501A1, whereas a similar N-terminal truncation does not seem to affect the efficiency of the internal signal of BCS1 protein.

The occurrence of positively charged residues is a common feature of mitochondrial targeting signals. The charged residues are usually distributed uniformly through out the length of the presequence, which ranges in size from about 15 to 45 amino acid residues (31). It is thought that the positively charged N-terminal domain might be involved in binding to the negatively charged surface of the mitochondrial membrane or regions of the import receptor/translocator proteins such as Tom20 or Tom22 (64-66). It has also been suggested that mitochondrial processing protease contains a negatively charged surface, which is critical for the binding of precursor proteins and subsequent cleavage of the presequence (67). Although the 1-44 as well as 33-44 sequence as part of the homologous 1A1 apoprotein or chimeric fusion with reporter proteins are not cleaved by the mitochondrial endoprotease (Figs. 3 and 4), the positively charged residues at 34 and 39 positions of the signal are critical for the mitochondrial targeting function.

Results of this study also point out some similarity in terms of targeting property but differences with respect to cytosolic processing of the 1-44 sequence in its native setting as part of P4501A1 as compared with its activity when fused to the N terminus of reporter protein, DHFR. Both in the native setting and as fusion of DHFR reporter, 1-44 sequence is a relatively inefficient signal for mitochondrial protein import as compared with the truncated 33-44 sequence. Previous results with endogenous P4501A1 and also that expressed in COS and C6 glioma cells (32, 34) showed that the mitochondrial targeting is associated with the N-terminal cleavage of the protein at the +4 and +32 positions. Results on the COS cell expression of 1-44/1A1-mc27 fusion protein, on the other hand, show no detectable shortening of the mitochondrial targeted protein (Figs. 3 and 4). Inability of the cytosolic protease to process the fusion protein in vitro (results not shown) suggests that internal sequence of P4501A1 may be necessary for binding to the endoprotease. Alternatively, we cannot rule out the possibility that the fusion protein is processed at the 4th position but not at the internal site. We were unable to test the latter possibility due to technical problems of isolating sufficient amount of mitochondrial targeted fusion protein for N-terminal sequencing under transient transfection conditions.

It is well established that the N-terminal 10-15 amino acid residues of various microsomal P450s are not only required for ER targeting, but they may also serve as stop transfer signals (8, 9, 12). This view is supported by mutations targeted to the two N-terminal most positively charged residues (2nd and 3rd residues) of P4502C2 (14), which resulted in the nearly complete translocation of the protein molecule into the ER lumen. As shown in Fig. 6, 1-44/1A1-mc27 targeted to the COS cell ER appears to assume a transmembrane topology. Furthermore, although results are not presented, the ER-associated fusion protein is highly sensitive to added protease suggesting the cytosolic orientation of bulk of the molecule. In view of this, membrane extrinsic organization of the +5/1A1 protein as well as 1-44/1A1-c27 fusion protein targeted to mitochondria is highly surprising, particularly since they contain nearly intact stop-transfer and transmembrane domains. It is likely that the mitochondrial protein transport machinery is unable to recognize the stop transfer signal of 1A1, which is functional in the ER membrane system. These results suggest inherent differences between the two membrane transport systems.

Results of this study also shed light as to how the same protein can assume different topological orientation in two different cytoplasmic membrane compartments, which could also be the basis for differences in their catalytic properties and also mode of interaction with different electron transfer proteins. A distinctive feature of the microsomal and mitochondrial P450s is thought to be their requirements for specific electron transfer proteins, the microsomal P450 reductase, and the soluble ferridoxin and ferridoxin reductase systems, respectively. In exception to this generality, microsomal P450c17, was shown to be fully active with bacterial flavodoxin and flavodoxin reductase (68). Furthermore, bacterially expressed N-terminal truncated P4501A2 showed activity in both bacterial flavodoxin + flavodoxin reductase, as well as ferridoxin + ferridoxin reductase (69). More recently, P450MT2 (N-terminal truncated 1A1) purified from BNF-induced liver mitochondria and also bacterially expressed +33/1A1 were shown to undergo functionally productive interaction with Adx and Adr proteins by multiple criteria. However, studies showing the productive interaction of bona fide mitochondrial P450s with P450 reductase have been relatively rare. A recent study showed that a chimeric fusion protein consisting of N-terminal signal sequence of microsomal P450c17 and mature protein of P450c27, when expressed in yeast cells, is correctly targeted to the ER membrane (70). The extent of mitochondrial targeting of this fusion protein was not investigated. Surprisingly, the microsomal associated P450c27 showed catalytic activity in a P450 reductase system. However, results in our own and others' laboratories (32, 46, 56) showed that bacterially expressed mature form of P450c27 or that purified from mitochondria exhibit negligible activity in a P450 reductase supported system. These contrasting results raised the possibility that a productive interaction with the microsomal P450 reductase is facilitated only when mitochondrial P450c27 assumes a transmembrane orientation. Our results with the 1-44/1A1-mc27 fusion protein expression system show that in a membrane-anchored orientation, P450c27 is able to interact with similarly organized P450 reductase, whereas in its membrane extrinsic orientation in the mitochondrial compartment, it can effectively interact with the soluble electron transfer proteins, Adx + Adr. This possibility is further supported by previous studies showing that N-terminal truncation of P450 reductase significantly reduced its ability to support the activity of microsomal P450s (71, 72).

In summary our results provide direct and conclusive evidence on the dual targeting properties of the N-terminal signal sequence of P4501A1. Additionally, in conjunction with the results of Sakaki et al. (70) results presented in this study provide new insights on the mode of interaction of mitochondrial P450s with P450 reductase, organized in a transmembrane orientation. These results have implications on the evolutionary relationship between the mitochondrial and microsomal P450 system and their specific electron transport proteins.

	ACKNOWLEDGEMENTS

We thank Drs. Debkumar Pain, Michael Waterman, and Frank Gonzales for generously providing some of the cDNA constructs and antibodies used in this study. We are also grateful to Dalila Markues for help with the 27-OH assays and all members of the Avadhani laboratory for helpful and valuable suggestions during this study and also their criticisms on the manuscript.

	FOOTNOTES

^* This research was supported in part by the National Institutes of Health Grant GM-34883.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Present address: Dept. of Experimental Oncology, St. Jude Children's Research Hospital, 332, N. Lauderdale, Memphis, TN 38105.

To whom correspondence should be addressed. Tel.: 215-898-8819; Fax: 215-573-6651; E-mail: narayan@vet.upenn.edu.

	ABBREVIATIONS

The abbreviations used are: P450, cytochrome P450; 1A1, CYP1A1; c27, CYP27; P450 reductase, NADPH cytochrome P450 reductase; Adx, adrenodoxin; Adr, adrenodoxin reductase; BNF, -naphthoflavone; DHFR, dihydrofolate reductase; ER, endoplasmic reticulum; 27-OH, cholesterol 27-hydroxylase; PAGE, polyacrylamide gel electrophoresis; FITC, fluorescein isothiocyanate.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES