Peroxisome Subpopulations of the Rat Liver: Isolation by Immune Free Flow Electrophoresis

Reagents

3-(N-morpholino)propanesulfonic acid (Mops), phenylmethylsulfonyl fluoride (PMSF), and dithiothreitol (DTT) were purchased from Sigma (Munich, Germany) and 5-((N-2,3-dihydroxypropylacetamido)-2,4,6-tri-iodo-N,N′-bis(2,3-dihydroxypropyl))-isophthalamide (Nycodenz) from Life Technologies (Eggestein, Germany). All other reagents were from Merck (Darmstadt, Germany) and of the highest grade available.

Antibodies

The preparation of antibodies against peroxisomal enzymes, i.e., catalase, acyl-CoA oxidase (AOx), and urate oxidase (UOx) and assessment of their monospecificity have been established previously (Beier et al. 1988).

A synthetic peptide of the C-terminal 10 amino acids of rat PO membrane protein PMP 70 was coupled to thyroglobulin (Van Regenmortel et al. 1988) and used to immunize rabbits. The appropriate sequence comprises the amino acids KITEDTVEFGS and was chosen because it was not homologous to or identical with any other sequence thus far published (Swiss Prot.). Immunization of the rabbits and purification of the antibody were carried out as outlined (Beier et al. 1988). The antibody was tested by immunoblotting, using the corresponding antigen for preabsorption.

The polyclonal antibodies against the rat hepatic PO membrane proteins PMP 22 and Pex 11 and that against PAF 2 and Pex 5 were kindly provided by Drs. T. Hashimoto (Matsumoto, Japan), T. Osumi (Osaka, Japan), and S. Subramani (San Diego, CA), respectively.

Treatment of Animals

Female SD rats weighing about 250 g were obtained from the Zentrale Versuchstieranlage (University of Heidelberg) where they were kept in accordance with the guidelines of the humane care and use of laboratory animals of the FRG. They were fasted overnight, anesthetized with chloral hydrate, and livers were drained of blood by perfusion with saline (0.9% NaCl, w/v), subsequently excised, dried with filter paper, and weighed. After removal of the connective tissue, they were minced in ice-cold homogenization buffer (HB): 250 mM sucrose, 5 mM Mops, 1 mM EDTA, 0.1% ethanol, 2 mM PMSF, 1 mM DTT, 1 mM ∊-aminocaproic acid.

Subfractionation of Liver Homogenates and Preparation of Heavy, Light, and Postmitochondrial Fractions

Homogenization of the minced tissue and subcellular fractionation by differential centrifugation were performed according to an established method (Völkl and Fahimi 1985), with a few modifications outlined below. Briefly, tissue was homogenized in an ice-cooled Potter-Elvehjem tissue grinder by a motor-driven teflon pestle (1000 rpm) in 5 ml/g HB. The homogenate was then subfractionated in consecutive steps (100 g, 1950 g, 25,500 g), giving rise to pellets (1950 g and 25,500 g) that were resuspended by means of a hand-driven teflon pestle in ice-cold HB (4 and 1 ml/g liver). The suspensions thus obtained correspond to heavy (HM) and light (LM) mitochondrial fractions, respectively. The supernatant recovered after the LM pelleting was layered on a cushion, prepared by dissolving Nycodenz in 5 mM Mops, 1 mM EDTA, 0.1% ethanol, 2 mM PMSF, 1 mM DTT, 1 mM ∊-aminocaproic acid (ρ = 1.20 g/cm³). It was spun in a fixed-angle type rotor (Ti 45; Beckman) at 37,000 × g for 20 min and the fluffy layer banding on top of the cushion, comprising microsomes, lysosomes, and peroxisomes was reloaded onto a fresh Nycodenz cushion and recentrifuged at 55,000 × g for 20 min in the same rotor. The pellet collected was resuspended in ice-cold HB (1.2 ml) using a glass rod and was designated the postmitochondrial fraction (PM). Aliquots of fractions HM, LM, and PM were processed by IFFE as detailed below. In addition, highly purified POs were isolated from an LM fraction by metrizamide density gradient centrifugation (Völkl and Fahimi 1985) and were designated LM_grad-PO.

Incubation of Fractions HM, LM, and PM with Anti-PMP 70 Antibody

One-ml aliquots of the HM, LM, and PM fractions corresponding to about 0.25-1 g of liver were mixed with 1 ml HB containing 1 mg polyclonal anti-PMP 70 antibody, 1 mg of BSA, and leupeptin (1 μM), and the mixtures were incubated at room temperature for >1 hr to immunocomplex the POs of these fractions. Immunoprecipitates and noncomplexed particles were collected by centrifugation for 20 min at 1950 × g (HM), 25,500 × g (LM), and 55,000 × g (PM), and the pellets were suspended in 2 ml medium buffer (MB), pH 8.0, containing 250 mM sucrose, 10 mM triethanolamine, 10 mM acetic acid, and 0.25 mg/ml bovine serum albumin.

Free Flow Electrophoresis

Free flow electrophoresis was performed employing the Octopus instrument (Dr. G. Weber; Kirchheim/Heimstetten, Germany). Just before each run, the separation chamber was washed with MB and electrophoresis was subsequently conducted at 4C in the same buffer with a field of 1000 V and 100 mA and a flow rate of ∼5 ml/fraction/hr. Samples were perfused into the separation chamber at ∼2 ml/hr and fractions were collected through a 96-channel peristaltic pump. Individual runs took approximately 60-90 min. Optical densities (ODs) of fractions were measured at 280 nm and they were concentrated by centrifugation for 20 min at g forces stated above. Pellets were suspended in HB and stored at −80C for enzyme analysis, SDS-PAGE, and immunoblotting.

Protein Assay

Protein was assayed using the Coomassie Blue binding method (Bradford 1976), with BSA as a standard.

SDS-PAGE and Immunoblotting

SDS-PAGE was performed under reducing conditions using 12.5% polyacrylamide gels (Laemmli 1970). Polypeptides resolved were transferred to nitrocellulose (Burnette 1981), which was incubated for 1 hr at 37C with 5% nonfat milk/10 mM Tris buffer, pH 7.4/0.05% Tween-20 to block unspecific binding sites. For immunocomplexing, 10 μg/ml each of PMP 22 and UOx antibodies were mixed with the same amount of one of the other antibodies used (PMP 70 and Pex 11 as well as Pex 5 and PAF 2 in the case of PMP 22; and Cat and AOx in the case of UOx), and incubation was performed at 4C overnight in the same buffer containing 1% nonfat milk. After repeated washing, a peroxidase-conjugated goat anti-rabbit antibody (1:10,000) was added for 1 hr at room temperature and enhanced chemiluminescence (ECL) was used according to the instructions of the manufacturer (Amersham; Braunschweig, Germany) to visualize the immune complexes.

Densitometric Analysis of Immunoblots

Specific signals revealed by ECL, representing the immune complexes of matrix and membrane proteins of PO derived from LM, HM, and PM fractions as well as from an LM density gradient centrifugation (LM_grad-PO), were analyzed by densitometry and ratios of intensities were calculated, with the PMP 22 signal serving as reference for the membrane proteins while the soluble PO matrix enzymes catalase and acyl-CoA oxidase were compared to the insoluble urate oxidase.

Isolation of Peroxisomal Subpopulations by IFFE

The electropherogram shown in Figure 1 exemplifies the fractionation of an LM fraction by means of IFFE. The respective peroxisomes isolated, LM-PO1 collected in fractions 32-34 and LM-PO2 in fractions 37-39, are indicated by circles. The distribution of protein across all fractions, reflected by the optical density at 280 nm (OD₂₈₀) and the determination of catalase, acyl-CoA oxidase, and urate oxidase activities (not shown) consistently reveal a clear-cut separation of nonperoxisomal organelles (first peak) and peroxisomes (second peak and shoulder). These findings, which are in line with corresponding results recently reported (Völkl et al. 1997), demonstrate the feasibility of the IFFE technique for isolation of POs in general and of PO subpopulations in particular.

OPEN IN VIEWER

Figure 1

Distribution of protein across IFFE fractions. LM, HM, and PM fractions were incubated with an antibody against the PO membrane protein PMP 70. The mixtures were centrifuged and the pellets obtained resuspended in MB, pH 8.0, and subjected to FFE as described in Materials and Methods. Consecutive fractions were pooled in pairs, numbered 1 (anode ⁺) through 45 (cathode ⁻), and assayed for protein (OD₂₈₀). Fractions enriched in PO, as revealed by the determination of enzyme activities, are indicated by open circles and designated LM-PO1 (32-34) and LM-PO2 (37-39).

Immunoblotting of the Subpopulations Isolated

PO subpopulations isolated by IFFE from an HM, LM, and PM preparation of a rat liver homogenate were analyzed by SDS-PAGE (Figure 2) and subsequent Western blotting (Figures 3–5) in comparison to highly purified POs (LM_grad-PO) obtained by concomitant metrizamide density gradient centrifugation of the same LM preparation. Whereas the polypeptide patterns of POs immunoisolated from the LM and PM fractions largely coincide with the LM_grad-PO, that of HM-POs does not (Figure 2). This might be due to polypeptides from contaminating organelles, although HM-PO might also represent a unique PO subpopulation.

OPEN IN VIEWER

Figure 2

SDS-PAGE of PO fractions isolated by IFFE. Five μg each of HM-, PM-, LM-PO, and of PO isolated by metrizamide density gradient centrifugation (LM_grad-PO) were loaded onto a 12.5% polyacrylamide gel (8 × 4 × 0.1 cm) and resolved at 25 mA. Polypeptides resolved were stained by Coomassie Blue. Note that the polypeptide patterns of PM- and LM-POs largely coincide with that of LM_grad-PO, whereas that of HM-PO does not.

OPEN IN VIEWER

Figure 3

Densitometric analysis of immunoblots of matrix proteins of POs isolated by IFFE. HM-, PM-, LM-POs and POs isolated by metrizamide density gradient centrifugation (LM_grad-PO) were resolved by SDS-PAGE, blotted onto nitrocellulose, and incubated with antibodies against catalase (Cat), acyl-CoA oxidase (AOx), and urate oxidase (UOx) as described. A peroxidase-conjugated goat anti-rabbit IgG antibody and enhanced chemiluminescence (ECL) were used to visualize immune complexes. Only Subunit B of AOx is shown. Specific signals revealed by ECL were quantified by densitometry and ratios of intensities were calculated, with the UOx signal serving as reference. Appropriate values depicted as bars are shown in the diagram.

Immunoblotting using antibodies against PO matrix enzymes (catalase, acyl-CoA oxidase, urate oxidase) as well as membrane proteins (PMP 22, PMP 70, Pex 11, Pex 5, and PAF 2) revealed differences between the protein-specific signals in a given subpopulation (Figures 3–5). More importantly, however, each protein investigated specifically varied in its signal intensity between LM-, HM-, PM-, and LM_grad-PO, suggesting a different composition of these PO populations. This interpretation is substantiated by the densitometric analysis of the blots and the calculation of relative intensities, with the UOx and PMP 22 signals, respectively, serving as references. As is indicated by the data depicted in the graphs related to the corresponding blots (lower halves of Figures 3–5), individual PO sub-populations indeed differ (a) in the ratios of the soluble matrix enzymes catalase and acyl-CoA oxidase to the insoluble matrix enzyme urate oxidase (Figure 3), (b) in the ratios of the membrane proteins PMP 70 and Pex 11 to the most abundant protein constituent of the PO membrane, i.e., PMP 22 (Figure 4), and (c) particularly in the PMP 22-related ratios of Pex 5 and PAF 2, proteins considered to contribute to the import of PO matrix enzymes. It should be noted in this context, that the highly purified POs (LM_grad-PO) obtained by density gradient centrifugation of an LM preparation contain only a minor amount of Pex 5 and PAF 2, which is almost consistent with the LM PO2 fraction (Figure 5).

OPEN IN VIEWER

Figure 4

Densitometric analysis of immunoblots of membrane proteins of POs isolated by IFFE. HM-, PM-, LM-POs, and POs isolated by metrizamide density gradient centrifugation (LM_grad-PO) were resolved by SDS-PAGE, blotted onto nitrocellulose, and incubated with antibodies against PO membrane proteins PMP 70, Pex 11, and PMP 22 as described. A peroxidase-conjugated goat anti-rabbit IgG antibody and enhanced chemiluminescence (ECL) were used to visualize immune complexes. Specific signals revealed by ECL were quantified by densitometry and ratios of intensities were calculated, with the PMP 22 signal serving as reference. Appropriate values depicted as bars are shown in the diagram.

OPEN IN VIEWER

Figure 5

Distribution of Pex 5 and PAF 2 in POs isolated by IFFE, as revealed by immunoblot and densitometric analysis. HM-, PM-, LM-POs, and POs isolated by metrizamide density gradient centrifugation (LM_grad-PO) were resolved by SDS-PAGE, blotted onto nitrocellulose, and incubated with antibodies against Pex 5, PAF 2, and PMP 22 as described. A peroxidase-conjugated goat anti-rabbit IgG antibody and enhanced chemiluminescence (ECL) were used to visualize immune complexes. Specific signals revealed by ECL were quantified by densitometry and ratios of intensities were calculated with the PMP 22 signal serving as reference. Appropriate values depicted as bars are shown in the diagram.

Distinct Subpopulations of POs Can Be Isolated by IFFE from the Main Subcellular Fractions of Rat Liver

In this study we have used IFFE to isolate POs from an LM (crude peroxisomal) fraction of rat liver, exactly as we described recently (Völkl et al. 1997). The same antibody that bound those peroxisomes (LM-POs) because of its interaction with the cytoplasmic domain of the peroxisomal membrane protein PMP 70 was incubated with two other subcellular preparations, an HM and a PM fraction, and the immunocomplexed particles were subjected to FFE. These main fractions were obtained by differential centrifugation of a rat liver homogenate at 1950 and 55,000 × g, respectively (Völkl and Fahimi 1985) and are mainly enriched in mitochondria (HM) and microsomes (PM). A representative electropherogram of the immunoelectrophoretic fractionation of an LM fraction is shown in Figure 1, with the respective peroxisomes isolated, i.e., LM-PO1 and LM-PO2, collected in the fractions indicated by circles. In accordance with the corresponding HM and PM electropherograms (not shown), it reveals a major peak comprising the bulk of organelles not immunocomplexed and a minor one (LM-PO1) containing POs as identified by marker enzyme activities (not shown). However, in contrast to the fractionation of HM and PM preparations, the minor peak in the LM electropherogram is followed by a shoulder representing a second PO fraction (LM-PO2) according to the results of appropriate enzyme assays (not shown). Apparently the anti-PMP 70 antibody is capable of binding to POs in all the main subfractions of a rat liver homogenate, and IFFE is a versatile tool to isolate such populations.

The POs immunoisolated as described above are expected to differ markedly in their buoyant densities, considering that heavy, light, and postmitochondrial fractions were recovered at quite divergent g forces, i.e., 1950, 25,000, and 55,000 × g, respectively. Indeed, in earlier studies PO fractions have been collected at densities other than ∼1.24 g/cm³ (Klucis et al. 1991; Lüers et al. 1993; Schrader et al. 1994; Wilcke et al. 1995), which is the range in which POs usually band. Considering the divergent concentrations of individual membrane proteins (e.g., Pex 11) as well as the variable ratios of Pex 11 and PMP 70 to PMP 22 in the PO subpopulations immunoisolated (Figure 4), differences in protein composition or membrane permeability might account for the distinct densities.

The Isolated Subpopulations Exhibit Significant Differences in Composition of the Matrix and Membrane Proteins

To figure out whether distinct physicochemical properties account for the distribution of POs to the hepatic subfractions investigated or whether they truly differ in their protein composition, we have used polyclonal antibodies directed against peroxisomal matrix and membrane proteins and immunoblotting to compare HM-, LM-, PM-, and LM_grad-POs (Figures 3–5).

It should be noted that we did not load identical quantities of HM, LM, and PM POs to the SDS gels, taking into account the heterogeneous composition in protein of these fractions; they putatively differ in their ratios of peroxisomal to contaminating organellar (e.g., mitochondrial, microsomal) proteins, and albumin and IgG protein, which are routinely present in the separation medium, probably stick to various extents to the different organelles. Therefore, we analyzed the immunoblots by densitometry and calculated the corresponding ratios to either the peroxisome membrane protein PMP 22 or the insoluble matrix protein UOx. The data presented in the lower halves of Figures 3–5 provide strong evidence that HM-PO, LM-PO1, LM-PO2, and PM-POs are distinct subpopulations because of the divergent proportions of their matrix and membrane proteins. Therefore, the ratios of the soluble matrix enzymes catalase and acyl-CoA oxidase to the insoluble urate oxidase are quite similar in LM-PO1 and LM-PO2, as well as in LM_grad-PO, yet markedly differ in PM- and HM-POs (Figure 3). This might just reflect different degrees of extraction due to the divergent g forces applied to pellet HM-, LM-, and PM-POs. Indeed, POs are quite fragile organelles and PO matrix proteins are commonly extracted at quite different rates; in particular, catalase leaks out quite easily (Alexson et al. 1985). However, the ratios of acyl-CoA oxidase to UOx are significantly lower than that of catalase in PM- and HM-POs, although acyl-CoA oxidase is less extractable than catalase (Alexson et al. 1985).

Differences were also observed in the distribution of the membrane proteins PMP 70, Pex 11, and PMP 22 (Figure 4). Whereas the latter was present in all fractions, albeit in different concentrations, only quite low amounts of PMP 70 and Pex 11 could be detected in HM-POs. Distinct ratios of Pex 11 and PMP 70 to PMP 22 were found in LM PO1 and LM PO2, in line with the distribution of Pex 5 and PAF 2 (Figure 5), which are mostly enriched in LM-PO1. In view of the role attributed to these proteins, which are considered as constituents of the PO import mechanism (Erdmann et al. 1997; Faber et al. 1998), the HM-, PM-, and LM-POs should exhibit differences in import competence, with LM-PO1 probably being the most competent.

The observations presented here gain particular importance in view of the recent debate on the role of the endoplasmic reticulum (ER) in the biogenesis of POs (Erdmann et al. 1997). The prevailing view on that issue has changed several times during the past three decades. The original model proposed that POs are formed by budding from the ER (Novikoff and Shin 1964). That hypothesis, however, could not be maintained after the discovery that PO proteins are synthesized on free ribosomes. The current concept envisages that new POs originate by division from preexisting ones and that POs grow by posttranslational import of matrix and membrane proteins synthesized on free ribosomes (Lazarow and Fujiki 1985). Recently, however, the potential involvement of the ER in the biogenesis of POs has gained some new support on the basis of genetic, biochemical, and morphological findings. Prompted by these recent observations, Erdmann et al. (1997) have hypothesized that there may be a vesicle-mediated transport from the ER to POs in the very early stages of PO biogenesis. Although this new view is intriguing, explaining some recent data that are difficult to reconcile with the current model, it is far from being settled. In particular, the preperoxisomal vesicles postulated to bud from the ER and to fuse either heterotypically with preexisting POs or to undergo homotypic fusion, thus forming new POs, still await identification and characterization. In this respect, immunochemical isolation methods such as IFFE, employing antibodies raised against specific membrane proteins involved (peroxins), might provide indispensable tools for isolation of such vesicles and elucidation of their role in the sequential events of PO biogenesis, e.g., by combination with pulse-labeling studies.