|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: LS TG BW WJ. Performed the experiments: LS TG CH DL SW KG. Analyzed the data: LS TG SW KG WJ. Contributed reagents/materials/analysis tools: HEM RS JP. Wrote the paper: WJ.
The current view of peroxisome inheritance provides for the formation of new peroxisomes by both budding from the endoplasmic reticulum and autonomous division. Here we investigate peroxisome-cytoskeleton interactions and show by proteomics, biochemical and immunofluorescence analyses that actin, non-muscle myosin IIA (NMM IIA), RhoA, Rho kinase II (ROCKII) and Rab8 associate with peroxisomes. Our data provide evidence that (i) RhoA in its inactive state, maintained for example by C. botulinum toxin exoenzyme C3, dissociates from peroxisomes enabling microtubule-based peroxisomal movements and (ii) dominant-active RhoA targets to peroxisomes, uncouples the organelles from microtubules and favors Rho kinase recruitment to peroxisomes. We suggest that ROCKII activates NMM IIA mediating local peroxisomal constrictions. Although our understanding of peroxisome-cytoskeleton interactions is still incomplete, a picture is emerging demonstrating alternate RhoA-dependent association of peroxisomes to the microtubular and actin cytoskeleton. Whereas association of peroxisomes to microtubules clearly serves bidirectional, long-range saltatory movements, peroxisome-acto-myosin interactions may support biogenetic functions balancing peroxisome size, shape, number, and clustering.
Mammalian peroxisomes associate with the microtubular cytoskeleton for intracellular transport , . Three distinct states of motility were recognized, long-range saltations, oscillations and arrest. Furthermore, we demonstrated that peroxisome motility is subject to regulation by extracellular ATP-lysophosphatidic acid (LPA) receptor co-stimulation. Signaling involves trimeric Gi/Go protein, PLC, Ca2+ influx, cPKC, MAP kinase and PLA2 and mediates peroxisomal arrest –. Via G12/13 the LPA receptor activates the Rho pathway preventing long-range peroxisomal motility that seems to be regulated by a complex signaling network , .
These studies did not show the involvement of the actin cytoskeleton in the motility of mammalian peroxisomes. However, plant and yeast peroxisomes associate with the actin cytoskeleton for saltatory movement and inheritance, respectively. In plant cells interference with the actin cytoskeleton results in loss of saltatory movement, aggregation, and complete cessation of peroxisome motility . In S.cerevisiae peroxisomes are targeted and segregated to the developing bud by a highly ordered process involving actin and Myo2p, a class V myosin motor protein , . Actin-based movement of mammalian organelles other than peroxisomes has however been demonstrated , . Two types of motility are described. One is based on actin polymerization itself propelling organelles by an “actin comet tail” toward the cell center. This type has been described for phagosomes and macropinosomes and may also be involved in transport between endosomes and lysosomes , . The other type of movement depends on actin-based myosin motors and utilizes myosins of the classes I, II, V or VI. This myosin-dependent movement has been implicated in dynamics of the ER, lysosomes, Golgi-derived vesicles, secretory granules, recycling endosomes and melanosomes –.
Some indirect evidence suggests that mammalian peroxisomes also associate with the actin cytoskeleton. Drp-1 known to be involved in mitochondrial fission in an actin cytoskeleton-dependent manner  was shown also to localize to peroxisomes. Over-expression of a dominant-negative mutant of Drp-1 prevents peroxisome division in a human hepatoma cell line –. Generally, members of the dynamin protein family are known regulators of vesicle trafficking implicated in constricting and severing membrane tubules .
Ultrastructural studies on peroxisome proliferation in rat liver suggest a sequential mechanism of membrane tubulation and constriction prior to fission , . Thus, a force-generating system such as the acto-myosin complex is likely to be involved. Frequently, these processes are regulated by small GTPases known to support membrane traffic as well as the local organization of both the microtubular and the actin cytoskeleton , . Members of the Rho, Rab and Sar1/Arf families are implicated ,  and Rho1p and Arf1 recently shown to associate with peroxisomes , , , , , . Mammalian peroxisomes in presence of GTP-primed cytosol recruit Arf1 and the COPI coat subunits and among the proteins specifically enriched with S. cerevisiae peroxisomes Rho1p was identified by a mass spectrometric screen.
So far our knowledge on the mechanism of peroxisome proliferation and the role small GTPases and the cytoskeleton play is rather incomplete. Peroxins, such as Pex11p, Pex25p and Pex27p are likely to be involved –. However, factors are still missing in order to fully understand the process. In the present study we initiated proteomics analyses to identify components that recruit onto peroxisomes from the cytosol. We found a significant number of candidate proteins some of which were further investigated. Here we report alternate binding of peroxisomes to microtubules and actin microfilaments dependent on the activation state of RhoA. Our studies indicate a sequence of events in which activated RhoA by binding to peroxisomes detaches the organelles from microtubules. Concomitantly, activated RhoA may favor peroxisome association with the actin cytoskeleton possibly by recruiting Rho kinase (ROCKII). Phosphorylation of myosin light chain induces actin-activated non-muscle myosin IIA (NMM IIA) ATPase activity enabling force generation.
Intracellular movement of peroxisomes along microtubular tracks is regulated by ATP/LPA receptor co-stimulation and activation of heterotrimeric G proteins. Signaling by both receptor co-stimulation and activation of G proteins caused peroxisomal arrest –. The first clues for a possible involvement of RhoA in this regulation came from LPA signaling that is known to activate RhoA by the trimeric G protein subfamily member G12/13 , . Therefore, we first tested C. botulinum exoenzyme C3 that is known to inactivate Rho proteins of the RhoA subfamily by ADP ribosylation . A significant increase in overall peroxisomal movements compared to untreated cells was observed independent of whether the toxin was added to the culture medium or expressed by transfection (Fig. 1A). As shown by the time series, most peroxisomes moved directionally with frequent changes in direction. Those organelles that did not saltate were at rest for short periods. Following the addition of Nocodazole to exoenzyme C3-treated cells, peroxisomes oscillated, similarly to what could be observed after Nocodazole treatment only (not shown). Both resting and saltating states presupposed association of the organelles to the microtubular cytoskeleton. The motility state elicited by exoenzyme C3 was in sharp contrast to the control state, as the frequency of peroxisomes moving longer distances was significantly increased. On the other hand, transfecting cells with myc-G14V-RhoA, the constitutively active form of RhoA exhibiting remarkably reduced intrinsic GTPase activity and unresponsiveness to RhoA GTPase-activating protein , resulted in random oscillations of peroxisomes that were indistinguishable from those observed after Nocodazole treatment (Fig. 1B). This regulatory effect was specific for RhoA. Dominant–active RhoD did not influence peroxisomal motility (Fig. 1B).
By electron microscopy peroxisomal constrictions were frequently seen preserving membrane continuities between segregated compartments (Fig. 2A–C). Both microtubules (Fig. 2B and C) and microfilament bundles (Fig. 2D–F) were regularly found in close proximity to the organelles (Fig. 2B and C) and smooth ER cisternae possibly delivering Ca2+ positioned close to the sites of constrictions. Taken together, these results suggest that peroxisomes contact both microtubules and cytoskeletal filaments. Activated RhoA may uncouple peroxisomes from microtubules as typical microtubule-dependent long-range saltations are no longer observed.
Following incubation of unlabeled rat liver peroxisomes with 35S-labeled rat hepatocyte cytosol, proteins with Mr of about 220, 60 and 40 kDa strikingly co-isolated with peroxisomes (not shown). These proteins were identified by nano-HPLC/ESI-MS/MS as non-muscle myosin heavy chain (MyH9) belonging to the class IIA of conventional myosins, α- and β-tubulin and β-actin, respectively. Additional proteins specifically recruiting from cytosol onto peroxisome membranes were detected by extended proteomics analysis. Table 1 lists the proteins identified and related to organelle dynamics and cytoskeletal functions.
To confirm specific peroxisomal recruitment of these proteins, an in vitro peroxisome-binding assay was developed. In this assay peroxisomes were incubated with GMP-PNP-primed cytosol in presence or absence of an ATP-regenerating system and recovered after flotation in a discontinuous Nycodenz gradient. Pex11αp, a peroxisomal membrane marker, floated up in the gradient to fraction F2 just between F1 (250 mM sucrose) and F3 (45% w/v Nycodenz) the latter retaining all the non-floating material. Using antibodies against b-actin and NMM IIA heavy chain, both myosin and actin were detected in the re-isolated peroxisomes. Binding of NMM IIA to peroxisomes required ATP and was observed in the presence of ATP plus GMP-PNP. However, actin recruitment required both nucleotides; the sole presence of GMP-PNP did not mediate recruitment of NMM or actin (Fig. 3A). In control experiments ran in the absence of peroxisomes neither NMM nor actin did float up (Fig. 3B, D, and F). Cytosol preparations contain endogenous guanine nucleotides that in presence of an ATP-regenerating system might become phosphorylated to GTP by nucleotide kinases. Therefore, the nucleotide dependence of NMM IIA and actin recruitment was investigated in presence of GDPγS, a metabolically inert GDP analog. These studies clearly demonstrated that binding of NMM IIA and actin onto peroxisomes requires both ATP and GTP (Fig. 3C). Replacing the ATP-regenerating system for AMP-PNP, a non-hydrolyzable ATP analog, prevented NMM IIA association to peroxisomes suggesting that the association requires hydrolyzable ATP (Fig. 3E).
Peroxisomal NMM IIA association was also proved by immunofluorescence. NMM IIA predominantly appeared as filamentous structures bearing peroxisomes closely attached like pearls on a string (Fig. 4A–C and inset in Fig. 4C) rather than having them distributed at random. Analyzing NMM IIA and actin filament stainings revealed nearly complete overlap of both of these structures (Fig. 4D–F). Thus, these immunofluorescence studies strongly suggest peroxisome acto-myosin associations.
The GTP dependence of NMM IIA and actin recruitment onto peroxisomes indicate small GTP-binding protein(s) to be involved. Using antibodies specific to RhoA, the GTP-dependent recruitment of RhoA to peroxisomes was observed in vitro (Fig. 5). In line with these biochemical assays, we also demonstrated co-localization of peroxisomes and RhoA by transfecting myc-G14VRhoA, a dominant-active mutant of RhoA, into mouse AT3 hepatoma cells (Fig. 6). Whereas both wild type and dominant-active RhoA associated to peroxisomes (Fig. 6A–F), Rho A was not detected on the organelles upon treatment of cell cultures with exoenzyme C3 (Fig. 6G, H, I). By closer examination of the RhoA topology, it appeared that activated RhoA concentrated at distinct sites on the peroxisomal membrane rather than exhibiting a continuous staining pattern (Fig. 6F and inset). In contrast to the RhoA staining, catalase, a peroxisomal matrix protein, constantly labeled the entire organelle. This localization of RhoA to discrete domains of the peroxisomal membrane is reminiscent to the distribution of Pex11αp and PMP69 in CHO cells overexpressing N-terminally myc-tagged Pex11βp . As Pex11βp expression was shown to cause peroxisome proliferation , it is tempting to speculate that segregation and concentration of distinct proteins within the peroxisomal membrane and peroxisome proliferation are functionally linked to each other.
A known effector of RhoA signaling is Rho-associated coiled-coil containing kinase (ROCK). By phosphorylation of myosin light chain (MLC) Rho kinase induces formation of stress fibers and focal adhesions . Two Rho kinase isoforms have been identified, ROCKI and ROCKII. Whereas ROCKI was recently shown to localize to plasma membrane and centrosomes , , ROCKII appears to be mainly distributed in the cytoplasm and translocates to membranes upon activation by GTP-bound RhoA . Therefore, we focused our attention on ROCKII. Again we used the proxisomal binding assay and observed a significant 3-4-fold increase in the amount of ROCKII recruited from cytosol onto peroxisomes in the presence of ATP plus GMP-PNP (Fig. 7). Interestingly, isolated peroxisomes without being further incubated already had bound ROCKII in low concentration (Fig. 7A and C, lane 6).
We also analyzed peroxisomal attachment of ROCKII by immunofluorescence. Peroxisomes were labeled by GFP carrying a C-terminal peroxisomal targeting signal 1 (GFP-SKL) and endogenous ROCKII was stained using a rabbit polyclonal antiserum (Fig. 8). According to the known broad cellular distribution of ROCKII, cellular background staining was high. Nevertheless, co-localization of ROCKII with the peroxisomal signal was convincingly evident. Quantitative evaluation of 256 GFP-labeled peroxisomes revealed 66 peroxisomes to co-localize with the ROCKII signal, i.e. more than 25% co-localizing structures were recognized.
As membrane docking of myosin motors is known to be controlled by Rab proteins , , , we next investigated by an additional proteomics screen the recruitment of small GTPases to the peroxisomal membrane focusing on membrane-associated proteins with Mr between 15 and 30 kDa. We detected members of the Rab family and the Rho family as well as ARF, ARF-related and SAR1 proteins. To increase the stringency of our analysis, the Rab proteins were semi-quantitatively profiled based on spectral counts  over the entire Nycodenz density gradient separating the post-nuclear fraction of a rat liver homogenate. In this profiling, Rab8a and Rab8b exhibited a gradient distribution strikingly similar to the peroxisomal matrix and membrane markers, bifunctional protein and Pex11ap, respectively (Fig. 9). To further investigate the peroxisomal localization of Rab8, Rab8a was cloned as 5′-GFP fusion construct and dominant-active (Q67L) and dominant-inactive (T22N) forms were stably expressed in Huh7 human hepatocellular carcinoma cells. For co-localization studies, peroxisomes and Golgi were stained with anti-catalase and anti-GM130 antibodies, respectively. The results of these experiments are shown in Fig. 10. (i) Rab8a clearly co-localizes with peroxisomes (Fig. 10A–I) and Golgi cisternae (Fig. 10J and K). (ii) The total average number of peroxisomes per cell changed by expressing wild type GFP-Rab8aWT (54±7; n=115; Fig. 10G–I), dominant-active Rab8aQ67L (35±6; n=119; Fig. 10A–C) and dominant-inactive Rab8aT22N (68±15; n=79; Fig. 10D–F). (iii) Applying organelle-based co-localization (OBCOL) analysis , we noted an about tenfold increase in the number of peroxisomes co-localizing with Rab8a in cells transfected with dominant-inactive Rab8aT22N compared with dominant-active Rab8aQ67L (Fig. 10L).
Our current view of the multiple signaling pathways regulating peroxisome microtubule interaction encompasses both activation of PLA2 via Gi/Go liberating arachidonic acid by ATP and RhoA via G12/13 by extracellular LPA , , . Both signals synergize in this process and RhoA appears to be a specific and important intermediate. Most of the activated cytosolic RhoA translocates to cellular membranes leading to saturation of RhoA binding sites and terminating microtubule-peroxisome interactions. Peroxisome oscillations are generated indistinguishable to those following depolymerization of microtubules , . On the other hand, exoenzyme C3 rendering RhoA in its inactive GDP-bound form strikingly increased the number of long-range microtubule-based peroxisome saltations. These observations are consistent with the idea that cycling of RhoA between its GDP- and GTP-bound states promotes attachment/detachment cycles of peroxisomes from microtubules. Similar attachment/detachment cycles were reported to affect the motility of pigment granules, Golgi vesicles, lysosomes and mitochondria , .
In previous work evidence has been provided that the actin cytoskeleton obviously is not involved in supporting peroxisome motility. However, a regulated short-range motor protein-based system may be advantageous, as it provides a mechanism tethering organelles at actin-rich intracellular sites and/or exerting mechanical forces . Although the actual mechanism facilitating peroxisome acto-myosin interaction is not known, binding of RhoA-GTP to the peroxisomal membrane may trigger this process possibly by activating NMM IIA. By uncoupling peroxisomes from microtubules, RhoA-GTP may concomitantly activate ROCKII, a serine/threonine kinase acting immediately downstream of RhoA. ROCKII is known to be involved in reorganizing actin structures preventing actin depolymerization. By phosphorylation of the myosin-binding subunit of myosin phosphatase and/or the myosin light chain of myosin II, ROCKII enhances myosin ATPase activity , . Moreover, ROCKII contains a well-defined pleckstrin homology domain binding to PtdIns(4,5)P2 the synthesis of which in the peroxisomal membrane was recently reported .
Myosin might also act in cargo protein assembly at donor membranes, as recently suggested for the biogenesis of melanosomes , . Scaffolding of melanosome proteins requires coat and adaptor proteins in addition to myosin. Similar processes may also occur in the peroxisomal system where different to catalase N-myc-tagged Pex11bp  and dominant-active RhoA (Fig. 6) both strikingly non-homogenously localize to peroxisomes. Interaction of Pex11βp and RhoA in the peroxisomal membrane and its possible downstream effects on peroxisome biogenesis remain to be established.
Two Rab8 isotypes have been described, Rab8a and Rab8b. They show distinct tissue distributions. Whereas endogenous Rab8a is predominantly expressed in lung, kidney, skeletal muscle and liver, Rab8b is mainly found in testis, spleen, brain, and heart . The high sequence identity of 83% of both genes might explain why exogenous expression of both isotypes in Huh7 cells show the same subcellular localization and develop rather similar phenotypes. In liver Rab8a might be the predominantly expressed endogenous isotype.
So far Rab8 has been implicated in a variety of functions including constitutive biosynthetic trafficking from the trans-Golgi network and the recycling endosome , , sorting of apical membrane proteins , constitutive and regulated melanosome traffic  and regulated ACTH secretion by interacting with TRIP8b . TRIP8b belongs to the tetratricopeptide domain proteins with homology to the peroxisomal targeting signal 1 (PTS1) receptor protein Pex5p known to be involved in peroxisomal matrix protein import –. In vesicle and organelle motility Rab GTPases were implicated in the association of actin motors with the corresponding organellar membranes. Examples are Rab27a and its associated protein MyRIP that mediate melanosome transport by cellular activation of MyoVa and MyoVIIa  and Rab11a involved in endosome recycling by linking the endocytic vesicles via Rab11FIP2 to myosin Vb  as well as Rab8-activated optineurin recruiting MyoVI . There are still other examples of direct and indirect motor protein effectors of Rab GTPases  elucidating multiple ways by which these proteins assure correct transport and vesicle/motor protein interaction.
Our finding of NMM IIA recruiting onto peroxisomes might be interpreted as actin-dependent intracellular peroxisome transport. However, previous studies addressing this topic did not reveal actin-dependent peroxisome motility , . Rather we believe that peroxisomal NMM IIA might locally act in force generation and/or contractile function similar as for example observed in cytokinesis and contractile ring formation , . The GTP dependence of this recruitment suggested the involvement of a GTPase. So far several small GTPases were related to peroxisome biogenesis and function. Among these are Rho1p , Arf1 and Arf3 in S. cerevisiae  and mammalian RhoA, Arf1 and Rab8b , , , . In addition, various Rho-GTPases have been implicated in intracellular vesicle trafficking. These include Cdc42 that inhibits recruitment of the microtubular motor dynein to COP I-coated Golgi vesicles  and RhoA and Rac affecting both clathrin-dependent  and clathrin-independent  endocytosis. The latter process was recently reported to involve Arf family GTPases that might be linked to the Rho family via GIT proteins, Arfaptins or ARAPs . Interestingly, mammalian peroxisomes were shown to bind Arf1 and the COP I coat, and in S.cerevisiae oleate-induced peroxisome proliferation depends on Arf1 and is strikingly stimulated by deleting Arf3 the ortholog of mammalian Arf6 .
In this context the role of Rab8 recruitment to peroxisomes is less clear. Overlap of the Rab8 function with that of Arf6 was reported to occur in the formation of cell protrusions, a process powered by actin polymerization , . As both dominant-active and -inactive Rab8 co-localize with peroxisomes and the dominant-inactive form goes along with an increase in peroxisome number, Rab8 seems to occupy a key position on the peroxisomal membrane regulating proliferation.
ATP and creatine kinase were from Roche Applied Science (Mannheim), creatine phosphate and exoenzyme C3 from Calbiochem (Frankfurt). Polyclonal rabbit anti-non-muscle myosin IIA and anti-ROCKII were obtained from Sigma (Munich). Monoclonal mouse anti-RhoA (26C4) was from Santa-Cruz Biotechnologies (San Diego) and the monoclonal mouse antibody raised against the c-myc epitope (clone 9E10) from Invitrogen (Karlsruhe). Polyclonal antibodies recognizing peroxisomal proteins were raised in rabbits and were previously described . The expression vector for dominant-active RhoA (pEXV-myc-G14VRhoA) was a gift from Dr. Alan Hall (MRC Laboratory for Molecular Cell Biology, University College, London). Exoenzyme C3 (Sigma, Munich) was added to the cell culture medium at a concentration of 5 mg/ml 24 h prior to analysis. The GFP-Rab8a and GFP-Rab8b constructs were previously described . The vector coding for EGFP-PTS1 was obtained from Clontech (Heidelberg). Other chemicals were from Sigma (Munich).
Time-lapse fluorescence microscopy using a cooled ccd camera was carried out exactly as described , . Briefly, time series consisted of 30 pictures taken every 16.5±0.5 s including the 1–2 s time of exposure automatically adjusted according to the fluorescence intensity. Peroxisomal motility was analyzed by animation of the entire time series using the KHOROS software package , . Major events of organelle motility were statistically evaluated by subtraction of two consecutive images using the PMIS software (Photometrics, Tucson, AZ, USA) and counting the number of peroxisomes either at rest or in the saltatory or oscillating state. Only the distance covered by the organelles during the time of two consecutive images (16.5±0.5 s) was taken as a measure to determine the number of saltating peroxisomes. Saltations were recognized as being separated by a distance greater than the maximum amplitude of oscillation that was determined in a separate set of experiments and at most corresponded to peroxisomes appearing attached in the overlay of two subsequent time frames . Statistics were based on counting 100 peroxisomes from 4 to 10 cells of 3–7 independent experiments. The displacement histogram was obtained by analyzing complete time series using particle-tracking velocimetry as described , . On the x-axis length of translations determined at 0.1 pixel distances was plotted versus frequency (y-axis) in percentage of total organelle counts.
Highly purified peroxisomes were isolated by a combination of rate zonal and Nycodenz equilibrium density gradient centrifugation as previously described . They were obtained in a concentration of 2.5–3.5 mg/ml and stored in small aliquots at −80°C prior to use. Rat liver cytosol was prepared by centrifugation of a rat liver homogenate in buffer H (50 mM Hepes/KOH, pH 7.55, 165 mM KAc, 2 mM MgAc2, 1 mM dithiothreitol) supplemented with protease inhibitors (1 mM leupeptin, 10 mM antipain, 100 mM phenylmethylsulfonyl fluoride) . After a final centrifugation at 180,000× g for 90 min (TFT 55.38 rotor, Kontron Instruments, Hanau), the resulting supernatant was concentrated about 10-fold in a MINITAN ultra concentration unit (molecular weight cut-off 10,000; Millipore, Eschborn). The cytosol obtained usually had a final concentration of 45–50 mg/ml. It was stored in aliquots at −80°Cuntil use.
Thawed cytosol (5 mg/ml) was cleared by centrifugation (20,000× g for 20 min at 4°C) and primed by incubation for 30 min at 37°C with either guanosin 5′-(b,-imido)triphosphate (GMP-PNP, 100 µM), or guanosin 5′-(β-thio)diphosphate (GDPγS, 100 µM) or buffer (50 mM HEPES/KOH, pH 7.5). Following centrifugation (200,000× g for 10 min at room temperature) an ATP-regenerating system consisting of 2 mM ATP, 20 mM creatine phosphate, 250 µg/ml creatine kinase in 50 mM HEPES/KOH, pH 7.5 was added. Peroxisomes were freed from Nycodenz by centrifugation and the pelleted peroxisomes (100 mg of protein) resuspended in the cleared cytosol and incubated at 37°C for 30 min. Some incubations contained ATP (2 mM) or AMP-PNP (100 µM) or buffer instead of the ATP-regenerating system. After incubation, organelles were left on ice for 30 min to depolymerize microtubules prior to flotation in a Nycodenz density gradient. To this end, incubated organelles were brought to 45% (w/v) Nycodenz by the addition of 70% (w/v) Nycodenz and overlayed by a layer of 40% (w/v) Nycodenz in sucrose buffer (250 mM sucrose in 10 mM glycylglycine, pH 7.4) followed by a layer of sucrose buffer. Total gradient volume was 600 ml. Centrifugation was carried out at 150,000× g in the SW55II, rotor (Beckman, Krefeld) for 3 h at 4°C. Three fractions indicated as F1, F2 and F3 were collected from top to bottom.
Proteins from reisolated peroxisomes were separated by SDS-PAGE using pre-cast Tris-Tricine 10–20% gradient gels (Anamed, Bad Ems) and subsequently transferred onto PVDF-membranes (Pall, Bad Kreuznach) for 2 hours at 400 mA. Proteins were visualized by enhanced chemiluminescence.
35S-labeled cytosol was isolated from metabolically labelled primary rat hepatocytes grown in DMEM/Hanks'F12 (11) containing 10% FCS, penicillin/streptomycin, 1% L-glutamine, 0.5 µM insulin, 1 µM dexamethasone and 1 µM triiodothyronine over night at 37°C and 5% CO2. For radioactive labelling, cells were incubated for 3 h in methionine/cysteine-free DMEM supplemented with 10 mM HEPES/NaOH, pH 7.4, 10% dialysed FCS, penicillin/streptomycin, 1% L-glutamine, 0.5 µM insulin, 1 µM dexamethasone and 1 µM triiodotyronine in the presence of 37.5 mCi/ml of 35S-methionine/cysteine (>1000 Ci/mmol, GE-Healthcare, Braunschweig). Cells were homogenized on ice in buffer H and the post-nuclear supernatant was centrifuged at 100,000× g in the TLA45 rotor (Beckman, Krefeld) for 1.5 hours at 4°C. Specific activity of the obtained 35S-labeled cytosol was 45 mCi/mg of protein.
Peroxisomes were incubated as described above using metabolically labelled rat liver cytosol in the presence of 2 mM ATP, 100 µM GTPγS and phosphatase inhibitor cocktail 2 (Sigma-Aldrich). Peroxisomes were separated from the bulk of cytosol by Nycodenz-density gradient flotation through a layer of 43% (w/v) Nycodenz and the peroxisomal proteins subjected to SDS-PAGE. To visualize 35S-labelled proteins, gels were dried on Whatman 3 MM paper and exposed to X-ray films (BioMax MR Scientific Imaging, Berlin).
Proteins of the reisolated peroxisomes visualized by autoradiography or by SDS-PAGE were identified by nano-HPLC/ESI-MS/MS . Tryptic peptide samples were analyzed on a Dionex LC Packings HPLC system (Dionex LC Packings, Idstein) directly coupled to a Bruker Daltronics HCT plus ion trap instrument equipped with nanoelectrospray ion source (Bruker, Bremen) and distal coated SilicaTips (New Objective, Wobern, MA, USA). For profiling of proteins associated with peroxisomal membranes, Nycodenz gradient fractions number 1 to 10 were individually treated with NaCO3, pH 11.5. The resulting membrane pellets were then subjected to SDS-PAGE and proteins stained by colloidal Coomassie Brilliant Blue G-250. For analysis of the low-molecular weight (<30 kDa) region the gel was equally cut into three slices and proteins were digested in-gel with trypsin and further subjected to proteomics analysis as described . Briefly, nano-HPLC/ESI-MS/MS analyses were performed on a 7-Tesla Finnigan LTQ-FT (Thermo Electron, Bremen) instrument equipped with a nanoelectrospray ion source. Samples were analyzed in a data-dependent mode fragmenting the three most intense ions in the linear ion trap by low-energy CID and accurately performing in parallel mass measurements in the FTICR cell.
For peptide and protein identification, MS/MS datasets were generally correlated with the rat International Protein Index (IPI; www.ebi.ac.uk) database using the MASCOT algorithm. Proteins were identified based on at least one unique peptide and a false positive rate below 5%. To semi-quantitatively follow proteins of interest across the density gradient fractions, the respective spectral counts were calculated  and plotted against the number of gradient fractions in order to generate protein profiles.
CHO-K1 and Huh7 cells were grown in DMEM containing 10% FCS, penicillin/streptomycin and 1% L-glutamine, whereas AT3F mouse hepatoma cells were maintained in DMEM/Hanks F12 (11) supplemented with 10% FCS, penicillin/streptomycin, 1% L-glutamine, 0.5 µM insulin, 1 µM dexamethasone, 1 µM triiodotyronine at 37°C and 5% CO2. Cells stably expressing EGFP—SKL were prepared as previously described , . Transient expression of myc-G14V-RhoA, a dominant-active mutant of RhoA  was achieved by transfection using the Nucleofector II device (Lonza, Basel) and an optimized protocol for mouse hepatocytes. For exoenzyme C3 treatment, the toxin (5 mg/ml) either was added to the medium two days prior to analysis or was expressed as a myc-tagged protein by transfection . Transfection of GFP-Rab8a and GFP-Rab8b in Huh7 cells was done using calcium phosphate. Cells were selected with 400 mg/l Geneticin for stable transfection and subsequently maintained in medium containing 250 mg/l Geneticin. For immunofluorescence, cells were grown on glass cover slips, fixed in 3% PFA and permeabilized with 1% TX-100 for 5 min at room temperature. After blocking with 10% BSA, cells were incubated with primary antibody for 1 h at 37°C followed by the appropriate FITC-, TRITC- or Alexa FLUOR-labeled secondary antibody (Invitrogen, Karlsruhe). Cover slips were mounted using Mowiol/p-phenylendiamine 101 in 0.1 M Tris/HCl, pH 8.7 (Sigma, Munich) or VectaShield hard set mounting medium (Vector Laboratories). Microscopy was carried out using the confocal laser-scanning microscope LSM Meta 510 (Zeiss, Jena) or an Axio Oberserver SD confocal microscope (Zeiss, Göttingen).
Prior to quantification of co-localizing structures, image stacks were filtered using the Gauss filter implemented in the image aquisition software (Axiovision 4.8, Zeiss, Göttingen) to remove noise. Stacks were subsequently exported as 8-bit greyscale images in tif-format and the analysis was performed using ImageJ (http://rsb.info.nih.gov/ij/) and the OBCOL plugin . The OBCOL plugin represents an automated pipeline for structure segmentation/aggregation into 3D organelles and subsequent statistical evaluation. OBCOL parameters were set to default of the OBCOL pipeline, a Pearson coefficient of ≥0,5 indicated co-localization. For counting the total number of organelles, the „nucleus counter“ plugin of ImageJ was used.
To generate peroxisome proliferating conditions, rats weighing about 150 g were treated with clofibrate (0.5% in the diet) and daily s.c. injections of 100 mg T4 for five consecutive days. Following anesthesia rats were fixed by transcardial perfusion with 2.5% glutaraldehyde in 0.1 M Na-cacodylate buffer, pH 7.6 containing 2% polyvinylpyrrolidone and 0.05% CaCl2. Microslicer liver sections (60–80 mm thick) were incubated in alkaline diaminobenzidine (pH 10.0) for 30 min. Samples were postfixed by cacodylate-buffered 1% osmium tetroxide for 1 h and stained en bloc in 1% uranyl acetate for 30 min. After dehydration in graded ethanol samples were embedded in Epon. Ultrathin sections were stained with lead citrate and analyzed using a Zeiss EM 906E .
We thank Susanne Reusing and Ingrid Kuhn-Krause, Magdalena Pawlas and Christian Bunse for skillful technical assistance and Dr. Allan Hall, MRC Laboratory for Molecular Cell Biology, University College London for the construct encoding myc-tagged exoenzyme C3. We are grateful to Patricia McCabe, FEBS Letters editorial office, for carefully reading the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was supported by the FP6 European Union Project “Peroxisomes in health and disease” (LSGH-CT-2004-512018) as well as by funds from the Deutsche Forschungsgemeinschaft (SFB 642). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.