A heterotrimeric G protein-phospholipase A₂ signaling cascade is involved in the regulation of peroxisomal motility in CHO cells

During the last few years much effort has been made to understand the intracellular translocation and positioning of membranous organelles. Various cytoskeletal structures, particularly the microtubules, play an important role in this process as they are the tracks along which organelles move specifically and over long distances (Vale et al., 1985; Allen et al., 1985; Cole and Lippincott-Schwartz, 1995; Rodionov and Borisy, 1997). For this purpose microtubules are equipped with a variety of associated proteins (MAPs) which drive organelle transport (for review see Mandelkow and Mandelkow, 1995) and mediate its regulation. Among these proteins the microtubule associated motor proteins, kinesin and cytoplasmic dynein, are well characterized (Walker and Sheetz, 1993; Holzbaur and Vallee, 1994). Each of them belongs to a family of proteins which generate movement at the expense of ATP- hydrolysis (Paschal and Vallee, 1987; Astumian and Bier, 1996). The motor proteins are heterooligomers and contain both a microtubule and a membrane binding site. First experiments with chicken brain microsomes have revealed that specific binding to kinesin is mediated by kinectin, an integral 160 kDa membrane protein of the endoplasmic reticulum and other subcellular organelles, which acts as an anchor for kinesin-dependent vesicle motility (Yu et al., 1992; Toyoshima et al., 1992; Vallee and Sheetz, 1996). An antibody directed against kinectin was able not only to inhibit kinesin-driven motility but to reduce cytoplasmic dynein-dependent motility by about 50% as well, suggesting that one and the same anchor interferes with both motor proteins (Yu et al., 1992).

Interaction between microtubule associated motor proteins and the membrane anchor is necessary but not sufficient in mediating organelle motility. Both biochemical reactions and additional accessory components may be essential to drive membrane movement (Schroer and Sheetz, 1991; Holzbaur and Vallee, 1994). Agents that affect intracellular levels of cAMP have been shown to influence the frequency of linear movements of vesicles in neurons and kidney cells (Hamm-Alvarez et al., 1993; Azhderian et al., 1994). The in vivo phosphorylation of kinesin heavy and light chains, kinesin associated proteins as well as dynein heavy, intermediate and light chains is well documented (Hollenbeck, 1993; Matthies et al., 1993; Lin et al., 1994; Dillman and Pfister, 1994). Furthermore, okadaic acid treatment of cytosolic extracts which increased kinesin motor activity also led to hyperphosphorylation of several proteins that coprecipitated with kinesin. Thus, phosphorylation not only of kinesin but also of kinesin-binding proteins and MAPs seems to be important in regulating organelle motility (for reviews see Cole and Lippincott- Schwartz, 1995; Vallee and Sheetz, 1996).

Recently, we described a system which allowed us to analyze peroxisomal motility in intact and permeabilized CHO cells (Rapp et al., 1996). The studies revealed that the majority of peroxisomes within a cell moves slowly by thermal oscillation, whereas about 15%-20% of the entire organelle population moves rapidly along microtubules. This latter movement is saltatory in that a stationary oscillating peroxisome after binding to a microtubule track is rapidly translocated along a certain distance before it is released from the microtubule and again becomes stationary oscillating. These data were recently confirmed by Wiemer et al. (1997) in a study utilizing CV1 cells. Actin filaments known to mediate movement of axoplasmic vesicles in squid and mammals (Kuznetsov et al., 1992; Fath et al., 1994) seem to influence peroxisomal motility only indirectly, possibly by interfering with micotubules.

Knowing that peroxisomes are transported intracellularly along microtubules, we became interested in the factors regulating peroxisomal movements. As mentioned above, protein kinases which are frequently regulated by G proteins seem to play an essential role in regulating organelle motility. Therefore, we focused our interest on the effects of GTP in this process. As a model system, we used CHO cells stably transfected with GFP carrying the carboxy-terminal peroxisomal targeting signal 1 (GFP-PTS1) and time-lapse imaging to record peroxisomal movements. In this study we provide evidence that peroxisomal motility in CHO cells is regulated by a heterotrimeric G protein of the class Gi/Go which acts within a phospholipase A₂ (PLA₂) signaling pathway.

CHO-K1 cells (ATCC, Rockville, USA) were grown in α-MEM containing 7.5% fetal calf serum, 100 units/ml penicillin and 100 μg/ml streptomycin (Biochrom KG, Berlin, Germany). For fluorescence microscopy, cells were plated on 32 mm dishes containing a central glass coverslip (MatTek Corp., Ashland, MA, USA).

Vectors containing the cDNA encoding wild-type GFP and the S65T mutant GFP (Cubitt et al., 1995) were kindly provided by Dr H.-H. Gerdes (Universität Heidelberg, Heidelberg, Germany). The C- terminal peroxisomal targeting signal 1 (PTS1) consisting of the amino acids serine-lysine-leucine (Gould et al., 1988) was introduced by PCR into the 170 basepair MunI fragment of the GFP cDNA. This fragment was inserted in frame behind the NcoI/MunI fragment of the S65T mutant cDNA. Subsequently, the entire cDNA was excised with EcoRI and inserted into the mammalian expression vector pcDNA3 (Invitrogen, De Schelp, The Netherlands).

CHO cells were lipofected using the DOTAP transfection reagent (Boehringer Mannheim, Mannheim, Germany). Cells were grown to 60-80% confluency on 10 cm plates. For transfection of one plate, 10 μg of vector DNA were linearized with BglII and purified with the Geneclean III kit (Bio 101, Vista, CA, USA). Transfected cells were selected by 0.5 g/l geneticin (Sigma, Deisenhofen, Germany) and stable transfectants were obtained after about 2-3 months of selection.

Nucleotidetriphosphates were depleted by incubating the cells for 7 minutes in a glucose-free medium containing 10 mM NaN₃ and 6 mM 2-deoxyglucose as described by Soto et al. (1993). The cytoskeletal active drugs nocodazole, colcemid, taxol and cytochalasin A, B and D were from Sigma (Deisenhofen, Germany) and prepared as 100- fold stock solutions (1 mg/ml) in dimethyl sulfoxide. Dibutyryl- cAMP (Sigma, Deisenhofen, Germany), A-3 (Biomol, Hamburg, Germany) and H-89 (Calbiochem, Bad Soden, Germany) were dissolved in water as 100-fold stock solutions and applied at a concentration of 200 μM, 50 μM and 50 μM, respectively. Okadaic acid (Boehringer Mannheim, Mannheim, Germany) was added to the cells from a 2,000-fold concentrated stock solution (0.2 mM) in ethanol. The aluminum fluoride complex was prepared by mixing aqueous solutions of AlCl₃ (10 mM) and NaF (6 M) in water immediately before use. The compounds were added to the medium at a concentration of 50 μM (AlCl₃) and 30 mM (NaF). Mastoparan, its inactive derivative Mas-17, kindly provided by Dr J. B. Helms (Universität Heidelberg, Heidelberg, Germany), and the more potent analog Mas- 7 were added at concentrations of between 15 and 50 μM from 1 mM stock solutions in 10 mM Tris-HCl, pH 7.4. Guanosine-5′-O-(3-thiotriphosphate) (GTPγS; Sigma, Deisenhofen, Germany) was introduced into the cells by microinjecting a 200 μM solution in microinjection buffer (Rapp et al., 1996). The calcium ionophores ionomycin (Biomol, Hamburg, Germany) and calcimycin (A23187; Sigma, Deisenhofen, Germany) were used at final concentrations of 10 μM, and time-lapse imaging was started 5 minutes after drug addition. In the case of inositol-1,4,5-trisphosphate (IP₃; Biomol, Hamburg, Germany), time-lapse imaging was started immediately after injection of a 2 mM solution in microinjection buffer. LiCl (Merck, Darmstadt, Germany) was given to the cells 20 minutes prior to analysis at a concentration of 60 mM. 1,2-Dioctanoyl-glycerol, ET-18-OCH₃ (edelfosine) and U-73122 were purchased from Biomol (Hamburg, Germany). Stock solutions of 1,2-dioctanoyl-glycerol and ET-18- OCH₃ in ethanol were added to the medium at final concentrations of 100 μM, whereas U-73122 was dissolved in dimethyl sulfoxide immediately before use and added to the cells at a final concentration of 10 μM. The PLA₂ inhibitors isotetrandrine, ONO-RS-082 and aristolochic acid (Biomol, Hamburg, Germany) were prepared as stock solutions in dimethyl sulfoxide and added to the medium at final concentrations of 20 μM to 40 μM, 12 to 25 μM, and 100 to 200 μM, respectively. Free arachidonic acid (Fluka, Neu-Ulm, Germany) or its sodium salt coupled to fatty-acid-free BSA (Serva, Heidelberg, Germany) in a molar ratio of 7:1 were given to the cells at a concentration of 1 mM. Lysophosphatidic acid (C18:1) (Sigma, Deisenhofen, Germany) was bound to fatty-acid-free BSA at a weight ratio of 1:10 in PBS and applied at a concentration of 50 μM. PLAP was dissolved in microinjection buffer at a concentration of 100 μg/ml and subsequently injected into the cells. If two drugs were applied in one experiment, they were added to the cells successively with a time span of 30 minutes in between. If not mentioned otherwise, time-lapse imaging was started 30 minutes after drug administration.

Cholera toxin and pertussis toxin (Biomol, Hamburg, Germany) were purchased as lyophilized preparations. They were dissolved in H₂O and added to the medium 24 hours prior to fluorescence microscopy in concentrations of 2 μg/ml and 50-100 ng/ml, respectively. Since GFP-PTS1 in the pcDNA3 vector is under the control of the cytomegalovirus promotor, expression could be stimulated by treating the cells with 5 mM sodium butyrate 24 hours before starting the experiment. Prior to fluorescence microscopy, the culture medium was replaced by a Hanks’ buffered salt solution containing 1.8 mM Ca²⁺. Only in the case of microinjection was a Ca²⁺-free Hanks’ solution used. When cells were depleted of ATP, analysis was performed in a glucose-free phosphate buffered salt solution. Cells were depleted of Ca²⁺ by a 45 minute incubation in Ca²⁺-free medium containing 5 mM EGTA. A constant temperature of 37°C was maintained during drug treatment and fluorescence microscopy analysis.

Time-lapse imaging and fluorescence microscopy were performed on a Zeiss Axiovert 10 inverted microscope (Zeiss, Oberkochen, Germany). Images were acquired with a Photometrics CH250 ccd camera (1,317×1,035 pixels) as described by Herr et al. (1993) using a 63-fold magnification objective (Zeiss Plan-Achromat, ×63, NA 1.4), each pixel thus corresponding to a 108×108 nm area. The series consisted of 31 pictures taken every 16.5±0.5 seconds including the 1-2 seconds time of exposure which was adjusted automatically according to the fluorescence intensity. The photostability of the GFP- PTS1 allowed acquisition of multiple time series from the same cell. Finally, peroxisomal motility was analyzed by animation of the entire time series using the KHOROS software package (Rasure et al., 1990; Herr et al., 1993). In order to document the major events of organelle motility visible in the films, statistics were carried out by subtraction of two consecutive images using the PMIS software (Photometrics, Tucson, AZ, USA) and counting the number of peroxisomes which were either at rest or in the saltating or oscillating state. Only the distance covered by the organelles during the time of two consecutive images (16.5±0.5 seconds) was taken as a measure to determine the state of motility. Resting peroxisomes are not visible on the difference image, whereas oscillating ones overlap or at most are attached to each other. This covered distance corresponds to the maximum amplitude of oscillation which was determined in a separate set of experiments. Peroxisomes which moved faster, i.e. which were separated by a distance greater than the maximum amplitude of oscillation, were regarded as moving in a saltatory manner. It has to be noted that a minor inaccuracy is inherent to these statistics resulting in a slight overestimation of the number of resting and saltating peroxisomes. An oscillatory peroxisome, on the one hand, which by chance occupies the same position on two consecutive images may be counted as resting. On the other hand, a saltating peroxisome moving for only a fraction of the time span in between two consecutive images (16.5±0.5 seconds) may be counted as an oscillatory one. In spite of these minor inaccuracies, the construction of difference images, in our hands, is a most suitable way to analyze a large number of organelles yielding statistical data that reliably describe the processes seen in the time series. For detection of mean displacements of randomly thermically oscillating peroxisomes, we confirmed the motility state of a given peroxisome by repeated animation of the time series prior to analyzing. If not mentioned otherwise, all statistics are based on counting 100 peroxisomes from 4 to 10 cells of 2 to 7 independent experiments. The displacement histogram was obtained by analyzing complete time series (31 frames) using particle tracking velocimetry as described by Hering et al. (1997).

Cells stimulated by sodium butyrate as described above were washed in PBS, pH 7.4, and after fixation for 20 minutes in 3.7% formaldehyde in PBS they were permeabilized by 1% Triton X-100 in PBS for 4 minutes. Subsequently, the cells were washed again with PBS containing 25 mM glycine and incubated with the primary antibody at 37°C for 30 minutes. Since GFP occupied the FITC-channel, TRITC-conjugated second antibodies (Dianova, Hamburg, Germany) could be used in order to decorate the first antibody. F-actin was visualized by FITC-phalloidin, which was kindly provided by Dr H. Faulstich (Max Planck-Institut für Medizinische Forschung, Heidelberg, Germany). Antibodies against the lysosomal marker cathepsin D were kindly provided by Dr F. Momburg (Deutsches Krebsforschungszentrum, Heidelberg, Germany).

Mutant GFP (S65T) was targeted to peroxisomes via the C- terminal PTS1 (SKL) which was introduced into the cDNA by PCR. The GFP-PTS1 encoding pcDNA3 construct was used to establish stable transfectants of wild-type CHO cells. Cytomegalovirus promotor driven expression was stimulated by sodium butyrate 24 hours prior to fluorescence microscopy (Palermo et al., 1991). Peroxisomes were brightly labelled under these conditions (Fig. 1A) and revealed the typical punctate pattern known from immunofluorescence staining (Soto et al., 1993; Rapp et al., 1993). Specific peroxisomal import of GFP-PTS1 was demonstrated by immunocytochemical colocalization with the peroxisomal marker catalase (Fig. 1B). Occasionally, in addition to the clear green fluorescence produced by GFP-PTS1, a yellow particulate fluorescence was observed in the transfected CHO cells. Frequently, this fluorescence colocalized with a lysosomal marker (cathepsin D) (not shown) and most likely is derived from partially degraded GFP-PTS1.

For fluorescence microscopical examination of peroxisomal motility, cells expressing GFP-PTS1 were plated onto glass coverslips and peroxisomal motility was analyzed by timelapse imaging. According to our own data (Rapp et al., 1996) and the most recent observations of Wiemer et al. (1997), two distinct movements were observed, an oscillating and a saltating one. The majority of peroxisomes moved slowly by local thermal oscillations, and about 15 to 20% of the entire peroxisomal population moved in a saltatory manner (Fig. 2A- C) exhibiting maximal velocities of up to 2 μm/second. This saltatory movement by which the organelles travel long distances within the cell is mediated by microtubules, as it is clearly abolished by colcemid and nocodazole, two drugs which depolymerize microtubules (Table 1). A third state of motility was observed under conditions of ATP depletion which was characterized as a completely motionless ‘frozen’ state (Fig. 2D-F), obviously caused by the inability of the attached organelles to be released from the microtubules (Rapp et al., 1996). This idea was strongly supported by the observation that peroxisomes bound to microtubules by either ATP depletion or, as shown below, by AlF₄⁻ (Table 1, Fig. 3A) or mastoparan, start to oscillate again following depolymerization of microtubules. Conversely, depolymerization of microtubules completely prevented peroxisomal arrest caused by these conditions (Table 1, Fig. 3B), leaving the organelles in the oscillating state. That about 25% of these organelles were classified as resting (Table 1) is explained by the high probability by which randomly oscillating peroxisomes occupy the same position on two consecutive images.

Although it has been reported that the microtubular system only contributes to a minor extent to cytosolic viscosity (Sato et al., 1988; Janmey et al., 1991; Tsai et al., 1996), we determined the influence of cytoskeletal poisons and various drugs used in this study on the amplitude of random peroxisomal oscillations in a separate set of experiments. Table 2 shows the results which demonstrate that the mean displacement of oscillating peroxisomes was hardly altered by these treatments. Thus, changes in cytoplasmic viscosity as they may occur as a result of cytoskeletal disruption (Tsai et al., 1996) seem to have only a minor influence on passive peroxisomal motility.

In the case of mitochondrial motility, a Ca²⁺-activated ATP- dependent proteolysis of MAPs, particularly MAP2, has been discussed to underly the mechanism of alternate periods of attachment and detachment of mitochondria and microtubules (Leterrier et al., 1994). We therefore studied the influence of Ca²⁺ on peroxisomal motility by incubation of the cells in the presence of 5 mM EGTA for 45 minutes which effectively lowers the level of intracellular Ca²⁺ (Capiod et al., 1982; Berthon et al., 1983). Under these conditions of low Ca²⁺ concentration, binding of peroxisomes to microtubules was abolished, leaving the organelles in an oscillating state. Binding was prevented even upon subsequent ATP depletion (Fig. 3C, Table 1), or the addition of AlF₄⁻ or mastoparan (not shown), conditions which in controls effectively led to attachment of peroxisomes (see below). On the other hand, a high level of intracellular Ca²⁺, as generated by the calcium ionophores calcimycin or ionomycin, lead to peroxisomal arrest (Table 1). These data suggest that Ca²⁺ may be involved in peroxisomal motility by mediating the binding of the organelles to microtubules.

Indications for the interaction of peroxisomes with cytoskeletal structures other than microtubules have not been obtained so far, although electron micrographs occasionally show peroxisomes to be closely associated with filamentous structures (Gorgas, 1985; Rapp et al., 1996). Various cytochalasins added to the incubation medium in concentrations of 10 μg/ml, which rapidly lead to depolymerization of actin microfilaments (Yahara et al., 1982), hardly influenced peroxisomal motility and only slightly decreased the number of saltating peroxisomes (Table 1).

As discussed previously, the motile behavior of a single peroxisome is best characterized by alternate cycles of binding and release from microtubules, hence alternating between oscillating and resting or saltating states of motility. This behavior is not only typical for peroxisomes but has also been described with minor variations for other subcellular organelles and vesicles (Couchman and Rees, 1982; Hollenbeck and Bray, 1987; Forman et al., 1987; Swanson et al., 1992; Morris and Hollenbeck, 1993; Lippincott-Schwartz et al., 1995). How are these processes coordinated and what are the underlying mechanisms of their regulation? The local Ca²⁺ concentration might be only one of many potential factors involved in this process.

In order to study the effect of GTP on peroxisomal motility, we microinjected the GFP-PTS1 expressing cells with various concentrations of GTPγS, a nonhydrolyzable GTP analog. Whereas a concentration of 0.5 mM was toxic for the cells, as manifested by rounding and detachment from the substratum, concentrations up to 0.2 mM were tolerated without recognizable detrimental effects. Microinjection of 0.2 mM GTPγS resulted in the complete loss of peroxisomal motility, resembling the ‘frozen’ state observed after ATP depletion (Fig. 4 and Table 1). However, loss of motility was not an immediate mastoparan mutant peptide Mas-17 was used that lacks the specific G protein activating properties but still is membrane active. As shown in Fig. 6C and Table 1, Mas-17 (50 μM) did not affect peroxisomal motility, suggesting that mastoparan specifically affects a G_i/G_o protein.

Both G_i and G_o proteins are substrates of pertussis toxin which catalyzes their ADP-ribosylation and hence prevents G protein activation by GTPγS and mastoparan (Tamura et al., 1982). Preincubation of the cells for 16-20 hours with pertussis toxin per se did not influence peroxisomal motility (Fig. 7A), but completely abolished arrest of peroxisomal motility by GTPγS and mastoparan (Fig. 7B,D). Since ADP-ribosylation process but became more and more pronounced over a time span of about 30 minutes after which all peroxisomes were immobilized (Fig. 5).

There are two main intracellular regulators which both might have been influenced by GTPγS, heterotrimeric and small GTP-binding proteins. The possible involvement of trimeric G proteins in mediating peroxisomal motility was investigated by utilizing the well established activating properties of both the aluminum fluoride complex (AlF₄⁻) and mastoparan. AlF₄⁻ is known to stimulate trimeric G proteins in general (Kahn, 1991), whereas mastoparan, a wasp venom toxin, specifically activates the G_i/G_o subgroup of trimeric G proteins (Higashijima et al., 1990). Similar to GTPγS, both AlF₄⁻ and mastoparan added to the medium in concentrations of 50 μM AlCl₃/30 mM NaF and 20-50 μM mastoparan completely blocked peroxisomal motility (Fig. 6A,B). NaF (30 mM) alone was without effect (Table 1). Since mastoparan is known also to interact unspecifically with membranes, as a control the of G_i/G_o proteins catalyzed by pertussis toxin does not prevent G protein activation by AlF₄⁻ (reviewed by Helms, 1995), the effect of AlF₄⁻ was not abolished (Fig. 7C).

The observation that GTPγS, AlF₄⁻ and mastoparan all affect peroxisomal motility raised the question as to how these drugs act on this process. For this reason, we tested the implication of the well known signal transduction pathways that utilize either cAMP, IP₃/DAG or arachidonic acid/lysophosphatidic acid as second messengers.

Several lines of evidence suggest that the cAMP cascade is not involved. Both cholera toxin which is proposed to catalyze ADP-ribosylation of G_sα and incubation of cells with the membrane permeable cAMP analog dibutyryl-cAMP had no effect on peroxisomal motility (Table 3). Similarly, the broadrange protein kinase inhibitors A-3 and H-89, the latter of which preferentially inhibits protein kinase A (Inagaki et al., 1986; Chijiwa et al., 1990), and okadaic acid (100 nM), a specific inhibitor of type 2A protein phosphatases (Cohen et al., 1990) previously shown to enhance vesicle motility (McIlvain et al., 1994), also did not affect peroxisomal motility (Table 3).

The involvement of the IP₃/DAG-dependent signal transduction pathway was examined by using the cell permeable phospholipase C inhibitors ET-18-OCH₃ (edelfosine) and U- 73122 (Powis et al., 1992; Smith et al., 1990). None of them did influence peroxisomal motility per se, nor could prevent AlF₄⁻-induced peroxisomal arrest (Table 3). In a second set of experiments, we directly investigated the effect of the two second messengers, diacylglycerol and IP₃, generated by G protein activated phospholipase C. Release of the second messengers was imitated by either incubating the cells with the diacylglycerol (DAG) analog 1,2-dioctanoyl-glycerol or microinjection of high concentrations of inositol-1,4,5-trisphosphate (IP₃), known to promote release of Ca²⁺ from intracellular stores (Majerus et al., 1986). These treatments as well as the incubation of cells in the presence of LiCl which inhibits the regeneration of phosphoinositol-4,5-bisphosphate did not cause any significant effect (Table 3), suggesting that an IP₃/DAG-dependent signal transduction pathway may not be responsible for the observed peroxisomal arrest.

A third widely distributed signaling cascade triggers the formation of arachidonic acid and lysophosphatidic acid from phospholipid stores by PLA₂. In CHO cells, this signaling pathway has been shown to be coupled to G proteins of the class G_i/G_o (Hunt et al., 1994). Only arachidonic acid but not lysophosphatidic acid abolished peroxisomal movements (Fig. 8A, Table 3). Consistent with this observation, microinjection of a PLA₂ activating protein peptide (PLAP) (Fig. 8B, Table 3) which specifically activates PLA₂ (Clark et al., 1991) completely inhibited peroxisomal motility, suggesting that PLA₂ is a major target of the G protein. This idea was further confirmed by the use of three specific inhibitors of the PLA₂ cascade, namely isotetrandrine, which blocks G protein-mediated activation of PLA₂ (Akiba et al., 1992), as well as aristolochic acid and ONO-RS-082, drugs both of which specifically inhibit PLA₂ enzymatic activity (Rosenthal et al., 1989; Banga et al., 1986). By themselves, these inhibitors did not visibly influence peroxisomal motility, a result which is in agreement with the observation that pertussis toxin per se also did not affect peroxisomal movements (see above). However, if CHO cells were pretreated with these drugs, the inhibitory effects of AlF₄⁻ and mastoparan on peroxisomal motility were largely abolished (Fig. 8C, Table 3). Peroxisomes were still motile under these conditions, with the number of oscillating peroxisomes similar to that in the controls. In cells treated with mastoparan, the percentage of saltating peroxisomes was also largely restored, whereras saltations rarely occurred in AlF₄⁻-treated cells (Table 3).

With respect to their motility, peroxisomes of CHO cells can occupy three distinct states, the resting, the oscillating or the saltating state. As revealed by the frequency versus displacement plot shown in Fig. 9, under control conditions most peroxisomes are oscillating (displacements of ∼0.05 to ∼0.35 μm) with smaller portions of the peroxisome population being in the resting (displacements <0.1 μm) or saltating state (displacements >0.35 μm). Resting peroxisomes are bound to microtubules as visualized by in vivo labeling of microtubules (Rapp et al., 1996), but are not actively moved. Consequently, this state appears under conditions that allow microtubular binding but exclude activation of the motors that drive peroxisomal motility. Since all motor proteins involved in organelle motility known to date are ATPases, depletion of ATP completely immobilizes peroxisomes.

Most of the time peroxisomes are in the oscillating state in which they seem not to be attached to any cytoskeletal structure and hence move passively by random thermal oscillations. Due to the small size of CHO cell peroxisomes (about 100 nm in diameter) the oscillating movements are easily recognized by time-lapse imaging fluorescence microscopy. The same may not hold for mitochondria which on average are at least 10 times larger than peroxisomes.

In the saltating state, peroxisomes are linked to microtubules and are actively transported within the cell. As with other cell organelles, transport may be mediated primarily by motor proteins of the kinesin superfamily (Hollenbeck and Swanson, 1990; Bloom and Endow, 1994; Rodionov et al., 1993) or cytoplasmic dynein family (Hirokawa et al., 1990; Lacey and Haimo, 1992; Aniento et al., 1993). Both motor proteins might be involved, since active movement of peroxisomes was observed in both directions towards the cell periphery as well as towards the cell center (Rapp et al., 1996; Wiemer et al., 1997). Mediation of saltatory transport of peroxisomes by microtubules, furthermore, is demonstrated by its selective sensitivity to drugs that affect microtubule assembly such as nocodazole and colcemid (Table 1). Addition of these drugs to the medium completely blocked saltations, leaving the organelles in the oscillating state. The computer analysis (Fig. 9) also reveals that the frequency of displacements in the range of oscillatory movements compared to the control decreases upon treatment with nocodazole which may be explained by the loss of active microtubule-based movements. Thus, the difference between control and nocodazole treatment would account for about 35% of the entire peroxisomal population being actively transported. However, in Tables 1 and 3, we only considered those peroxisomes as being actively moving which exhibit displacements of ≥0.35 μm. Within this range the nocodazole histogram declines to almost zero. Thus, all peroxisomes exhibiting displacements of ≥0.35 μm are actively moving. They account for approximately 18% of the total population (Fig. 9).

Although depolymerization of microtubules and particularly of actin microfilaments has been reported to decrease cytoplasmic viscosity (Tsai et al., 1996), oscillatory movements of peroxisomes as measured by their average amplitude (Table 2) are not substantially altered in comparison to untreated cells (cf. Wiemer et al., 1997). Hence, disruption of microtubules rather than changes in cytoplasmic viscosity may be the cause for the alterations in peroxisomal motility. In addition, peroxisomes immobilized by AlF₄⁻ or ATP-depletion return to the oscillating state upon disrupting the microtubules, suggesting that the peroxisomes are resting due to their immobilization on intact microtubules.

Although the number of saltating peroxisomes compared to the controls is somewhat smaller in cells in which the actin network has been removed by various cytochalasins (Table 1), this does not necessarily prove a direct interaction of peroxisomes with actin microfilaments, but rather may relate to the manifold structural and functional links between the actin and the microtubule system (Sattilaro et al., 1981; Fath and Lasek, 1988; Kuznetsov et al., 1992; Fath et al., 1994). The in any case only slightly reduced number of saltating peroxisomes in cytochalasin-treated cells, therefore, is likely to be the result of an indirect influence of the drugs on microtubular structure and function.

Three main aspects of the regulation of peroxisomal motility emerged from our results.

• Ca²⁺ is involved in attachment of peroxisomes to microtubules, as suggested by the observation that treatment of CHO cells with Ca²⁺ ionophores results in immediate and complete immobilization of peroxisomes (Table 1). Does this mean that the observed G protein-mediated peroxisomal arrest is caused by a high increase in intracellular Ca²⁺ levels? To answer this question, we tested the influence of several G protein activating drugs on the level of intracellular Ca²⁺ by using FURA-2 microfluorometry (Grynkiewicz et al., 1985; Nobiling and Bührle, 1989). Neither AlF₄⁻, mastoparan nor arachidonic acid altered the level of intracellular Ca²⁺ concomitant with their effect on peroxisomal motility (not shown). On the other hand, both application of extracellular ATP during time-lapse imaging (not shown), which causes a transient increase of intracellular Ca²⁺ in CHO cells (Iredale and Hill, 1993), and microinjection of IP₃ (Table 3), known to promote release of Ca²⁺ from intracellular stores (Majerus et al., 1986), did not affect peroxisomal motility. From these results we might conclude that the G protein activating drugs used in this study mediate attachment of peroxisomes to microtubules by signaling events distinct from those attributed to increases of intracellular Ca²⁺. Compared to the effects of the Ca²⁺ ionophores, we might assume that the ATP- or IP₃-induced increase in the Ca²⁺ level was not sufficiently high or did not affect peroxisomal motility due to intracellular channelling.

  However, physiological levels of intracellular Ca²⁺ seem to be important for peroxisomal-microtubular interactions, since incubation of the cells with EGTA which is known to effectively lower intracellular Ca²⁺ (Capiod et al., 1982; Berthon et al., 1983) completely prevented peroxisomal arrest caused by ATP depletion or treatment of the cells with AlF₄⁻. Several reports (Breuer et al., 1992; Matthies et al., 1993; Leterrier et al., 1994; Rivera et al., 1995) suggest a requirement for Ca²⁺ in organelle motility. Whereas the data of Matthies et al. (1993) indicate Ca²⁺ to modulate kinesin ATPase activity in vitro by the interaction of Ca²⁺-calmodulin with kinesin light chains, Leterrier et al. (1994), investigating mitochondrial motility, postulate a Mg²⁺/Ca²⁺ dependent proteolysis step of MAPs, predominantly MAP2, which may facilitate subsequent binding of kinesin (Lopez and Sheetz, 1993; Hagiwara et al., 1994).

• A heterotrimeric G protein of either the class G_i or G_o is involved in the regulation of peroxisomal motility, as derived from the inhibitory activity of GTPγS, AlF₄⁻ and mastoparan. Whereas GTPγS activates G proteins in general, AlF₄⁻ only acts on heterotrimeric ones (Kahn, 1991) and mastoparan more specifically interacts with G_i/G_o proteins resulting in their activation (Higashijima et al., 1990). The inhibitory effect on peroxisomal motility of mastoparan and GTPγS, but not of AlF₄⁻ was abolished by pretreatment of the cells with pertussis toxin. This toxin selectively catalyzes the ADP-ribosylation of G_i/G_o, thus preventing their activation (Tamura et al., 1982). Since AlF₄⁻ can activate even ADP-ribosylated G proteins (reviewed by Helms, 1995), its activating properties were not abolished by pertussis toxin.

  Our data indicate that a heterotrimeric G protein of the class G_i/G_o participates in the regulation of peroxisomal motility raising the question as to the cellular location of the G protein. In addition to the plasma membrane, where G_i/G_o are confined to the inner aspect and are known to operate in signal transduction, G proteins were also found on various intracellular membranes, like the Golgi apparatus, the trans-Golgi network, chromaffin granules, rough and smooth endoplasmatic reticulum, and synaptic vesicles, where they may participate in the regulation of vesicular transport, exocytosis or budding of vesicles from the trans-Golgi network (Toutant et al., 1987; MelanHon et al., 1987; Audigier et al., 1988; Ercolani et al., 1990; Leyte et al., 1992; Ohashi and Huttner, 1994). There are also some reports dealing with the interaction of G_i/G_o with tubulin (Roychowdhury et al., 1993; Yan et al., 1996) as well as with microtubules (Wu and Lin, 1994). Consequently, G_i/G_o involved in peroxisomal motility might be localized to the plasma membrane, however, location to peroxisomes or to the interacting microtubules cannot be excluded. Similar to the mechanism of the regulation of kinectin function, recently proposed by Sheetz and Yu (1996), peroxisomal or microtubular G_i/G_o could interfere with either motor protein or kinectin function as a central control point for the coordinated regulation of both motors. Location to the plasma membrane, on the other hand, may be indicative of a classical G protein triggered signal transduction cascade.

• The G_i/G_o protein mediates peroxisomal arrest by activating PLA₂. Specific information on the involvement of this signaling pathway was derived by pharmacologically affecting PLA₂ activity as well as by studying the effects of the PLA₂ products, lysophosphatidic acid and arachidonic acid. Whereas lysophosphatidic acid apparently had no influence, elevation of cellular levels of arachidonic acid either by microinjecting a PLA₂ activating protein peptide (Clark et al., 1991) or by adding arachidonic acid to the medium imitated the effect of G protein activation causing peroxisomal arrest. On the other hand, drugs that inhibit PLA₂ activity per se did not influence microtubulemediated peroxisomal motility. However, they largely abolished peroxisomal arrest induced by mastoparan, Mas-7 or AlF₄⁻, and restored oscillating movement of peroxisomes. In the case of mastoparan or Mas-7, motility was similar to that observed in control cells, except a somewhat reduced number of saltating peroxisomes (Table 3). In cells treated with AlF₄⁻, however, oscillations but not saltations could be restored, suggesting AlF₄⁻ to exert side effects or to influence additional components participating in the regulation.

There are several indications arguing against the involvement of the cAMP or the IP₃/DAG signaling cascades in regulating peroxisomal motility. First, incubation of the cells, for example, with dibutyryl-cAMP, a membrane permeable cAMP analog, did not show any significant influence on peroxisomal motility. Accordingly, cholera toxin which is known to constitutively activate adenylate cyclase through ADP-ribosylation of G_sα, A- 3 and H-89, two potent broad-range inhibitors of protein kinase activity (Inagaki et al., 1986; Higashijima et al., 1990), as well as the inhibition of type 2A protein phosphatases by okadaic acid were all without effect (Table 3). In a similar way, peroxisomal motility was not affected by mimicking IP₃/DAG- dependent signal transduction by microinjection of IP₃ or treatment of the cells with 1,2-dioctanoyl-glycerol, and was not influenced by the application of the phospholipase C inhibitors U-73122 and ET-18-OCH₃, or by LiCl. All these drugs did not prevent peroxisomal arrest caused by AlF₄⁻, indicating that the IP₃/DAG-dependent signal transduction cascade is not involved. As verified by FURA-2 microfluorometry measurements, the observed G_i/G_o triggered signal transduction events are clearly not elicited by Ca²⁺ which may operate downstream of IP₃/DAG in cell signaling. Therefore, we assume that the effects of Ca²⁺ depletion or ionophore treatment are caused by interaction with other components involved in peroxisomal motility, particularly the motor proteins. This idea is in line with observations that demonstrate redistribution of cytoplasmic dynein from lysosomes to the cytoplasm upon Ca²⁺ depletion (cf. Lin et al., 1994), and binding of Ca²⁺-calmodulin to kinesin which has been shown to inhibit the ATPase activity of native kinesin (Matthies et al., 1993).

GTP plays an important role not only in the regulation of peroxisomal motility but may also be involved in the regulation of the transport of other cell organelles. Bloom et al. (1993) recently reported on the involvement of GTP in the regulation of fast axonal transport which was inhibited by GTPγS. Similar to our observations on peroxisomes, fast axonal transport was not affected by protein phosphatase inhibitors such as okadaic acid, or broad spectrum inhibitors of protein kinases such as K-252a. However, in contrast to our results, fast axonal transport was not perturbed by AlF₄⁻, indicating that the regulation of this type of motility includes GTP-dependent processes others than those related to trimeric G proteins. Most interestingly, intracellular distribution of pigment granules by directional migration in chromatophores of various fish, amphibians, reptiles and crustaceans is subject to G protein-mediated regulation. Following extracellular stimuli, these systems frequently respond to elevated intracellular Ca²⁺ levels by granule aggregation and to cAMP by granule dispersion (Oshima et al., 1986; Kotz and McNiven, 1994), and have been used to assay G protein coupled receptor functioning (Lerner, 1994).

The necessity for active and fast organelle transport is obvious in certain specialized cells like neurons or chromatophores.

Synaptic vesicles, for example, have to be transported from the soma to the synapse, the site of their action, and pigment granules have to be dispersed or concentrated in the case of color change. However, active organelle transport may also serve housekeeping functions like maintainance of the perinuclear location of the Golgi apparatus and the lysosomal compartment, the active extension of the endoplasmic reticulum towards the cell periphery as well as maintainance of steady states present in endomembrane systems that continuously communicate by budding and fusion events. Accordingly, peroxisome-microtubule interactions may serve a biogenic purpose, since at least in some tissues such as sebaceous glands, or in cultured HepG2 cells, peroxisomes are frequently found in close contact to microtubules and undergo severe morphological changes upon partial differentiation and exponential growth, respectively (Gorgas and Zaar, 1984; Gorgas, 1984; Rapp et al., 1996; Schrader et al., 1996).

In various tissues, peroxisomes are found at preferential intracellular locations, e.g. highly ordered at the surface of lipid droplets in sebaceous glands (Gorgas, 1987), or concentrated within the basolateral cytoplasm of polarized kidney cells (Zaar et al., 1984), indicating a requirement for their translocation and positioning. During mitosis a regular dispersion of peroxisomes in the mother cell caused by microtubular transport may guarantee an even distribution of peroxisomes to the daughter cells. Transport of peroxisomes to the cell periphery may also ensure the rapid metabolism of incoming peroxisomal substrates such as eicosanoids (Jedlitschky et al., 1991) which exert potent regulatory functions, whereas transport towards the cell center may be a prerequisite for the organelle’s own catabolism. At first glance, it may be puzzling that motility of subcellular organelles and cellular vesicles is the subject of diverse regulation. However, as all these organelles serve distinct functions within the cell, differential regulation of organelle transport may be of vital importance for correct cell functioning.

We gratefully acknowledge the expert assistance of Dr Rainer Nobiling (Universität Heidelberg, Chirurgische Universitätsklinik, Abteilung Experimentelle Chirurgie) in FURA-2 microfluorometry and Dr Frank Hering (Universität Heidelberg, Interdisziplinäres Zentrum für wissenschaftliches Rechnen) in particle tracking velocimetry. Furthermore, we thank Drs Felix Wieland and Bernd Helms for critically reading the manuscript and many stimulating comments throughout the work. Ursula Jäkle is gratefully acknowledged for her skillful technical assistance. This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB 352) and the Biochemical Instrumentation Programme at the European Molecular Biology Laboratory (EMBL).