Identification of the Kinesin Kifc3 As a New Player for Positioning Of

Biochimica et Biophysica Acta 1833 (2013) 3013–3024

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

journal homepage: www.elsevier.com/locate/bbamcr

Identiﬁcation of the kinesin KifC3 as a new player for positioning of peroxisomes and other organelles in mammalian cells

Denise Dietrich a, Florian Seiler a, Frank Essmann b, Gabriele Dodt a,⁎ a Interfaculty Institute of Biochemistry, Cell Biochemistry, University of Tuebingen, D-72076 Tuebingen, Germany b Interfaculty Institute of Biochemistry, Molecular Medicine, University of Tuebingen, D-72076 Tuebingen, Germany article info abstract

Article history: The attachment of organelles to the cytoskeleton and directed organelle transport is essential for cellular Received 25 February 2013 morphology and function. In contrast to other cell organelles like the endoplasmic reticulum or the Golgi Received in revised form 19 July 2013 apparatus, peroxisomes are evenly distributed in the cytoplasm, which is achieved by binding of peroxisomes Accepted 2 August 2013 to microtubules and their bidirectional transport by the microtubule motor proteins kinesin-1 (Kif5) and cyto- Available online 13 August 2013 plasmic dynein. KifC3, belonging to the group of C-terminal kinesins, has been identiﬁed to interact with the human peroxin PEX1 in a yeast two-hybrid screen. We investigated the potential involvement of KifC3 in perox- Keywords: Kinesins isomal transport. Interaction of KifC3 and the AAA-protein (ATPase associated with various cellular activities) Peroxisomes PEX1 was conﬁrmed by in vivo colocalization and by coimmunoprecipitation from cell lysates. Furthermore, Peroxins knockdown of KifC3 using RNAi resulted in an increase of cells with perinuclear-clustered peroxisomes, indicating Cytoskeleton enhanced minus-end directed motility of peroxisomes. The occurrence of this peroxisomal phenotype was cell cycle phase independent, while microtubules were essential for phenotype formation. We conclude that KifC3 may play a regulatory role in minus-end directed peroxisomal transport for example by blocking the motor function of dynein at peroxisomes. Knockdown of KifC3 would then lead to increased minus-end directed peroxisomal transport and cause the observed peroxisomal clustering at the microtubule-organizing center. © 2013 Elsevier B.V. All rights reserved.

1. Introduction distributed in mammalian cells due to binding of peroxisomes to microtubules [3–5] and transport of peroxisomes along microtubules in mam- Bidirectional transport of organelles, vesicles, protein complexes and malian and Drosophila cells by the motor proteins cytoplasmic dynein mRNA by motor proteins is essential for cellular morphology and func- and kinesin-1 [6–9]. Live cell imaging after staining with ﬂuorescent tion. Myosins that move along microﬁlaments are responsible for slow dyes [6,10] revealed that only a small part of peroxisomes is directed short-range transport whereas the microtubule motors of the kinesin by microtubule-associated motor proteins in a fast long-range manner superfamily and cytoplasmic dynein perform long-range movements [4,5]. in mammalian cells. Compared to dynein, representing a minus-motor Besides uniform cellular distribution, peroxisomal motility also and therefore transporting cargoes to the cell center, the kinesin's direc- allows response to varying metabolic conditions by transport to certain tion of transport is dependent on the position of their motor domain cellular areas. Interactions of peroxisomes with other cellular organelles [1,2]. N-kinesins, like the so-called conventional kinesin (kinesin-1, like the endoplasmic reticulum (ER) or lipid droplets have been shown Kif5), harbor their motor domain at the N-terminus and are involved [11–13] and seem to play an important role in peroxisomal lipid metab- in plus-end directed transport to the cell periphery whereas C-kinesins olism [14]. Furthermore, peroxisomes are able to interact and form com- are typically minus-motors. In contrast to other cell organelles like the plex peroxisomal networks [6]. Recent live cell imaging data showed Golgi apparatus or the endoplasmic reticulum, peroxisomes are evenly long-range movements of peroxisomes resulting in close contacts with other peroxisomes [15]. However, exchange of proteins or metabolites between single peroxisomes has not been detected yet [15–17]. Abbreviations: AAA-protein, ATPase associated with diverse cellular activities; CLIP, A yeast two-hybrid screen revealed KifC3, a kinesin family protein, cytoplasmic linker protein; CoIp, Coimmunoprecipitation; EEA, early endosomal antigen; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; HEPES, 4-(2-hydroxyethyl)-1- as potential binding partner of the AAA-protein (ATPase associated piperazineethane sulfonic acid; Kif, kinesin superfamily protein; MDCK-II cells, Madin with various cellular activities) PEX1. KifC3 is a C-terminal kinesin of Darby canine kidney II cells; MAPK, mitogen-activated protein kinase; PEX, peroxin; the kinesin-14 family and is ubiquitously expressed [18,19]. The corre- PMP, peroxisomal membrane protein; PTS-1, peroxisomal targeting sequence-1; TGN, sponding human gene is located on chromosome 16q13–q21 [20] a trans-Golgi network locus that is also affected in Bardet–Biedl syndrome [21], which is char- ⁎ Corresponding author at: Interfaculty Institute of Biochemistry, Hoppe-Seyler-Str. 4, 72076 Tuebingen, Germany. Tel.: +49 7071 2973349; fax: +49 7071 295191. acterized by retinal dystrophy amongst other symptoms [22].AsKifC3 E-mail address: [email protected] (G. Dodt). maps within the critical region of this disease and constitutes one of

0167-4889/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbamcr.2013.08.002 3014 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 the most abundant expressed kinesins in retina and retinal pigment 10% fetal calf serum, 2 mM L-glutamine and 0.1 mM (50 mg/mL) genta- epithelium [23], involvement of KifC3 in Bardet–Biedl syndrome micin. For transient DNA transfection cells in 25 cm2 dishes or flasks was considered [20,23] but has not been confirmed yet. However, were treated with 5 μg plasmid DNA using jetPEI (PeqLab) according KifC3−/− knockout mice did not show any physiological phenotype to the manufacturer's instruction. The transfected cells were harvested [24] appearing healthy and developing normally. Minus-end directed 24 to 48 h post transfection or prepared for immunofluorescence mi- motor activity of KifC3 was followed in vitro and a role of KifC3 in api- croscopy, respectively. The endoplasmic reticulum (ER) was labeled cal transport of TGN (trans-Golgi network)-derived vesicles was by transfection of 1 μg pDsRed2-ER vector 24 h prior to fixation, shown in vivo in polarized MDCK II cells (Madin Darby canine accordingly. kidney) [19]. Xu et al. also postulated that KifC3 participates in posi- Microtubules were depolymerized by treating COS-7 cells with 5 μM tioning of the Golgi apparatus under cholesterol depleted conditions [25]. nocodazole (stock: 10 mM in DMSO (dimethyl sulfoxide)) for 4 h prior Mutations in the PEX1 gene are one of the most common causes of to fixation. Cytochalasin B was added at a concentration of 4 μM(stock: Zellweger syndrome [26,27] and the identification of KifC3 as potential 10 mM in DMSO (dimethyl sulfoxide)) and was incubated for 48 h until binding partner suggested involvement of this kinesin in peroxisomal cells were prepared for immunofluorescence microscopy. Synchroniza- motility. The interaction of PEX1 and KifC3 was analyzed in vivo as tion of cells was achieved by adding 10 mM thymidine for 24 h prior to well as in vitro using coimmunoprecipitation. The role of KifC3 in perox- fixation. isomal transport was investigated by means of knockdown experiments using KifC3-mRNA targeting siRNA and peroxisomes were analyzed for 2.4. Indirect immunofluorescence their distribution. For indirect immunofluorescence microscopy cells were seeded on 2. Material and methods coverslips 24 h before preparation as described before [32]. Primary antibodies were diluted as indicated and were applied to the cells for 2.1. Plasmids 30 min. After extensive washing cells were treated with secondary goat or donkey α-mouse or α-rabbit antibodies conjugated with Alexa The plasmid containing the coding sequence for human KifC3 in the Fluor 488 or Alexa Fluor 594 (Molecular Probes, Invitrogen) for 10 min. retroviral pLPCX vector and the plasmid coding for EGFP-tagged murine Rabbit polyclonal antibodies against human KifC3 (ProteinTech KifC3 in the C1 vector were kind gifts of George Stark (Cleveland, USA), Group) were used in a 1:50 dilution. For detection of β-tubulin I + II Masatoshi Takeichi (Kobe, Japan) and Nobutaka Hirokawa (Tokyo, or α-tubulin monoclonal mouse antibodies (Sigma) were applied in a Japan) and have been described previously [28,29]. The pDsRed2-ER 1:100 dilution. Polyclonal antibodies, raised against the first N-terminal vector used for fluorescent labeling of the endoplasmic reticulum 133 amino acids of human PEX14 in rabbit, were diluted 1:400 [33]. encodes the red fluorescent protein DsRed2 fused to the ER-targeting The sheep α-PMP70 antibodies were a kind gift of Stephen Gould and sequence of calreticulin and the ER retention sequence coding for KDEL were diluted 1:100. The myc-tag was detected using monoclonal mouse and was purchased from Clontech. PEX1-myc in pcDNA3.1zeo that antibodies (9B11, epitope: EQKLISEEDL) from Cell Signaling in a 1:2000 expresses the open reading frame of PEX1 with a C-terminal myc-tag dilution. Monoclonal mouse α-EEA-1 and α-GM130 antibodies, pur- was a gift of Herma Portsteffen (Bochum, Germany). Changes to the chased from BD Biosciences, were diluted 1:100 and 1:50, respectively. reference PEX1 sequence that do not affect complementation efficiency Actin filaments were detected using phalloidin, tagged with are G90R and A113T. tetramethylrhodamine B isothiocyanate (Sigma), in a dilution of 1:500 analogous to primary antibodies. For staining of mitochondria 2.2. Yeast two-hybrid screening cells were treated with MitoTracker Red CMXRos (Molecular Probes) at a concentration of 37.6 μM in medium for 30 min at 37 °C prior to The human PEX1 open reading frame was amplified by PCR from fixation. pBM65 using GD 311(5′-CGC GGA TCC GTT TGT GGG GCA GCG ATC Immunofluorescence stained cells were analyzed using a Zeiss GCC TGG CG-3′) and GD 312 (5′-CGC GGA TCC TTA TGC TAA AGT TAC Axiovert 200 M fluorescence microscope equipped with an AxioPlan TTT CTG TCC AGG TCG-3′) as primer and cloned into the BamHI site of Apochromat 1.4 63× oil objective and an AxioPlan Neofluar 1.3 100× the yeast two-hybrid vector pGBKT7 (Clontech). This plasmid coding oil objective and images were acquired with an AxioCam MRm camera for PEX1 N-terminally tagged with a GAL4 DNA-binding domain was in combination with AxioVision 4.7.2 software. transformed into Saccharomyces cerevisiae AH 109 (mating type a) using the lithium acetate method by Schiestl and Gietz [30]. The haploid 2.5. siRNA knockdown yeast strain Y187 (mating type α) carrying a HeLa cDNA library (pGADT7-rec) was purchased from Clontech and the titer was deter- COS-7 cells were electroporated twice at a 24 h interval using mined with 1.38 ∗ 108 cfu/mL. After mating, interaction of proteins 1nmol(20μM) siRNA (siKifC3 sense: 5′-GGCUGUGCACGAGAAUCU resulted in expression of the GAL4 regulated reporter gens HIS3, ATT-3′, siKif5B sense: 5′-GCCUUAUGCAUUUGAUCGGTT-3′ or AllStars ADE2, lacZ and MEL1. Positive clones were first selected using drop- Negative Control siRNA, Qiagen) as described before [32]. Cells were out plates TDO (-Leu/-Trp/-His) and QDO (-Leu/-Trp/-His/-Ade) and harvested 24 h and 48 h after the second electroporation for RNA then assayed for α-D-galactosidase and β-D-galactosidase activity. In extraction and protein analysis, respectively or prepared for immunoflu- total 1.01 ∗ 107 clones were screened and the prey plasmid DNA of orescence staining 48 h after the second siRNA treatment. 162 positive clones was sequenced. All yeast two hybrid methods followed the protocols as described in the Yeast Protocols Handbook 2.6. RNA extraction, cDNA synthesis and quantitative real-time PCR (Clontech PT3024-1). measurements

2.3. Cell culture and transient DNA transfection Total RNA of siRNA-treated COS-7 cells was extracted (RNA spin Kit, PrepEase, USB) and 500 ng of total RNA were transcribed to cDNA SV40T-transformed PEX1-deficient human fibroblasts (PBD009T) (iScript, BioRad) according to the manufacturer's instructions. For [31] were a kind gift of Stephen Gould (Baltimore, USA). COS-7 (ATTC, relative quantification of cDNA amounts real-time PCR reactions were CRL-1651) and A549 cells (ATTC, CCL-185) were obtained from Hans performed by use of MaximaTM SYBR Green/ROX qPCR Master Mix Probst (Tübingen, Germany). All cell lines were cultured at 37 °C and (Fermentas) in the StepOne Plus Real-time PCR System (Applied

8.5% CO2 in Dulbecco's modiﬁed Eagles' medium complemented with Biosystems) under following conditions: 50 °C for 2 min, 95 °C for D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 3015

10 min, 40 cycles at 95 °C for 15 s and 60 °C for 1 min followed by using buffer D with 0.1% BSA (w/v) and incubated with mouse α-myc melting curve analysis. Data were collected at the 60 °C step. The antibodies (Cell Signaling) in a 1:100 dilution overnight at 4 °C. The mRNA knockdown levels of KifC3 and Kif5B were calculated via the beads were washed using buffer D containing 0.1% BSA (w/v) and 2−ΔΔct method using GAPDH (Glyceraldehyde 3-phosphate dehy- 0.5% Triton X-100 (v/v) and resuspended in 200 μLofcelllysates. drogenase) as reference [34]. The primer pairs were designed as: After incubation for 2 h at 4 °C the beads were separated by a magnet glyceraldehyde 3-phosphate dehydrogenase (GAPDH) forward: and the supernatants were mixed with SDS-loading buffer (125 mM 5′-CATCAAGAAGGTGGTGAAGCAG-3′ and reverse: 5′-CATCAAGAA Tris pH 6.8, 138 mM SDS, 10% (v/v) β-mercaptoethanol, 20% (v/v) glyc- GGTGGTGAAGCAG-3′ and for kinesin superfamily protein 5B (Kif5B) erin, 0.29 mM bromphenol blue) and heated to 80 °C for 5 min. After forward: 5′-gagtgctgagattgattctgatg-3′ and reverse: 5′-cggagatctgcat several washing steps using buffer D containing 1% Triton (v/v), 0.5% tatcacg-3′. The primer pair for the Kinesin superfamily protein BSA (w/v), 0.2% sodium deoxycholic acid (w/v) and protease inhibitor C3 (KifC3) was purchased from Qiagen (Hilden, Germany) as cocktail (Sigma, 1:200) SDS-loading buffer was added to the beads, QuantiTect®Primer Assay, Hs_KifC3_1_SG. heated to 80 °C for 5 min and the eluate was separated from the beads. All samples were stored frozen at −20 °C. 2.7. Western blotting 3. Results Cell extracts containing 40 μg of total protein were separated on 8% SDS-PAGE gels, modified according to Laemmli [35]. Separated proteins A yeast two-hybrid screen with human PEX1, an AAA-ATPase that were transferred onto nitrocellulose membranes via semidry blotting is involved in the late steps of peroxisomal matrix protein import, (BioRad) for 1 h at 20 V in 20 mM Tris, 150 mM Glycin, 1.73 mM SDS revealed KifC3 as potential binding partner. To gain independent data and 20% (v/v) methanol. Membranes were blocked with 5% (w/v) for the interaction of these proteins we coexpressed both proteins in non-fat dry milk in TBS-Tween (1 M Tris, pH 7.4, 1 M NaCl, 0.1% (v/v) mammalian cells, analyzed their localization in vivo and performed Tween20) overnight at 4 °C. Rabbit polyclonal α-KifC3 antibodies coimmunoprecipitations. Furthermore, KifC3 was transiently over- (Proteintech Group) were diluted 1:333 in TBS-Tween containing 5% expressed in COS-7 cells and knockdown experiments were performed (w/v) non-fat dry milk and incubated for 2 h at room temperature. using RNAi to obtain information on KifC3 function in peroxisomal For detection of MAPK (mitogen activated protein kinase) as loading motility. control mouse α-MAPK antibodies (QIAexpress® Tag•100, epitope of MAP kinase 2: EETARFQPGYRS, Qiagen) were diluted 1:1000 in TBS- 3.1. PEX1 and KifC3 are able to interact in vivo Tween containing 5% (w/v) non-fat dry milk. After washing with TBS and TBS-Tween membranes were decorated either with goat α-rabbit A HeLa cDNA library (1.38 ∗ 108 cfu/mL) in S. cerevisiae Y187 was or goat α-mouse antibodies conjugated with horseradish peroxidase screened using PEX1 as bait. PEX1 (pGBKT7) was transformed into (Sigma) at a 1:15,000 dilution in TBS-Tween containing 5% (w/v) non- AH109. The expressed prey proteins harbored the GAL4 transcription- fat dry milk. After conducting the same washing procedures membranes activating domain so that interaction of proteins resulted in expression were developed by enhanced chemiluminescence (Pierce ECL Western of GAL4 regulated reporter gens after mating of the two different yeast Blotting substrate) and exposed to a photosensitive film (Amersham strains. In total 107 clones were screened and the strains were mated Hyperfilm™ ECL, GE Healthcare). with an efficiency of 11%. After selection the prey plasmid DNA of 162 positive clones was sequenced. 2.8. Analysis of relative DNA content by flow cytometry Among the positive clones four different PEX6 fragments

(PEX6389–980,PEX6430–980,PEX6665–980 and PEX6706–980)could COS-7 cells of a 35 mm dish were trypsinized 48 h after the second be identiﬁed, indicating the principle functionality of this screen. siRNA treatment and washed with PBS buffer (10 mM Na2HPO4, The shortest detected PEX6 fragment (PEX6706–980), comprising 1.8 mM KH2PO4, 140 mM NaCl, 2.7 mM KCl). Cell pellets were thesecondAAA-domainD2andtheC-terminus,reﬂects the es- resuspended in 0.5 mL PBS and transferred into 4.5 mL ice cold 70% sential PEX1 binding region. Tamura et al. [36] described a some-

(v/v) ethanol. After 2 h of ﬁxation on ice cells were washed again with what longer region in PEX6593–980 that showed a week interaction to PBS buffer. Pellets were resuspended in 1 mL of propidium iodide stain- PEX1 using a yeast two-hybrid assay, although a stronger binding was ingsolution(PBSbuffercontaining0.1%TritonX-100,0.2mg/mLRNAse achieved if both AAA-domains were present. The binding of our detected

Aand50μg/mL propidium iodide) and incubated for 15 min at 37 °C. shorter PEX6 (PEX6706–980) fragment to PEX1 might just be strong At least 104 stained cells were analyzed on a LSR II (BD Biosciences) enough to be caught with the screen. flow cytometer and cell cycle phase subpopulations were quantified One potentially interesting candidate found was KifC3 that could using the FACS Diva Software. be identified four times. The interacting fragments all basically corre- spond to the C-terminal 160 amino acids of KifC3 and cover the 2.9. Coimmunoprecipitation from cell lysates C-terminal motor domain. Besides PEX6- and KifC3-fragments, several cytoskeleton-associated proteins were found. For preparation of cell lysates COS-7 cells of a 25 cm2 flask were To test for in vivo interaction of PEX1 and KifC3, PEX1-deficient transiently transfected with 5 μg PEX1-myc and 2.5 μg pcDNA3.1zeo, human fibroblasts were transiently transfected with plasmids coding 2.5 μg KifC3-pLPCX and 5 μg pcDNA3.1zeo or 5 μg PEX1-myc and for PEX1-myc and human KifC3, respectively. Cellular localization of 2.5 μg KifC3-pLPCX, respectively. 48 h after transfection cells were the expressed proteins was analyzed 48 h post transfection by indirect harvested, washed twice with Hank's buffer and lysed for 30 min on immunofluorescence microscopy (Fig. 1A). Overexpressed KifC3 local- ice using buffer D (20 mM HEPES, 110 mM potassium acetate, 5 mM ized at microtubules confirmed by colocalization with tubulin α stain- sodium acetate, 2 mM magnesium acetate, 1 mM EDTA, pH 7.3) con- ing whereas PEX1-myc was mainly detected at peroxisomes together taining 0.5% Triton X-100 (v/v), 0.5% BSA (w/v), 0.2% sodium with PEX14 as a peroxisomal membrane protein marker. However, deoxycholic acid (w/v) and protease inhibitor cocktail (Sigma, 1:200). coexpression of PEX1-myc and KifC3 resulted in a microtubular staining After centrifugation for 15 min at 16,100 ×g and 4 °C supernatants pattern of PEX1-myc colocalizing with KifC3 positive structures. This were subjected to coimmunoprecipitation. indicates that PEX1 is able to interact with KifC3 in human fibroblasts For coimmunoprecipitation analysis Dynabeads® M-280 coupled to when overexpressed. Coexpression of PEX1 and KifC3 in ΔPEX1 cells sheep α-mouse IgG (Invitrogen) were used according to the did not abolish peroxisome complementation visualized by peroxisomal manufacturer's instruction. 50 μL beads were prepared as indicated localization of α-catalase and α-SKL antibodies, respectively. However, 3016 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024

Fig. 1. KifC3 and PEX1 interact in vivo. (A) PEX1-deficient human fibroblasts were transiently transfected with PEX1-myc and human KifC3, either with one plasmid alone or in combination. 48 h after transfection cells were prepared for immunofluorescence and incubated with α-KifC3, α-tubulin α, α-myc, α-PEX14 antibodies and the corresponding fluorescence- tagged secondary antibodies. Overexpressed KifC3 was localized at microtubules (compare Fig. 2) whereas PEX1-myc was mainly found at peroxisomes. Upon coexpression of KifC3 and PEX1-myc, PEX1-myc was detected at microtubules where it colocalized with KifC3. (B) Interaction of KifC3 and PEX1-myc was analyzed by coimmunoprecipitation (CoIp) using cell lysates of COS-7 cells transiently expressing PEX1-myc and KifC3, either alone or in combination. PEX1-myc was captured to magnetic beads via α-myc antibodies and was able to precipitate KifC3. Representative Western blots of CoIp input and supernatant samples (right panel) and eluates (left panel) using α-KifC3 and α-PEX1 antibodies are shown (repeated four times). Intensity of KifC3 eluate bands was quantified and revealed an intensity doubling when PEX1-myc and KifC3 were expressed in combination (100%) compared to samples expressing KifC3 alone (50%). Marked bands at 55 kDa (*) are caused by the applied IgGs and were used as loading controls for quantification. Scale bar: 10 μm. complementation rates were slightly decreased compared to samples significantly enriched in the presence of PEX1-myc in four independent expressing PEX1 alone (data not shown). experiments. Quantification of KifC3 eluate bands in Fig. 1B(leftpanel) In addition, coimmunoprecipitation (CoIp) experiments were revealed an intensity doubling from 50% to 100% upon expression of performed. COS-7 cell lysates transiently expressing either PEX1-myc KifC3 and PEX1-myc compared to samples expressing KifC3 alone. The or human KifC3, or both proteins in combination were treated with observed shift of KifC3 bands to higher molecular weights in the input α-myc antibodies coupled to magnetic beads to precipitate PEX1-myc. and supernatant samples (Fig. 1B, right panel) is caused by high concen- KifC3 and PEX1 protein levels were analyzed using α-KifC3 and trations of BSA (bovine serum albumin) with similar molecular weight α-PEX1 antibodies after SDS page and blotting onto nitrocellulose. used to minimize unspecific binding. For the same reason, no KifC3 PEX1-myc successfully associated with the beads and KifC3 was eluted double bands could be detected (compare Figs. 2Band3A) in the eluate in samples that coexpressed PEX1-myc and KifC3 (Fig. 1B, left panel), in- samples (Fig. 1B, left panel) whereas replicates with reduced BSA levels dicating binding of PEX1 and KifC3 in vivo. Although KifC3 unspecifically during CoIp elution showed KifC3 double bands indicating interaction of bound to the beads under various conditions (data not shown), it was PEX1 with different KifC3 isoforms (data not shown). D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 3017 A

Fig. 2. Localization of endogenous and overexpressed KifC3. (A) COS-7 cells were processed for immunofluorescence microscopy using α-KifC3 antibodies. During interphase endogenous KifC3 localized near the nucleus and concentrated at perinuclear spots, which were most likely the microtubule organizing center (MTOC) as indicated by tubulin α staining. Visualization of peroxisomes by use of α-PMP70 antibodies did not reveal any colocalization of KifC3 with peroxisomes. During metaphase KifC3 was detected at the spindle fibers and spindle poles. (B) SDS-PAGE and western blot analysis using α-KifC3 antibodies revealed distinct expression levels of endogenous KifC3 in different cell lines with highest expression levels in A549 and COS-7 cells. To correct for the different loading of the total protein amount (HELA: 40 μg, A549: 10 μg, HEK 293T: 40 μg, HepG2: 40 μg, GM5756T: 40 μg, COS-7: 13 μg, COS-7 + KifC3: 83 ng) additional analysis with α-MAPK (mitogen activated protein kinase) antibodies is shown. (C) Overexpression of KifC3 in COS-7 cells resulted in a microtubular staining pattern 24 h after transient transfection confirmed by colocalization with tubulin α. Just like endogenous KifC3, overexpressed KifC3 was not located at peroxisomes and it did not influence the peroxisomal distribution. Scale bar: 10 μm.

3.2. Localization of endogenous KifC3 was cell cycle phase dependent patterns depending on the corresponding cell cycle phase (Fig. 2A). In interphase cells, KifC3 was localized near the nucleus and concentrated In order to reveal potential involvement of KifC3 in peroxisomal at the centrosome as indicated by merging with the tubulin α staining transport or distribution we examined the localization of endogenous (Fig. 2A, first row). This was not unexpected as KifC3, a C-terminal KifC3. COS-7 cells were fixed and prepared for immunofluorescence kinesin, is supposed to function as a microtubule minus-motor and microscopy using α-KifC3 antibodies and Alexa Fluor 594 labeled has been detected at perinuclear regions in Caco2 cells [29].Incontrast, secondary antibodies. KifC3 showed diverse intracellular distribution mitotic cells, especially those in metaphase, showed an enhanced 3018 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 AB

Fig. 3. Clustering of peroxisomes after knockdown of KifC3. COS-7 cells were electroporated twice with siRNA targeting endogenous KifC3 or with non-target control siRNA (sictr), respectively. (A) Knockdown efficiencies on mRNA level were verified using quantitative real-time PCR measurements 24 h after the second electroporation (means and standard deviations are shown). KifC3-mRNA was reduced by 78 ± 0.08% (N = 14) whereas mRNA levels of Kif5B remained unaffected (N = 6). (B) Western blot analysis using α-KifC3 antibodies and α-MAPK antibodies as loading control revealed knockdown levels of KifC3 protein of 86 ± 12% (N = 11), a representative blot is shown. (C) 48 h after the second siRNA treatment cells were fixed and peroxisomes were stained using α-PEX14 antibodies. KifC3 knockdown samples showed increased amounts of cells with clustered peroxisomes compared to control cells whereas knockdown of Kif5B (knockdown Kif5B mRNA: 78.6 ± 3.34%, N = 2) even reduced the amount of cells with clustered peroxisomes. Means and standard deviations (KifC3: N = 4) or spreads (Kif5B: N = 2) are depicted. (D) Staining of peroxisomes and microtubules using α-PEX14 and α-tubulin α antibodies did not reveal any loss of microtubule-associated peroxisomes upon KifC3 knockdown compared to control cells. Scale bars: 10 μm. occurrence of KifC3 at the spindle poles and spindle fibers (Fig. 2A, third 3.3. Overexpression of KifC3 did not affect peroxisomal distribution row), which may argue for a role of KifC3 during cell division. Other C-terminal kinesins like Kar3p in S. cerevisiae or XCTK2 in Xenopus laevis To get insight into the role of KifC3 in peroxisomal transport, KifC3 are also known to be involved in mitosis or meiosis [37–40].KifC3was was overexpressed and cells were analyzed for peroxisomal distribution. more abundant in A549 (carcinomic human alveolar basal epithelial COS-7 cells were transiently transfected with plasmids coding for human cells) and COS-7 (immortalized kidney cells from African green mon- KifC3 and prepared for immunofluorescence microscopy 24 h post key) cells compared to e.g. HepG2 (liver carcinoma derived human transfection. As depicted in Fig. 2C using primary α-KifC3 antibodies epithelial cells) or HeLa (cervix carcinoma derived human epithelial and Alexa Fluor 594 labeled secondary antibodies, overexpressed KifC3 cells) cells (Fig. 2B). This is in accordance with published KifC3 expres- was localized to microtubules (compare also Fig. 1A). This indicates sion profiles showing higher KifC3 levels in kidney and lung derived that KifC3 is able to bind to microtubules and most likely functions as a tissues and lower KifC3 levels in tissues from liver and uterus of microtubule motor protein. Indeed, KifC3 minus-motor activity has human and mice [19,23,24]. However, human embryonic kidney cells been confirmed before in vitro as well as in vivo following transport of (HEK) seemed to express KifC3 at lower protein levels (Fig. 2B). The TGN-derived vesicles [19]. Just like endogenous KifC3, overexpressed used α-KifC3 antibody detected double bands at 70 kDa, as reported KifC3 could not be detected at peroxisomes, which were visualized before [25,28], which were verified to be specific for KifC3 protein by using α-PMP70 antibodies (Fig. 2C, second row). The magnifi cation detecting overexpressed human KifC3 in COS-7 cells. As the endogenous of the overlay (inlay, second row) shows that peroxisomes are often KifC3 levels were highest in A549 and COS-7 cells, these cell lines were located in close proximity to microtubules (in this case marked by used for further experiments. Endogenous KifC3 could not be detected staining patterns of overexpressed KifC3) and are most likely associated at peroxisomes in COS-7 cells marked by α-PMP70 antibodies (Fig. 2A, with them. However, this overlap of peroxisomes with microtubules second row). This could be due to the detection limit of fluorescence seemed not to be enhanced in cells overexpressing KifC3 compared to microscopy as only 5–10% of peroxisomes are actually moving along those harboring endogenous KifC3 levels (not shown). microtubules [4,5] and motor proteins are transiently bound in small Furthermore, overall peroxisomal distribution was not affected by amounts (1–10 molecules per peroxisome [7]). overexpression of either human KifC3 or murine EGFP-tagged KifC3 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 3019

(Table 1). In most cells the peroxisomes were evenly distributed in the Kif5B would then result simultaneously in inactivation of dynein and re- cytoplasm. However, KifC3 overexpressing samples as well as control duced peroxisomal motility explaining the increase of cells with evenly samples transfected with the corresponding empty vectors showed a distributed peroxisomes (Fig. 3C). small portion of cells with perinuclear-clustered peroxisomes of the same magnitude. Peroxisomes are able to respond to alternating 3.5. Microtubules were necessary for peroxisomal clustering metabolic needs and the observed clustering of peroxisomes in some cells could represent secondary stress effects caused by the transfection In order to demonstrate that the observed increase of peroxisomal procedure and the overexpression of protein. clustering after KifC3 knockdown was indeed caused by a microtubule motor and thus was dependent on intact microtubules, the cytoskeleton 3.4. Knockdown of KifC3 led to perinuclear clustering of peroxisomes polymerization was manipulated. SiRNA-treated cells were incubated either with nocodazole to depolymerize microtubules or with cytocha- fi As overexpression of KifC3 did not yield further information about lasin B to disturb micro lament formation as indicated in Fig. 4A. the function of KifC3 on peroxisomes, RNA interference was applied to Controls were incubated with the same volume of DMSO, which was reduce KifC3 protein levels and cells were analyzed for a peroxisomal used as solvent. DMSO-treated samples showed an increased amount distribution phenotype. COS-7 cells were electroporated twice with of cells with perinuclear-clustered peroxisomes after KifC3 knockdown siRNA targeting KifC3-mRNA, with siRNA targeting Kif5B-mRNA as pos- as described before (Fig. 4B). Furthermore, addition of cytochalasin B itive control or with non-target control siRNA. Knockdown efficiencies were controlled on mRNA and protein levels (Fig. 3A,B). Quantitative A real-time PCR revealed reduced KifC3-mRNA levels of 22.3 ± 0.08% (N = 14) compared to control cells whereas Kif5B-mRNA levels remained unaffected by siKifC3 treatment (Fig. 3A). Knockdown of KifC3-mRNA decreased KifC3 protein levels by 86.4 ± 11.99% (N = 11) as exemplified by Western blot analysis of siRNA-treated cell lysates using α-KifC3 antibodies with α-MAPK (mitogen-activated protein kinase) antibodies as loading control (Fig. 3B). To scan for KifC3 knockdown phenotypes, siRNA-treated cells were fixed 48 h after the second electroporation. Peroxisomes were visualized using α-PEX14 antibodies and Alexa 594 labeled secondary antibodies (Fig. 3C). The knockdown of KifC3 increased the number of cells with perinuclear-clustered peroxisomes from 5% to 10% compared B to control cells whereas the number of peroxisomes per cell seemed to remain constant. Microtubules were not affected by KifC3 knockdown and peroxisomes were still associated with microtubules (Fig. 3D). This indicates that microtubule minus-end directed peroxisomal motility is favored upon lowered KifC3 levels, which could be due to increased activity of the microtubule minus-motor dynein. In contrast, siKif5B treatment (knockdown Kif5B-mRNA: 78.6 ± 3.34%, N = 2) resulted in an increased amount of cells with evenly distributed peroxisomes and in less cells with perinuclear-clustered peroxisomes compared to cells treated with control siRNA (Fig. 3C).Hence,peroxisomalminus- end directed motility seemed to be decreased after Kif5B knockdown. Actually, peroxisomal motility phenotypes have been reported after knockdown or inhibition of the main peroxisomal motors kinesin-1 (Kif5) and dynein, respectively [6–8]. In these studies lowered amounts or restricted activity of kinesin-1 or dynein decreased long-range peroxisomal movements in both directions without affecting the even distribution of peroxisomes in the cell. This has been explained by the so called tug-of-war theory [41–43] which states that opposite polarity 8 motors have to be bound simultaneously at the same cargo to create a certain kind of tension that is essential for bidirectional motility. If one 6 of these motors is missing or inactivated, this tension can no longer be 4 applied and motility is completely lost. In our case, knockdown of 2

Table 1 Overexpression of KifC3 did not influence peroxisomal distribution. COS-7 cells were transiently transfected with the plasmids and cells with Fig. 4. Peroxisomal knockdown phenotype depends on microtubules. (A) Microtubules of clustered peroxisomes of one experiment were counted (only apparently COS-7 cells were depolymerized by incubation with nocodazole (noc) whereas addition of transfected cells were considered). At least 500 cells were analyzed. cytochalasin B (cyto B) degraded the actin skeleton. Controls were treated with DMSO, which was used as solvent. Microtubules were visualized using α-tubulin α antibodies Transfected plasmid Amount of cells with and the actin skeleton was labeled by binding to rhodamine tagged phalloidin (ph–rh). clustered peroxisomes (B) SiRNA-treated COS-7 cells were incubated with cytochalasin B for 48 h, while nocodazole was added 4 h prior to fixation. Controls were treated with DMSO for the re- KifC3(m)–pEGFP-C1 3.0% spective times. Peroxisomes were stained using α-PEX14 antibodies. The observed in- pEGFP-N1 2.6% crease of cells with perinuclear-clustered peroxisomes after KifC3 knockdown was KifC3(h)–pLPCX 1.2% independent of the actin skeleton, whereas microtubules were necessary. Cells with pcDNA3.1zeo 1.4% clustered peroxisomes of one experiment were quantified. At least 500 cells were counted. (m): murine; (h): human. sictr: non-target control siRNA, scale bar: 10 μm. 3020 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 directly after the second siRNA-treatment did not influence formation of KifC3 in COS-7 and A549 cells increased the amount of cells with of this peroxisomal phenotype. Cells with clustered peroxisomes were perinuclear peroxisomal clustering (Fig. 5C). Accumulation of both cell still increased upon lowered KifC3 levels compared to control cells. types in S-phase did not influence this increase of cells with clustered Thus, the actin skeleton was not needed for minus-end directed perox- peroxisomes in KifC3 knockdown samples. Consequently, alterations isomal transport towards the centrosome and the observed peroxisomal in the distribution of cells in the particular cell cycle phases did not clustering after KifC3 knockdown was independent of microfilaments. affect formation of the peroxisomal phenotype indicating that the KifC3 In contrast, depolymerization of microtubules in siRNA-treated cells knockdown phenotype is not cell cycle phase dependent. However, the prevented formation of the KifC3 knockdown mediated perinuclear overall amount of cells with clustered peroxisomes was increased in clustering of peroxisomes. In most cells peroxisomes were evenly KifC3 knockdown samples as well as in control samples after thymidine distributed and just a small part of cells was positive for the typical per- treatment. This might be induced by modified metabolic conditions due oxisomal nocodazole phenotype [5,6,44] showing several small clusters to the thymidine incubation. of peroxisomes throughout the cytosol. In KifC3 knockdown samples as well as in control samples the same amounts of cells with the 3.7. KifC3 knockdown also affected morphology of mitochondria and ER nocodazole phenotype were counted, excluding the possibility that these small clusters were generated from earlier KifC3 knockdown- It is known that kinesins and cytoplasmic dynein are able to bind dependent perinuclear clusters. This led to the conclusion that microtu- diverse cargoes and are involved in transport of different cellular organ- bules are necessary to transport peroxisomes to the centrosome where elles. Therefore, the morphology and distribution of different organelles the peroxisomal clusters are formed after KifC3 knockdown. These was examined after KifC3 knockdown. COS-7 cells were treated with results were reproduced several times using different nocodazole siRNA as described before and prepared for immunofluorescence incubation periods (data not shown). microscopy 48 h after the second electroporation. Peroxisomes, early endosomes and the Golgi apparatus were detected using α-PEX14, 3.6. The peroxisomal KifC3 knockdown phenotype was independent of the α-EEA-1 (early endosomal antigen-1) and α-GM130 antibodies, cell cycle phase respectively. Mitochondria were visualized by incubating the cells with MitoTracker Red and the endoplasmic reticulum (ER) was labeled Many motor proteins, especially C-terminal kinesins, play important by transfection with the pDsRed2-ER vector. The KifC3 knockdown not roles in cell division and organization of microtubules. It is likely that only affected the distribution of peroxisomes, but also influenced the KifC3 is also involved in these processes. Hence, we wanted to exclude morphology of mitochondria and the ER (Fig. 6). Mitochondria were the possibility that the observed peroxisomal clustering is caused by clustered near the nucleus to the same extent as peroxisomes. ER struc- accumulation of KifC3 knockdown cells in a certain cell cycle phase tures were also clearly accumulated in the same area though the single e.g. by stabilization of microtubules. COS-7 cells treated with siRNA tubular ER structures were still distributed throughout the cell and the were incubated with propidium iodide 48 h after the second electropo- ER network was intact. Kinesin-1 and dynein not only are responsible ration to label the DNA. These DNA-stained cells were then subjected to for bidirectional transport of peroxisomes but also constitute the princi- flow cytometry and the percentage of cells in the particular cell cycle ple motors for retrograde and anterograde transports of mitochondrial phases was determined (Fig. 5A). KifC3 knockdown cells as well as and ER structures [45–49].Thisarguesforamutualmechanismto control cells showed an equal distribution in different cell cycle phases. develop the KifC3 knockdown phenotype dependent on the minus- The lowered KifC3 levels did not lead to accumulation in a certain cell motor dynein. In contrast, KifC3 knockdown did not influence the distri- cycle phase. To confirm these results COS-7 cells and A549 cells were bution of early endosomes and the Golgi apparatus. The observed over- blocked in S-phase after siRNA treatment by addition of thymidine lay of α-GM130 and α-PEX14 positive structures after KifC3 knockdown and the peroxisomal distribution was analyzed (Fig. 5B). Knockdown (Fig. 6) resulted from the physiological localization of Golgi stacks at the

Fig. 5. Cell cycle phase dependency of peroxisomal clustering. DNA of siRNA-treated A549 and COS-7 cells was stained using propidium iodide 48 h after the second electroporation. A histogram of fluorescence intensities was recorded by flow cytometry to quantify cells in a particular cell cycle phase corresponding to the amount of DNA. (A) Flow cytometry analysis of KifC3 knockdown COS-7 cells and control cells (sictr) did not show any differences concerning the distribution of the cells in the different cell cycle phases. (B) A549 cells were incubated with thymidine (thym) for 24 h to accumulate the cells in S-phase. The quantified relative amounts of cells in the particular cell cycle phases of thymidine-treated and untreated samples are depicted. The used thymidine block resulted in a shift of cell cycle phase distribution from G1- to S-phase and cells in S-phase were doubled. (C) SiRNA-treated A549 and COS-7 cells were incubated with thymidine (as in B) and the amount of cells with clustered peroxisomes was compared to unblocked cells. Means and standard deviations (COS − thym: N = 4) or spreads (COS + thym, A549 ± thym: N = 2) are depicted. At least 500 cells were counted. D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 3021

Fig. 6. Altered morphology of mitochondria and the endoplasmic reticulum after KifC3 knockdown. COS-7 cells were electroporated twice with KifC3-siRNA and with control siRNA (sictr), respectively and prepared for immunofluorescence microscopy 48 h after the second knockdown. Peroxisomes, early endosomes and the Golgi apparatus were stained using α-PEX14, α- EEA-1 and α-GM130 antibodies. To mark mitochondria cells were incubated with MitoTracker Red CMXRos for 30 min and the endoplasmic reticulum (ER) was visualized by transient transfection of the pDsRed2-ER vector 24 h prior fixation. Mitochondria and ER are artificially colored in green. Scale bar: 10 μm. 3022 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 centrosome in COS-7 cells (compare control cells) and the clustering of Actually, increased KifC3 expression levels have been detected in peroxisomes in the same region after reduced KifC3 levels. Xu et al. docetaxel- and paclitaxel-resistant breast cancer cells counteracting showed that KifC3 was involved in positioning of the Golgi apparatus the microtubule stabilizing effect of these cytostatic drugs [28,50] and under cholesterol depleted conditions [25].However,noGolgipheno- suggesting a microtubule destabilizing function of overexpressed KifC3. type was observed in adrenocortical KifC3−/− knockout cells under a KifC3 knockdown could therefore lead to accumulation of cells in a physiological cholesterol level. We describe here perinuclear clustering certain cell cycle phase e.g. by stabilization of microtubules. However, of peroxisomes, mitochondria and ER structures upon KifC3 knock- flow cytometry after staining of DNA with propidium iodide revealed down, indicating enhanced retrograde transport of these organelles, that KifC3 knockdown did not affect the distribution of the cells in the most likely resulting from increased activity of the minus-motor dynein. different cell cycle phases (Fig. 5A). Accumulation of siRNA-treated COS-7 and A549 cells in S-phase by adding thymidine did not influence 4. Discussion the occurrence of the KifC3 knockdown-dependent peroxisomal phenotype. This leads to the conclusion that the observed increase of cells with The uniform distribution of peroxisomes in the cytosol is essential perinuclear-clustered peroxisomes after KifC3 knockdown is formed for peroxisomal ROS (reactive oxygen species) protecting function and independently of the cell cycle phase. Not much is known about distri- transport of peroxisomes to a certain cellular area can be necessary bution of peroxisomes in mammalian cells during cell division. Wiemer under specific metabolic conditions e.g. for peroxisomal lipid metabo- et al. described stochastic disordered inheritance of peroxisomes in lism [14] or to generate inter-peroxisomal contacts [15]. Bidirectional CV-1 cells expressing GFP-PTS1 (peroxisomal targeting sequence-1) in transport of peroxisomes in mammalian and Drosophila cells is mainly contrast to ordered inheritance by association to cytoskeletal elements achieved by the motor proteins kinesin-1 (Kif5) and cytoplasmic dynein [5]. During mitosis peroxisomes were evenly distributed in these cells [6–9]. We show here the involvement of the C-terminal kinesin KifC3 in and mostly not associated with microtubules. The peroxisomal popula- positioning of peroxisomes. In COS-7 cells used in this work KifC3 was tion seemed to be randomly divided between the two developing cells detected at microtubule minus-ends close to the nucleus at the by segregation of the cytoplasm. In contrast, Kredel et al. observed microtubule-organizing center (Fig. 2A). Overexpression of KifC3 led peroxisomal clustering at mitotic spindle poles in HEK 293, HELA and to a prominent localization at microtubules and did not show any NIH3T3 cells after staining peroxisomes by expression of PTS-mRuby alterations regarding the peroxisomal distribution (Fig. 2C). However, suggesting ordered inheritance of peroxisomes [51].However,our knockdown of KifC3 by siRNA resulted in an increase of cells with data indicate that the observed perinuclear clustering of peroxisomes perinuclear-clustered peroxisomes from 5 to 10% (Fig. 3C). Depolymer- after KifC3 knockdown is not correlated to mitosis. ization of the cytoskeleton using nocodazole or cytochalasin B proved the need of microtubules to form the KifC3-dependent peroxisomal 4.2. Is KifC3 involved in binding of peroxisomes to microtubules? phenotype whereas microfilaments were not necessary. Minus-end directed activity of KifC3 has been described in vitro and Binding of peroxisomes to microtubules has been shown in vitro in vivo by demonstrating participation of KifC3 in apical transport of as well as in vivo and binding capacity in vitro was dependent on TGN-derived vesicles in polarized epithelial cells and by tracking of pH, temperature and ATP [4,10,52]. CLIP (cytoplasmic linker protein) EGFP-stained KifC3 in Caco2 cells [19,29]. Epithelial cells harbor apically proteins have been assumed to serve as static linkers as KCl- minus-end anchored microtubules in addition to centrosomal microtu- pretreated peroxisomal eluate fractions reacted with antibodies bules. In these cells KifC3 was localized at microtubule minus-ends on against CLIP-115 [10], a brain-enriched protein responsible for bind- small cholesterol-rich vesicles directly beneath the plasma membrane ing of dendritic lamellar bodies to microtubules [53]. Furthermore, [19]. Expression of a dominant negative, motor domain-lacking KifC3 direct binding of the peroxisomal membrane protein PEX14 to tubu- mutant resulted in loss of this apical KifC3 localization whereas lin has been demonstrated recently and PEX14-deficient human overexpression of intact KifC3 increased the apical appearance of KifC3. fibroblasts showed severe reduction in motility of peroxisomal Assuming KifC3 minus-motor activity at peroxisomes, one would structures [54]. The authors concluded that PEX14 serves as a perox- expect impaired microtubule minus-end directed transport after KifC3 isomal anchor to microtubules and thereby enables peroxisomal knockdown resulting in increased plus-end directed transport and motility. clustering of peroxisomes in the cell periphery. Actually, increased KifC3 did not seem to act as a typical minus-motor on peroxisomal − − Golgi fragmentation has been observed in adrenocortical KifC3 / transport and therefore one can speculate that KifC3 could play a similar knockout mice under cholesterol-depleted conditions, which was role in statically connecting peroxisomes to microtubules. This seems caused by reduced minus-end directed transport. On the other hand, if reasonable as only a small part of peroxisomes performs long-range KifC3 was involved in a tug-of-war, knockdown of KifC3 would not directed movements whereas most peroxisomes are statically linked lead to a peroxisomal distribution phenotype, as also described after to microtubules [4,5]. The observed perinuclear retraction of mitochon- kinesin-1 knockdown (Fig. 3Cand[7,8]), due to simultaneous inactiva- dria and ER after KifC3 knockdown (Fig. 6) also supports the idea of tion of the opposite-directed motor protein. However, we could not KifC3 acting as a static microtubule linker on these organelles. It is observe any of these expected peroxisomal phenotypes after KifC3 tempting to speculate that PEX1 could serve as peroxisomal adaptor knockdown. Treatment of COS-7 and A549 cells with siRNA targeting for KifC3 via the peroxins PEX6 and PEX26. KifC3-mRNA rather seemed to enhance minus-end directed transport We show here interaction of KifC3 with the AAA-ATPase PEX1 in vitro of peroxisomes resulting in perinuclear peroxisomal clustering. Hence, using coimmunoprecipitation (CoIp) (Fig. 1B) as well as in vivo in human one can assume that KifC3 does not act as a typical minus-motor on fibroblasts lacking PEX1. In vivo, KifC3 was not detected at peroxisomes peroxisomal transport. In the following part different possibilities are but transiently expressed PEX1-myc colocalized with overexpressed discussed that may explain the observed peroxisomal phenotype upon KifC3 at microtubules (Fig. 1A). As the CoIp was conducted using COS- KifC3 knockdown. 7 cell lysates overexpressing PEX1-myc and human KifC3, indirect interaction via another protein could be possible. The CoIp conditions needed 4.1. Is KifC3 involved in cell division? to be precisely adapted in order to detect KifC3 in the eluates suggesting weak, transient binding. The fact that KifC3 has been detected in a yeast Increased peroxisomal clustering after KifC3 knockdown could be two-hybrid screen with PEX1 also indicates an interaction of these induced for example by secondary effects resulting from KifC3 functions proteins. during cell division. This idea is supported by the fact that KifC3 was The AAA-ATPases PEX1 and PEX6 are responsible for recycling of the localized at spindle fibers and spindle poles in mitotic cells (Fig. 2A). peroxisomal matrix protein import receptor PEX5 from the peroxisomal D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024 3023 membrane back into the cytosol [55,56]. PEX1 and PEX6 are partially (cis-Golgi network) to microtubules [67], increased the Golgi complex located in the cytosol and are anchored at the peroxisomal membrane in size whereas expression of a dominant negative GMAP-210 mutant by the membrane protein PEX26 [57].PEX1-lackinghumanfibroblasts resulted in Golgi size reduction [68]. It was concluded that GMAP-210 (ΔPEX1) are deficient in peroxisomal matrix protein import and harbor may regulate the size of the Golgi complex by blocking anterograde dysfunctional empty membrane structures, so called “ghosts”.The vesicle transport from the CGN to the medial Golgi after binding of the number and distribution of these “ghosts” are disturbed, often resulting CGN via GMAP-210 to microtubules. In contrast, the microtubule- in clustering at different cellular areas. Nguyen et al. showed reduced associated protein Tau selectively inhibited kinesin-1 motor activity binding of peroxisomes to microtubules in ΔPEX1 cells compared to [69]. Overexpression of Tau in N2a cells reduced plus-end directed normal fibroblasts [44] which would support the idea of PEX1 acting peroxisomal motility and resulted in clustering of peroxisomes at the as a peroxisomal adaptor protein for KifC3. However, we were able to MTOC (microtubule organizing center) [70]. observe long-range directed movements of peroxisomal “ghosts” by YFP-detection after PEX16-YFP expression (data not shown) and also another study did not detect any decrease of peroxisomal motility in 5. Conclusion ΔPEX1 cells compared to normal fibroblasts [54]. In addition, it has fi been demonstrated that expression of PEX11β in ΔPEX1 cells not only We show here for the rst time involvement of the C-terminal increased the numbers of peroxisomal structures by stimulating prolif- kinesin KifC3 in microtubule-based peroxisomal transport. Knockdown eration [58–60], but also resulted in a uniform distribution of these of KifC3 using siRNA increased the amount of cells with perinuclear- “ghosts” in the cytoplasm [44]. These two examples indicate that the clustered peroxisomes. We consider a role of KifC3 as static linker PEX1 KIFC3 binding does not constitute the only connection of peroxi- between peroxisomes and microtubules unlikely and rather favor a somes to microtubules. The expression of PEX16-YFP or PEX11-myc role of KifC3 in minus-end directed peroxisomal transport regulation might trigger different adapters. Furthermore, the yeast two-hybrid by acting as a break for dynein motor protein action. Knockdown of screen using PEX1 as bait identified C-terminal residues of KifC3 KifC3 would then cause enhanced minus-end directed motility of as binding region (data not shown). Motor proteins normally bind to peroxisomes leading to the observed clustering of peroxisomes at cen- microtubules via their motor domain, which is located at the C- trosomes. Furthermore, interaction of PEX1 with KifC3 could constitute terminus in C-kinesins like KifC3, and bind their cargoes at the cargo bind- another regulating element for peroxisomal motility e.g. by competing ing domain, usually located at the N-terminus of the protein, often via with KifC3 binding to microtubules. The results of this study indicate adaptor proteins. Therefore, binding of PEX1 to a C-terminal area of that microtubule-based peroxisomal transport is much more complex KifC3 points more to a regulating function on KifC3. than expected so far and that numerous other factors are involved in these processes besides the motor proteins kinesin-1 and dynein. 4.3. KifC3 acts as a break on dynein-based organelle transport Acknowledgements KifC3 seems to selectively affect peroxisomal minus-end directed motility and therefore might be directly involved in kinesin-1 and We thank Petra Krensel for establishing siRNA technology and George dynein dependent peroxisomal transport. Knockdown of KifC3 resulted Stark, Masatoshi Takeichi, Nobutaka Hirokawa, Herma Portsteffen and in an increase of cells with perinuclear-clustered peroxisomes most Britta Möllers for providing KifC3 and PEX1 plasmids, respectively. probably due to enhanced minus-motor activity at peroxisomes. The Furthermore, we are grateful to Stephen Gould and Hans Probst for used siRNA targeting KifC3-mRNA did not reduce Kif5B-mRNA (Fig. 3A) providing cell lines and antibodies and to Karin Steiger for technical as- and also did not constantly upregulate dynein-mRNA (data not shown). sistance. The authors declare no conflict of interests. As KifC3 did not seem to act as a typical motor protein on peroxisomes, KifC3 may play a regulatory role during peroxisomal minus-end directed transport. Regulation of peroxisomal motility by a GTP-binding protein References via a peroxisome-specific signaling pathway has been reported [61,62] [1] N. Hirokawa, Y. Noda, Y. Tanaka, S. Niwa, Kinesin superfamily motor proteins and and the small GTPase RhoA seems to be involved in association of per- intracellular transport, Nat. Rev. Mol. Cell Biol. 10 (2009) 682–696. oxisomes to microtubules [63]. [2] H. Miki, Y. Okada, N. Hirokawa, Analysis of the kinesin superfamily: insights into – Until now the only detected minus-motor protein acting on peroxi- structure and function, Trends Cell Biol. 15 (2005) 467 476. – [3] M. Schrader, J.K. Burkhardt, E. Baumgart, G. Luers, H. Spring, A. Völkl, H.D. Fahimi, somes is cytoplasmic dynein [6 8]. Based on the observed peroxisomal Interaction of microtubules with peroxisomes. Tubular and spherical peroxisomes KifC3 knockdown phenotype, it is tempting to speculate that binding of in HepG2 cells and their alterations induced by microtubule-active drugs, Eur. J. KifC3 to peroxisomes could constrict dynein function e.g. by simple Cell Biol. 69 (1996) 24–35. [4] S. Rapp, R. Saffrich, M. Anton, U. Jäkle, W. Ansorge, K. Gorgas, W.W. Just, steric hindrance. Existence of such mechanisms to regulate organelle Microtubule-based peroxisome movement, J. Cell Sci. 109 (Pt 4) (1996) 837–849. transport has already been discussed before [64,65]. KifC3 could block [5] E.A. Wiemer, T. Wenzel, T.J. Deerinck, M.H. Ellisman, S. Subramani, Visualization of the pathway in minus-end direction at microtubules and therefore the peroxisomal compartment in living mammalian cells: dynamic behavior and association with microtubules, J. Cell Biol. 136 (1997) 71–80. could function as a break until minus-end directed peroxisomal trans- [6] M. Schrader, S.J. King, T.A. Stroh, T.A. Schroer, Real time imaging reveals a peroxi- port is activated. This process could be regulated by phosphorylation, somal reticulum in living cells, J. Cell Sci. 113 (Pt 20) (2000) 3663–3671. as Western blot analysis of different cell lysates using α-KifC3 antibodies [7] C. Kural, H. Kim, S. Syed, G. Goshima, V.I. Gelfand, P.R. Selvin, Kinesin and dynein move a peroxisome in vivo: a tug-of-war or coordinated movement? Science 308 partly detected an additional band potentially representing phosphory- (2005) 1469–1472. lated KifC3 (Fig. 2B). In fact, phosphoproteom studies using purified [8] S. Ally, A.G. Larson, K. Barlan, S.E. Rice, V.I. Gelfand, Opposite-polarity motors mitotic spindle proteins of HELA cells revealed KifC3 phosphorylation activate one another to trigger cargo transport in live cells, J. Cell Biol. 187 (2009) – [66]. Activation of minus-end directed peroxisomal transport could 1071 1082. [9] L.C. Kapitein, M.A. Schlager, W.A. van der Zwan, P.S. Wulf, N. Keijzer, C.C. lead to release of KifC3 into the cytosol and dynein-mediated minus- Hoogenraad, Probing intracellular motor protein activity using an inducible cargo end directed transport of peroxisomes. If KifC3 levels were reduced trafficking assay, Biophys. J. 99 (2010) 2143–2152. due to siRNA treatment at least this regulation of peroxisomal transport [10] M. Thiemann, M. Schrader, A. Völkl, E. Baumgart, H.D. Fahimi, Interaction of peroxisomes with microtubules. In vitro studies using a novel peroxisome-microtubule would be missing. This would cause enhanced minus-end directed binding assay, Eur. J. Biochem. 267 (2000) 6264–6275. motility of peroxisomes leading to clustering of peroxisomes at the [11] K. Gorgas, Morphogenesis of peroxisomes in lipid synthesizing epithelia, in: H.D. centrosome. Fahimi, H. Sies (Eds.), Peroxisomes in Biology and Medicine, Springer Verlag, Heidelberg, 1987, pp. 3–17. Such selective blocking functions on motor proteins have been [12] A.B. Novikoff, P.M. Novikoff, O.M. Rosen, C.S. Rubin, Organelle relationships in postulated before. Overexpression of GMAP-210, anchoring the CGN cultured 3T3-L1 preadipocytes, J. Cell Biol. 87 (1980) 180–196. 3024 D. Dietrich et al. / Biochimica et Biophysica Acta 1833 (2013) 3013–3024

[13] E.J. Blanchette-Mackie, N.K. Dwyer, T. Barber, R.A. Coxey, T. Takeda, C.M. Rondinone, [42] M.A. Welte, S.P. Gross, M. Postner, S.M. Block, E.F. Wieschaus, Developmental regu- J.L. Theodorakis, A.S. Greenberg, C. Londos, Perilipin is located on the surface layer of lation of vesicle transport in Drosophila embryos: forces and kinetics, Cell 92 (1998) intracellular lipid droplets in adipocytes, J. Lipid Res. 36 (1995) 1211–1226. 547–557. [14] M. Schrader, Tubulo-reticular clusters of peroxisomes in living COS-7 cells: dynamic [43] P.A. Sims, X.S. Xie, Probing dynein and kinesin stepping with mechanical manipula- behavior and association with lipid droplets, J. Histochem. Cytochem. 49 (2001) tion in a living cell, ChemPhysChem 10 (2009) 1511–1516. 1421–1429. [44] T. Nguyen, J. Bjorkman, B.C. Paton, D.I. Crane, Failure of microtubule-mediated per- [15] N.A. Bonekamp, P. Sampaio, F.V. de Abreu, G.H. Luers, M. Schrader, Transient oxisome division and trafficking in disorders with reduced peroxisome abundance, complex interactions of Mammalian peroxisomes without exchange of matrix or J. Cell Sci. 119 (2006) 636–645. membrane marker proteins, Traffic 13 (2012) 960–978. [45] Y. Tanaka, Y. Kanai, Y. Okada, S. Nonaka, S. Takeda, A. Harada, N. Hirokawa, Targeted [16] A.M. Motley, E.H. Hettema, Yeast peroxisomes multiply by growth and division, disruption of mouse conventional kinesin heavy chain, kif5B, results in abnormal J. Cell Biol. 178 (2007) 399–410. perinuclear clustering of mitochondria, Cell 93 (1998) 1147–1158. [17] S.J. Huybrechts, P.P. Van Veldhoven, C. Brees, G.P. Mannaerts, G.V. Los, M. Fransen, [46] A.D. Pilling, D. Horiuchi, C.M. Lively, W.M. Saxton, Kinesin-1 and Dynein are the Peroxisome dynamics in cultured mammalian cells, Traffic 10 (2009) 1722–1733. primary motors for fast transport of mitochondria in Drosophila motor axons, [18] Z. Yang, D.W. Hanlon, J.R. Marszalek, L.S. Goldstein, Identification, partial character- Mol. Biol. Cell 17 (2006) 2057–2068. ization, and genetic mapping of kinesin-like protein genes in mouse, Genomics 45 [47] K.E. Miller, M.P. Sheetz, Axonal mitochondrial transport and potential are correlated, (1997) 123–131. J. Cell Sci. 117 (2004) 2791–2804. [19] Y. Noda, Y. Okada, N. Saito, M. Setou, Y. Xu, Z. Zhang, N. Hirokawa, KIFC3, a microtu- [48] C.M. Waterman-Storer, E.D. Salmon, Endoplasmic reticulum membrane tubules bule minus end-directed motor for the apical transport of annexin XIIIb-associated are distributed by microtubules in living cells using three distinct mechanisms, Triton-insoluble membranes, J. Cell Biol. 155 (2001) 77–88. Curr. Biol. 8 (1998) 798–806. [20] E.H. Hoang, J.L. Whitehead, A.C. Dose, B. Burnside, Cloning of a novel C-terminal [49] M.J. Wozniak, V.J. Allan, Cargo selection by specific kinesin light chain 1 isoforms, kinesin (KIFC3) that maps to human chromosome 16q13–q21 and thus is a candi- EMBO J. 25 (2006) 5457–5468. date gene for Bardet–Biedl syndrome, Genomics 52 (1998) 219–222. [50] M.H. Tan, S. De, G. Bebek, M.S. Orloff, R. Wesolowski, E. Downs-Kelly, G.T. Budd, G.R. [21] E.A. Bruford, R. Riise, P.W. Teague, K. Porter, K.L. Thomson, A.T. Moore, M. Jay, M. Stark, C. Eng, Specific kinesin expression profiles associated with taxane resistance Warburg, A. Schinzel, N. Tommerup, K. Tornqvist, T. Rosenberg, M. Patton, D.C. in basal-like breast cancer, Breast Cancer Res. Treat. 131 (2012) 849–858. Mansfield, A.F. Wright, Linkage mapping in 29 Bardet–Biedl syndrome families [51] S. Kredel, F. Oswald, K. Nienhaus, K. Deuschle, C. Rocker, M. Wolff, R. Heilker, G.U. confirms loci in chromosomal regions 11q13, 15q22.3–q23, and 16q21, Genomics Nienhaus, J. Wiedenmann, mRuby, a bright monomeric red fluorescent protein for 41 (1997) 93–99. labeling of subcellular structures, PLoS One 4 (2009) e4391. [22] J.S. Green, P.S. Parfrey, J.D. Harnett, N.R. Farid, B.C. Cramer, G. Johnson, O. Heath, P.J. [52] M. Schrader, M. Thiemann, H.D. Fahimi, Peroxisomal motility and interaction with McManamon, E. O'Leary, W. Pryse-Phillips, The cardinal manifestations of Bardet– microtubules, Microsc. Res. Tech. 61 (2003) 171–178. Biedl syndrome, a form of Laurence–Moon–Biedl syndrome, N. Engl. J. Med. 321 [53] C.I.DeZeeuw,C.C.Hoogenraad,E.Goedknegt,E.Hertzberg,A.Neubauer,F. (1989) 1002–1009. Grosveld, N. Galjart, CLIP-115, a novel brain-specific cytoplasmic linker pro- [23] E. Hoang, L. Bost-Usinger, B. Burnside, Characterization of a novel C-kinesin (KIFC3) tein, mediates the localization of dendritic lamellar bodies, Neuron 19 (1997) abundantly expressed in vertebrate retina and RPE, Exp. Eye Res. 69 (1999) 57–68. 1187–1199. [24] Z. Yang, C. Xia, E.A. Roberts, K. Bush, S.K. Nigam, L.S. Goldstein, Molecular cloning [54] P. Bharti, W. Schliebs, T. Schievelbusch, A. Neuhaus, C. David, K. Kock, C. Herrmann, and functional analysis of mouse C-terminal kinesin motor KifC3, Mol. Cell. Biol. H.E. Meyer, S. Wiese, B. Warscheid, C. Theiss, R. Erdmann, PEX14 is required for 21 (2001) 765–770. microtubule-based peroxisome motility in human cells, J. Cell Sci. 124 (2011) [25] Y. Xu, S. Takeda, T. Nakata, Y. Noda, Y. Tanaka, N. Hirokawa, Role of KIFC3 motor 1759–1768. protein in Golgi positioning and integration, J. Cell Biol. 158 (2002) 293–303. [55] N. Miyata, Y. Fujiki, Shuttling mechanism of peroxisome targeting signal type 1 [26] S.J. Steinberg, G. Dodt, G.V. Raymond, N.E. Braverman, A.B. Moser, H.W. Moser, receptor Pex5: ATP-independent import and ATP-dependent export, Mol. Cell. Peroxisome biogenesis disorders, Biochim. Biophys. Acta 1763 (2006) 1733 –1748. Biol.25(2005)10822–10832. [27] M.S. Ebberink, P.A. Mooijer, J. Gootjes, J. Koster, R.J. Wanders, H.R. Waterham, Genetic [56] H.W. Platta, S. Grunau, K. Rosenkranz, W. Girzalsky, R. Erdmann, Functional role of classification and mutational spectrum of more than 600 patients with a Zellweger the AAA peroxins in dislocation of the cycling PTS1 receptor back to the cytosol, syndrome spectrum disorder, Hum. Mutat. 32 (2011) 59–69. Nat. Cell Biol. 7 (2005) 817–822. [28] S. De, R. Cipriano, M.W. Jackson, G.R. Stark, Overexpression of kinesins mediates [57] N. Matsumoto, S. Tamura, Y. Fujiki, The pathogenic peroxin Pex26p recruits the Pex1p- docetaxel resistance in breast cancer cells, Cancer Res. 69 (2009) 8035–8042. Pex6p AAA ATPase complexes to peroxisomes, Nat. Cell Biol. 5 (2003) 454–460. [29] W. Meng, Y. Mushika, T. Ichii, M. Takeichi, Anchorage of microtubule minus ends to [58] M. Schrader, B.E. Reuber, J.C. Morrell, G. Jimenez-Sanchez, C. Obie, T.A. Stroh, D. adherens junctions regulates epithelial cell–cell contacts, Cell 135 (2008) 948–959. Valle, T.A. Schroer, S.J. Gould, Expression of PEX11beta mediates peroxisome [30] R.H. Schiestl, R.D. Gietz, High efficiency transformation of intact yeast cells using proliferation in the absence of extracellular stimuli, J. Biol. Chem. 273 (1998) single stranded nucleic acids as a carrier, Curr. Genet. 16 (1989) 339–346. 29607–29614. [31] H. Portsteffen, A. Beyer, E. Becker, C. Epplen, A. Pawlak, W.H. Kunau, G. Dodt, Human [59] C.C. Chang, S. South, D. Warren, J. Jones, A.B. Moser, H.W. Moser, S.J. Gould, Metabolic PEX1 is mutated in complementation group 1 of the peroxisome biogenesis disor- control of peroxisome abundance, J. Cell Sci. 112 (Pt 10) (1999) 1579–1590. ders, Nat. Genet. 17 (1997) 449–452. [60] X. Li, S.J. Gould, PEX11 promotes peroxisome division independently of peroxisome [32] F. Schmidt, D. Dietrich, R. Eylenstein, Y. Groemping, T. Stehle, G. Dodt, The role of metabolism, J. Cell Biol. 156 (2002) 643–651. conserved PEX3 regions in PEX19-binding and peroxisome biogenesis, Traffic13 [61] C. Huber, R. Saffrich, M. Anton, M. Passreiter, W. Ansorge, K. Gorgas, W. Just, A (2012) 1244–1260. heterotrimeric G protein-phospholipase A2 signaling cascade is involved in the [33] G.K. Will, M. Soukupova, X. Hong, K.S. Erdmann, J.A. Kiel, G. Dodt, W.H. Kunau, R. regulation of peroxisomal motility in CHO cells, J. Cell Sci. 110 (Pt 23) (1997) Erdmann, Identification and characterization of the human orthologue of yeast 2955–2968. Pex14p, Mol. Cell. Biol. 19 (1999) 2265–2277. [62] C.M. Huber, R. Saffrich, W. Ansorge, W.W. Just, Receptor-mediated regulation of [34] K.J. Livak, T.D. Schmittgen, Analysis of relative gene expression data using real-time peroxisomal motility in CHO and endothelial cells, EMBO J. 18 (1999) 5476– quantitative PCR and the 2(−delta delta C(T)) method, Methods 25 (2001) 402–408. 5485. [35] U.K. Laemmli, Cleavage of structural proteins during the assembly of the head of [63] L. Schollenberger, T. Gronemeyer, C.M. Huber, D. Lay, S. Wiese, H.E. Meyer, B. bacteriophage T4, Nature 227 (1970) 680–685. Warscheid, R. Saffrich, J. Peranen, K. Gorgas, W.W. Just, RhoA regulates peroxisome [36] S. Tamura, S. Yasutake, N. Matsumoto, Y. Fujiki, Dynamic and functional assembly of association to microtubules and the actin cytoskeleton, PLoS One 5 (2010) e13886. the AAA peroxins, Pex1p and Pex6p, and their membrane receptor Pex26p, J. Biol. [64] S.P. Gross, M. Vershinin, G.T. Shubeita, Cargo transport: two motors are sometimes Chem. 281 (2006) 27693–27704. better than one, Curr. Biol. 17 (2007) R478–R486. [37] V. Mountain, C. Simerly, L. Howard, A. Ando, G. Schatten, D.A. Compton, The kinesin- [65] V.J. Allan, T.A. Schroer, Membrane motors, Curr. Opin. Cell Biol. 11 (1999) 476–482. related protein, HSET, opposes the activity of Eg5 and cross-links microtubules in [66] B. Wang, R. Malik, E.A. Nigg, R. Korner, Evaluation of the low-specificity protease elas- the mammalian mitotic spindle, J. Cell Biol. 147 (1999) 351–366. tase for large-scale phosphoproteome analysis, Anal. Chem. 80 (2008) 9526–9533. [38] M. Hatsumi, S.A. Endow, Mutants of the microtubule motor protein, nonclaret [67] R.M. Rios, A.M. Tassin, C. Celati, C. Antony, M.C. Boissier, J.C. Homberg, M. Bornens, A disjunctional, affect spindle structure and chromosome movement in meiosis and peripheral protein associated with the cis-Golgi network redistributes in the mitosis, J. Cell Sci. 101 (Pt 3) (1992) 547–559. intermediate compartment upon brefeldin A treatment, J. Cell Biol. 125 (1994) [39] W. Saunders, D. Hornack, V. Lengyel, C. Deng, The Saccharomyces cerevisiae 997–1013. kinesin-related motor Kar3p acts at preanaphase spindle poles to limit the number [68] C. Infante, F. Ramos-Morales, C. Fedriani, M. Bornens, R.M. Rios, GMAP-210, a and length of cytoplasmic microtubules, J. Cell Biol. 137 (1997) 417–431. cis-Golgi network-associated protein, is a minus end microtubule-binding protein, [40] C.E. Walczak, S. Verma, T.J. Mitchison, XCTK2: a kinesin-related protein that J. Cell Biol. 145 (1999) 83–98. promotes mitotic spindle assembly in Xenopus laevis egg extracts, J. Cell Biol. 136 [69] M. Vershinin, B.C. Carter, D.S. Razafsky, S.J. King, S.P. Gross, Multiple-motor based (1997) 859–870. transport and its regulation by Tau, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 87–92. [41] G.T. Shubeita, S.L. Tran, J. Xu, M. Vershinin, S. Cermelli, S.L. Cotton, M.A. Welte, S.P. [70] K. Stamer, R. Vogel, E. Thies, E. Mandelkow, E.M. Mandelkow, Tau blocks trafficof Gross, Consequences of motor copy number on the intracellular transport of organelles, neurofilaments, and APP vesicles in neurons and enhances oxidative kinesin-1-driven lipid droplets, Cell 135 (2008) 1098–1107. stress, J. Cell Biol. 156 (2002) 1051–1063.