The armadillo protein p0071 controls KIF3 motor transport

The protein p0071 (also known as plakophilin 4, PKP4), p120-catenin (also known as CTNND1), NPRAP (also known as CTNND2), ARVCF and the more distantly related plakophilins 1–3 (PKP1, PKP2 and PKP3) belong to the p120 family of armadillo repeat proteins (McCrea and Gu, 2010). p120 catenins act as scaffolds in large protein complexes and contain an array of 40–45 amino acid sequence repeats, the so-called armadillo (arm) repeats (Alema and Salvatore, 2007; Keil et al., 2013; Schackmann et al., 2013). In addition, p0071 also harbors a PDZ domain at its C-terminal end (Keil et al., 2013). p0071 has been implicated in various cellular processes, including intercellular adhesion, neurite outgrowth, cytoskeletal organization, cell division (Keil et al., 2007; Nolze et al., 2013; Wolf et al., 2006) and Rab11-dependent endosomal recycling (Keil and Hatzfeld, 2014). p0071 contributes to these cytoskeleton-dependent processes by controlling Rho GTPase signaling. Upon interaction with RhoA and the RhoGEF Ect2, p0071 was shown to foster contractile ring formation/maintenance during mitosis. Both, knockdown or high ectopic expression of p0071 resulted in mitotic dysregulation (Keil et al., 2013; Wolf et al., 2006).

Intracellular transport of p0071 to the midbody or adhesion sites, as well as proper localization of p0071 during cell division, depends on the microtubule cytoskeleton and is mediated by direct interaction with the kinesin protein KIF3B. KIF3B is a motor subunit of the anterograde heterotrimeric KIF3 transport complex, which belongs to the kinesin-2 family and comprises KIF3A, KIF3B and the cargo adaptor protein KAP3 (Keil et al., 2009, 2013; Verhey and Hammond, 2009). KAP3 preferentially binds the C-terminal rod of the KIF3A–KIF3B heterodimer and promotes the efficient binding of membranous cargo vesicles (Deacon et al., 2003; Jimbo et al., 2002; Takeda et al., 2000). In contrast to vesicle-packaged cargo proteins, the interaction of p0071 with KIF3B was shown to be independent of KAP3 (Keil et al., 2009). Additional functions of the p0071–KIF3 complex are as yet elusive.

Apart from its function regulating mitosis, the KIF3 complex is involved in many fundamental biological processes. In non-neuronal mammalian cells, this complex transports membranous cargo vesicles between the endoplasmatic reticulum and the Golgi complex (Stauber et al., 2006). Altered vesicular sorting or transport of specific cargos has been repeatedly linked to various types of cancer and therefore represents a potential target for therapeutic interventions (Patel et al., 2014). The KIF3 complex is also implicated in the transport of dense core granules towards the plasma membrane (Scholey, 2013). Trafficking of dense core vesicles has been previously investigated in BON cells, a cell line established from a human pancreatic neuroendocrine tumor (NET) (Evers et al., 1994; von Wichert et al., 2005). BON cells are of neuroectodermal origin and exhibit hypersecretion of various substances that include bioamines and neuropeptides, such as IGF-I or the secretogranin Chromogranin A (CgA) (von Wichert et al., 2005, 2000). CgA is also an important marker of NETs in vivo. The secretion of CgA can therefore serve as a tool to monitor the molecular mechanisms of protein transport from the trans-Golgi network (TGN) to the plasma membrane.

The involvement of p0071 in processes associated with cytoskeletal organization as well as its direct interaction with KIF3B, prompted us to investigate whether p0071 also contributes to intracellular transport of membranous cargo vesicles.

p0071 is ubiquitously expressed and is primarily detectable at cellular adhesion sites, but also seen in the cytoplasm and the perinuclear area (Hatzfeld, 2005; Hatzfeld and Nachtsheim, 1996; Jakobsen et al., 2013; Wolf et al., 2006). In line with these data, immunofluorescence analysis of endogenous p0071 in BON cells demonstrated localization to adhesion contacts as well as strong cytoplasmic and perinuclear signals (Fig. 1A). There was no specific colocalization with the perinuclear trans-Golgi marker TGN46 (also known as TGOLN2), suggesting that p0071 is not involved in the control of vesicle biogenesis from the trans-Golgi network (TGN) (Fig. 1A).

The kinesin-2 microtubule motor protein KIF3B regulates the intracellular distribution of p0071 (Keil et al., 2009). Indeed, we observed a partial colocalization of p0071 with microtubules in the cytoplasm (Fig. 1A). Since peptide hormone-containing vesicles are transported along microtubules to their release sites (Elias et al., 2012), we examined whether p0071 would colocalize with CgA-positive vesicular structures. Interestingly, we observed a partial colocalization of p0071 with CgA granules in the cytoplasm, which was lost upon RNAi-mediated knockdown of p0071 (Fig. S1A), suggesting a potential role of p0071 in the intracellular transport of CgA-containing vesicles (Fig. 1B).

Thus, we next investigated whether p0071 depletion would impact on CgA intracellular transport in BON cells as measured by time-lapse microscopy with transiently transfected CgA-eGFP (Fig. 2A). Retention of CgA cargo at the TGN was induced at low temperature and transport reinitiated at 37°C. Control cells showed a depletion of the intracellular GFP signal corresponding to the secretion of CgA–eGFP into the supernatant within 25 min. In contrast, upon p0071 knockdown, CgA–eGFP was almost completely retained within the cell at the permissive temperature, suggesting a transport block. The intracellular retention of CgA–eGFP observed upon depletion of p0071 was similar to the phenotype observed following knockdown of protein kinase D2 (PKD2, also known as PRKD2), a Ca²⁺/calmodulin kinase known to regulate release of CgA vesicles from the TGN (von Wichert et al., 2008) (Fig. 2A). The striking effects of p0071 depletion on CgA transport were confirmed by quantitative ELISA from BON culture supernatants (Fig. 2B). Knockdown of p0071 resulted in a significant reduction of CgA in BON supernatants that was again comparable to what was observed upon depletion of PKD2 (von Wichert et al., 2008) (Fig. 2B). To determine the specificity of p0071-mediated effects on CgA vesicle transport in BON cells, we performed rescue experiments by ectopically expressing p0071–HA following depletion of endogenous p0071 using an shRNA directed against the 3′UTR (shp0071 #1). Knockdown of p0071 significantly impaired CgA release, whereas re-expression of p0071 almost completely restored CgA secretion back to the level of controls (Fig. 2C).

p0071 has been implicated in cytoskeletal organization (Hatzfeld and Nachtsheim, 1996; Keil et al., 2013; Wolf et al., 2006) and partially colocalized with microtubules in the cytoplasm of cells (Fig. 1A). Transmission electron microscopy revealed a marked intracellular accumulation of chromaffin vesicles in the cytoplasm of BON cells upon p0071 depletion (Fig. 2D). These data are in agreement with results obtained from our CgA–eGFP experiments (Fig. 2A–C). Moreover, the intracellular transport of CgA vesicles to the plasma membrane along microtubules is affected by microtubule stability (Erickson, 1980; Joseph et al., 2008; Xie et al., 2010). Microtubule-targeting agents impair vesicle transport by affecting microtubule stability, and their long-term application causes microtubule network reorganization (Erickson, 1980; Thuret-Carnahan et al., 1985; Xie et al., 2010). We hypothesized that p0071 might influence CgA vesicle transport by modulating microtubules. Treatment of BON cells with the microtubule-stabilizing drug taxol (Thuret-Carnahan et al., 1985) induced a perinuclear retention of CgA–eGFP as determined by time-lapse microscopy (Fig. S2A). Accordingly, CgA levels in culture supernatants as determined by ELISA assays were significantly reduced by taxol and also by the microtubule-depolymerizing agent vinblastine even at low concentrations (Fig. S2B). Thus, microtubule stability clearly impacts on intracellular cargo transport. Tubulin acetylation has been proposed as a marker for increased tubulin stability (Asthana et al., 2013; Webster and Borisy, 1989). We therefore tested whether p0071 would affect the stability of the microtubule network by assessing tubulin acetylation in the presence and absence of p0071 by using an acetylation-specific antibody. Taxol significantly enhanced microtubule acetylation and thus stability, in line with the literature (Erickson, 1980; Thuret-Carnahan et al., 1985; Xie et al., 2010), whereas knockdown of p0071 did not cause differences in tubulin acetylation (Fig. S2C). This suggests that p0071 does not regulate vesicle transport by modifying tubulin stability. Microtubules are polar structures that originate from microtubule-organizing centers (MTOCs) and constantly undergo a process of disassembly and regrowth. While their inward-bound ends (towards the MTOC) are referred to as minus-ends, the less stable outward oriented ends are termed plus-ends. End-binding protein 3 (EB3, also known as MAPRE3) binds to the plus-end of microtubules and stimulates microtubule growth (Akhmanova and Steinmetz, 2008). It is therefore used as a marker for microtubule polymerization. To investigate whether p0071 depletion would interfere with microtubule polymerization we employed mEOS–EB3 tracking microscopy (Fig. S2D; Movies 1 and 2). Tracking of EB3-positive plus-end comets in HeLa cells indicated no significant differences in polymerization rates upon p0071 depletion, whereas polymerization was significantly inhibited by taxol (Fig. S2D; Movies 1 and 2). Golgi membranes exist in an intimate relationship with microtubules, since microtubule stability and Golgi membranes influence each other (Chabin-Brion et al., 2001). To determine whether p0071 would interfere with microtubule-associated Golgi complex assembly, structure and maintenance, we performed nocodazole washout experiments. HeLa cells were either transduced with p0071-directed or scrambled shRNAs, and subsequently treated with nocodazole to depolymerize microtubules and induce Golgi dispersion. Subsequently, effects were reversed by nocodazole washout. Immunofluorescence analysis of the trans-Golgi marker TGN46 as well as α-tubulin indicated a strong dispersion of the Golgi complex and microtubule network at 2 h after addition of nocodazole (Fig. S2E). Following extensive washout, the Golgi complex and microtubule cytoskeleton reformation was completed after 2 h, regardless of the p0071 expression status (Fig. S2E). In summary, these data suggest that the regulation of cargo transport by p0071 appears to be independent of microtubule organization, polymerization and stability, and of structural changes at the Golgi compartment. Thus, the striking differences in CgA vesicle transport observed upon p0071 depletion are likely not directly dependent on the modulation of microtubule properties.

KIF3B forms a heterotrimeric motor complex together with KIF3A and the cargo adapter KAP3 (Verhey and Hammond, 2009). This complex transports proteins, including large dense core granules (Scholey, 2013), nucleic acids and organelles, towards the plus-ends of microtubules and the plasma membrane. KIF3B also directs intracellular positioning of p0071 (Keil et al., 2009) in a manner that is mediated through the interaction of the p0071-arm repeat domain with the C-terminus of KIF3B (Keil et al., 2009). We confirmed binding of p0071 to KIF3B by co-immunoprecipitation of p0071–HA and KIF3B–Flag proteins in HEK293T cells (Fig. 3A). Furthermore, endogenous KIF3B also colocalized with CgA-positive vesicles in BON cells (Fig. 3B), and this specific colocalization was lost upon shRNA-mediated KIF3B depletion (Fig. S1B). We next examined whether KIF3B was required for the transport of CgA-containing vesicles. Knockdown of KIF3B using two different shRNAs resulted in a marked reduction of CgA release from BON cells, which was comparable to the reduction observed upon p0071 depletion (Fig. 3C). Thus, p0071 and its interaction partner KIF3B, as a component of the KIF3 motor complex, are both involved in the transport of CgA vesicles.

Next, we asked whether p0071 has a general function in cargo transport mediated by the KIF3 complex. Therefore, we assessed whether p0071 would regulate other KIF3 cargos. Intracellular shuttling of matrix metalloprotease 9 (MMP9) is known to depend on KIF3 (Hanania et al., 2012). MMP9 is involved in extracellular matrix degradation as well as tumor angiogenesis, and thereby fulfills key functions in cancer cell invasion and metastasis (Egeblad and Werb, 2002; Wille et al., 2014). To explore a potential role of p0071 in MMP9 transport, we performed MMP9 ELISAs of culture supernatants from Panc1 pancreatic cancer cells (Fig. 4A). MMP9 levels in Panc1 cell supernatants were significantly reduced upon p0071 depletion. Effects were comparable to knockdown of PKD2, which blocks the shedding of MMP9-containing vesicles at the TGN and was used a respective control (Fig. 4A). To test whether p0071 would specifically regulate KIF3-dependent transport and not shuttling by other kinesin motors, we examined transport of vesicular stomatitis virus glycoprotein (VSV-G) vesicles, which is accomplished by kinesin-1 (Cai et al., 2009; Hirokawa et al., 2009; Kreitzer et al., 2000). Knockdown of p0071 did not interfere with VSV-G–eGFP transport in HeLa cells (Fig. 4B,C). In summary, transport of the KIF3 cargos CgA and MMP9 was impaired in different cell models (BON and Panc-1 cells) upon knockdown of p0071 whereas the transport by kinesin-1 was not affected. This suggests that p0071 plays a specific and vital role in anterograde microtubule transport mediated by the KIF3 motor complex.

In order to assess how p0071 might regulate KIF3-mediated CgA transport along microtubules, we tracked the movement of CgA–GFP-containing storage vesicles by time-lapse microscopy in BON cells (Fig. 5A–C; Movies 3 and 4). Both control and p0071-depleted cells exhibited spontaneous movement of vesicles in the cytosol after temperature-mediated cargo release from Golgi membranes (Fig. 5A,B; Fig. S3A). We also did not detect any changes in the overall velocity of CgA vesicles upon p0071 depletion (Fig. S3B). However, tracking of vesicles in the cytosol between the Golgi complex and the plasma membrane indicated significant deficiencies in directional vesicle movement upon p0071 knockdown. The Euclidean distance from the starting point to the end point of each track was significantly reduced by ∼50% in p0071-depleted cells compared to that in controls (Fig. 5A,B), pointing to a substantial loss of directional persistence (Fig. 5C). These effects could be demonstrated using two independent p0071-directed shRNAs (Fig. 5A–C). Again, depletion of the p0071-binding partner KIF3B (Keil et al., 2009) generated similar effects (Fig. 5A–C), suggesting a role of p0071 in the directional movement of CgA carriers transported by the KIF3 motor complex.

Kinesins constantly compete with dynein-mediated transport processes. This so-called ‘tug-of-war’ determines the directionality of vesicle movement (Gross et al., 2002; Soppina et al., 2009). Our data show, that upon p0071 knockdown directional persistence of KIF3-mediated transport is impaired. This results in an imbalance of transport processes so that they proceed at the same speed, but without a stringent direction towards the plasma membrane. At the molecular level, this phenotype may be caused by a change in the affinity of the kinesin motor domains towards microtubules, or by impairing the interaction of the KIF3 motor with cargo vesicles. In order to investigate the underlying molecular mechanisms, we performed quantitative acceptor-photobleach fluorescence resonance energy transfer (AB-FRET) analyses in BON cells to determine the relative proximity (binding) of the endogenous KIF3 complex (labeled with anti-KIF3B primary and Alexa Fluor 488 secondary antibodies) to microtubules stained by α-tubulin–mCherry. Our results indicated that FRET efficiencies for the interaction of KIF3B with microtubules were not significantly different between control and p0071-depleted cells (Fig. S3C), suggesting that p0071 does not modulate the KIF3B–microtubule association. In order to further verify these findings, we additionally performed in vitro microtubule-binding studies with KIF3B–FLAG that was affinity purified from lysates of scrambled shRNA-treated and p0071-depleted HEK293T cells (Fig. S3D–F). Indeed our in vitro experiments demonstrated that the amount of KIF3B–FLAG present in the 100,000 g microtubule pellet fractions was not significantly different for the respective conditions (Fig. S3D–F). This is in line with the literature indicating that p0071 does not directly interact with the motor domain of KIF3B (Keil et al., 2009).

Since p0071 did not modulate the interaction of KIF3B with microtubules, it might regulate the interaction of the KIF3 motor with the cargo vesicles. The heterotrimeric KIF3 complex comprises of the motor units KIF3A and KIF3B as well as the associated protein KAP3 (KIFAP3) (Doodhi et al., 2009; Kondo et al., 1994; Scholey, 2013; Takeda et al., 2000; Yamazaki et al., 1995, 1996). Many functions for KAP3 are still elusive, but it has been suggested to be required for effective cargo binding (Doodhi et al., 2009; Scholey, 2013). KAP3 also interacts with cargo molecules, such as fodrin, APC (Jimbo et al., 2002; Takeda et al., 2000) or dynactin in melanosomes (Deacon et al., 2003). We therefore tested whether depletion of KAP3 would also affect secretion, similar to results obtained upon knockdown of KIF3B and p0071 (Figs 2B and 3C). In line with that data, knockdown of KAP3 also significantly impaired secretion of CgA from BON cells (Fig. 5D). We next explored whether p0071 was involved in the KIF3–KAP3 interaction. First, we examined whether p0071 would bind to KAP3 by co-immunoprecipitation experiments for endogenous proteins from BON lysates (Fig. 5E). Indeed, we were able to show endogenous KAP3 in p0071 immunoprecipitates (Fig. 5E). This interaction was further confirmed by AB-FRET experiments demonstrating a significant increase in the fluorescence donor signals of p0071 stained with p0071 primary and Alexa-Fluor-488-conjugated secondary antibodies (donor molecule) after depletion of the acceptor KAP3–V5–His marked by anti-V5 primary and Alexa-Fluor-568-conjugated secondary antibodies (Figs S3G,I and S4A) at CgA-positive structures (marked with Alexa Fluor 647). In contrast, we could not detect a direct interaction of CgA marked with anti-CgA primary and Alexa Fluor 647 secondary antibodies, packaged in granules, with the cargo adaptor KAP3–V5–His stained with anti-V5 primary and Alexa Fluor 568 secondary antibodies in corresponding FRET experiments (Fig. S3H,I). In order to map the interaction of p0071 with KAP3, we performed GST-pulldown experiments using bacterially purified p0071-head, p0071-armadillo-repeat and p0071-tail domains (Fig. 6A,B). KAP3–V5–His was expressed in HEK293T cells, and was only pulled down from lysates by the p0071 head, but not the p0071 armadillo or tail, domains (Fig. 6B). We confirmed the specificity of our pulldown experiments by investigating the binding of p0071 domains to the KAP3 cargo MMP9–GFP. As cargo stored in vesicles, MMP9–GFP should not directly bind to p0071. In line with this assumption, we did not detect any signal in our pulldown experiments (Fig. S4B). To additionally verify a direct interaction between KAP3 and p0071, we performed in vitro binding experiments using purified bacterially expressed KAP3 and p0071-head–GST proteins (Fig. 6C). Indeed, we were able to show that both proteins directly bind to each other. Subsequently, we analyzed whether depletion of p0071 would affect the interaction of KIF3B and KAP3. Interestingly, knockdown of p0071 with two specific shRNAs significantly impaired the interaction of KIF3B and KAP3 in BON cells. Co-immunoprecipitation experiments demonstrated that, upon p0071 depletion, the amount of KAP3 associated with KIF3B and, vice versa, the portion of KIF3B detected in KAP3 immunoprecipitates was significantly less with respect to cells treated with control shRNA (Figs 6D,E and 7A,B; Fig. S4C–E). These results were confirmed in Panc1 cells (Fig. 7C,D), that were used to study KIF3-mediated MMP9 secretion in Fig. 4B (Hanania et al., 2012; Wille et al., 2014). Again, depletion of p0071 significantly impaired the interaction of KIF3B and KAP3 as indicated by a markedly weaker co-immunoprecipitation of KAP with KIF3B (Fig. 7C,D), suggesting that p0071 functions as a linker to stabilize the KIF3–KAP3 complex. In order to further explore such a stabilization at cargo vesicles, we quantified KIF3B signal intensities at CgA-positive structures in the cytoplasm of BON cells (Fig. 7E,F; Fig. S4F). Indeed, depletion of p0071 resulted in a significant loss of KIF3B signal at CgA-positive vesicles with respect to that of control cells (Fig. 7E,F). In summary, these data indicate that p0071 fosters the interaction of the KIF3B and KAP3 subunits of the KIF3 motor complex thereby regulating directional transport of KIF3 cargos.

In this study, we describe an entirely new regulatory function of the p120-catenin family protein p0071 in intracellular transport of CgA in neuroendocrine tumor cells and of MMP9 cargo in Panc-1 cells. Transport of both cargos was severely impaired upon p0071 knockdown (Figs 2 and 4A). This was surprising, since p0071 had not been previously implicated in cargo transport and the secretory pathway. Our data indicate that the microtubule cytoskeleton has a vital function in p0071-regulated intracellular vesicle transport (Figs 1A and 5, Figs S2 and S1A). However, p0071 did not affect microtubule stability, polymerization or Golgi organization (Figs S2A–E; Movies 1 and 2). Transport of dense core granules and of MMP9 to the plasma membrane depends at least partially on the KIF3 motor complex (Hanania et al., 2012; Hirokawa et al., 2009). Depletion of the KIF3B motor subunit resulted in a similar transport phenotype to that seen upon p0071 knockdown (Figs 3C and 5). p0071 can directly interact with KIF3B (Keil et al., 2009, 2013). Since knockdown of p0071 had no effect on cargo transport to the plasma membrane mediated by other microtubule motors, such as kinesin-1 (Fig. 4B), p0071 seems to be specifically required for the transport of cargo shuttled to the plasma membrane by KIF3 motors.

The heterotrimeric KIF3 motor complex comprises the KIF3A and KIF3B motor subunits as well as the associated protein KAP3, which was suggested to serve as a tightly bound adaptor for cargo binding (Scholey, 2013). KIF3 motors mediate plus-end-directed anterograde transport towards the plasma membrane. Transport in the reverse, minus-end, direction is processed by dynein. Activities of these opposing microtubule motors then produce bidirectional movement of cargo vesicles (De Rossi et al., 2015; Deacon et al., 2003; Fu and Holzbaur, 2013; Gross et al., 2002; Scholey, 2013). Two models have been postulated to explain these processes. In the ‘coordination model’ distinct mechanisms are proposed that switch different sets of motors on or off (Fu and Holzbaur, 2013; Gross et al., 2002). In the ‘tug-of-war’ model a competition between opposed polarity motors is postulated (Blehm and Selvin, 2014; Hendricks et al., 2010; Soppina et al., 2009). Here, the ‘stronger team’ wins the competition and defines the direction of the cargo. It is currently accepted that there is a complex interplay between local tug-of-war interactions on the cargo vesicles and larger regulatory events that bias the outcome of the tug-of-war (De Rossi et al., 2015). Our data fit both models. Analysis of CgA cargo vesicle movement along microtubules revealed that p0071 is required for the directional persistence of vesicle transport, since in the absence of p0071 vesicles moved erratically (Fig. 5A–C; Movies 3 and 4), whereas their overall velocity was unchanged (Fig. S3B). p0071 interacts with the KIF3B motor protein via its armadillo repeats (Keil et al., 2009). We identified a novel interaction of p0071 with KAP3 via the p0071 head domain (Figs 5E and 6A–C). The presence of p0071 stabilized the interaction of the KIF3B motor subunit with the cargo-binding protein KAP3 (Figs 6D,E and 7A–D; Fig. S4C–E). If KIF3 motors are not properly loaded with cargo or the affinity of cargo binding is reduced (Hill et al., 2004; Luby-Phelps et al., 1987), KIF3-transported carriers could easily lose the local tug-of-war and move with dyneins towards microtubule minus-ends. On the other hand, the presence of p0071 might also act as an activating switch that allows for proper activation of the KIF3 motor complex by fostering an interaction within the heterotrimeric complex. In this context, it was recently shown that intermolecular communication within the KIF3A–KIF3B heterodimer modulates entry into the processive run of motors (Albracht et al., 2016). Recent models also suggest that one kinesin, but multiple dyneins, are associated with a cargo vesicle (De Rossi et al., 2015). Interestingly, dyneins are able to ‘step backwards’ into the direction of kinesin motors, whereas kinesins cannot do the same [e.g. during melanosome transport kinesin-2 is suggested to temporarily detach from microtubules but stays connected to the cargo carrier during dynein-favored movement (De Rossi et al., 2015)]. Our finding may be part of an alternate molecular mechanism that might help to control these competing processes. If the interaction between KIF3B and KAP3 is lost, we also lose directional persistence during movement of cargo (Fig. 5A–C). We show that depletion of p0071 results in a weakened KAP3–KIF3B binding (Figs 6D,E and 7A–D; Fig. S4C-E), while the interaction of the KIF3 motor subunit with the microtubules remains intact, as demonstrated by FRET interaction studies and in vitro microtubule binding assays (Fig. S3C–F). In addition, we show that depletion of p0071 also results in impaired KIF3B signals quantified at CgA-positive vesicular structures (Fig. 7E,F; Fig. S4F). Our data therefore suggest a mechanism by which the motor more readily detaches from the cargo vesicle in the absence of p0071, which also might explain the observed loss of directional persistence during plasma membrane-directed transport.

In conclusion, the data presented in this study suggest the following model (Fig. 8). The p120-catenin family protein p0071 binds to the microtubule motor subunit KIF3B with its armadillo repeats and interacts with the associated cargo-binding protein KAP3 via its head domain, thereby stabilizing the interaction of KIF3B and KAP3. If p0071 is lost from the complex, there is a strongly aberrant movement of KIF3–motor cargos, indicated by a significant reduction in directional persistence of CgA vesicles on microtubules transported towards the plasma membrane. This ultimately results in impaired secretion of the KIF3 cargos CgA or MMP9 from BON and Panc1 cells, respectively.

p0071 therefore is a novel and crucial partner in the molecular regulation of transport processes that depend on the KIF3 motor complex. Our data are compatible with, and expand the current models that try to integrate the regulation of plus- and minus-end-directed transport by dyneins and kinesins. This is accomplished through an additional regulatory function of p0071 stabilizing the interaction of the heterotrimeric KIF3 motor complex with cargo vesicles.

Human BON (authenticated in February 2014; Münzberg et al., 2015) and Panc1 (freshly purchased from the European Collection of Cell Cultures through Sigma-Aldrich, St Louis, MO) pancreatic tumor cell lines, and HeLa cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum in a humidified atmosphere of 5% CO₂:95% air at 37°C. Transfection of BON and HEK293T cells was performed using PEI (Polysciences Inc., Warminster, PA). HeLa cells were transfected with HeLa Monster reagent (Mirus Bio, Madison, WI). Panc1 cells were transfected using FuGENE HD (Promega, Madison, WI) according to the manufacturer's instructions. Stable cell lines were generated by lentiviral transduction and selected using 6 µg/ml puromycin. For virus production, HEK293FT cells were transfected with lentiviral expression and packaging constructs as described previously (Sroka et al., 2016). Virus containing supernatants were harvested after 48 h and used for transduction of BON, HeLa and Panc1 cells, respectively.

The pVen2-HA-p0071(wt) and pcDNA3-KIF3B-wt-Flag eukaryote expression plasmids as well as the bacterial expression vector p0071-R(510-993)-pGex5 were described previously (Keil et al., 2009). P71-head(1-509)-pGex5.1 and p0071(tail)-pGex-5X-1 were generated by cloning the p0071 head domain (amino acids 1–509) into pGEX5.1 using the MunI and EcoRI restriction site and the tail domain (amino acids 994–1149) into pGEX-5X-1 using the EcoRI and XhoI restriction sites, respectively. The plasmid pEF-DEST51-KAP3-V5-His was acquired from Source bioscience (Nottingham, UK). For bacterial protein expression of KAP3, the gene sequence was cloned into pGEX-6P3 using EcoRI and Xho1 restriction sites introduced by the following oligonucleotides: 5′-GCTATGAATTCGCAAGGGGAGGACGCCAGATACCTC-3′ as forward primer and 5′-ATTCTCGAGTCAAGATCCATAGCCATAG-3′ as reverse primer, respectively. The pcDNA3-tubulin-mCherry and mEOS-EB3 expression vectors were a kind donation from Prof. Franz Oswald (Ulm University, Ulm, Germany). The CgA–GFP expression construct has been described in von Wichert et al. (2008). p0071- and KIF3B-specific shRNA constructs were purchased from Thermo Scientific (GE Dharmacon, Lafayette, CO). Two p0071 hairpins, one directed against p0071 3′ UTR (TRCN0000123179, referred to in text as shp0071 #1 in experiments) and one directed against p0071 cDNA (TRCN0000123181, referred to as shp0071 #2) were selected for experiments. Two KIF3B-specific shRNAs (TRCN0000116458, referred to as shKIF3B #1 and TRCN0000116460, referred to as shKIF3B #2) were used for the experiments. Constructs used to generate KAP3 knockdown cell lines as follows: clone IDs TRCN0000117237 (referred to as shKAP3 #1) and TRCN0000117238 (referred to as shKAP3 #2). The PKD2 shRNA (NM_016457.x-294s1c1]) was obtained from Sigma-Aldrich, St Louis, MO.

Unless otherwise indicated, cells were serum starved and treated with 10 nM paclitaxel (taxol) (Sigma-Aldrich) or 5 nM vinblastine (Sigma-Aldrich), respectively. CgA secretion assays were carried out after 16 h of incubation. Tubulin acetylation was assessed by western blotting from total cell lysates after incubation times of up to 72 h. For nocodazole treatment, cells were incubated with 2.5 µg/ml nocodazole (Sigma-Aldrich) for 2 h at 37°C. Washout was achieved by rinsing five times and transferring to fresh medium for 2 h before fixation.

BON and HeLa cells were seeded on glass coverslips, and after 24 h (unless indicated otherwise) fixed and permeabilized with cold methanol for 5 min. Samples were blocked and stained in PBS containing 5% fetal bovine serum (FBS) and 0.05% Tween-20. Primary and secondary Alexa Fluor-conjugated (Invitrogen, Carlsbad, CA, USA) antibodies were added for 2 h at room temperature followed by washing samples with PBS five times for 5 min. Coverslips were mounted using Fluoromount-G (Southern Biotech, Birmingham, AL). Primary antibodies used for immunofluorescence staining include antibodies against p0071, KIF3B, KAP3, α-tubulin, TGN46 and anti-chromogranin A. A full list of primary and secondary antibodies used for immunofluorescence including dilution, catalog numbers, vendors and literature references is available in Table S1. Signal intensities were measured by quantifying the mean of the region of interest (ROI) intensities within a defined sub-ROI using the ImageJ software (Schneider et al., 2012), and statistical analysis was performed using Prism 6.0 (GraphPad, San Diego, CA). AB-FRET experiments were performed and quantified as described previously (Eiseler et al., 2016; Sroka et al., 2016; Wille et al., 2014). In brief, pre-bleach and post-bleach images of donor and acceptor were acquired in an automated time series. Acceptor bleaching was performed using an intensive 561-nm laser line in a defined region of interest as representatively shown in Fig. S3. Single percentage FRET values, as well as mean FRET efficiency and s.e.m. of non-thresholded raw data were calculated, and statistical significance was determined by using a two-tailed unpaired Student's t-test.

For live-cell imaging, cells were seeded onto 35 mm microscopy dishes (ibidi GmbH, Martinsried, Germany) 24 h prior to treatment. For nocodazole washout experiments and microtubule end point tracking, an inverted Olympus IX71 microscope (Olympus, Hamburg, Germany) equipped with an ORCA R2 camera (Hamamatsu Orca, Hamamatsu, Japan) and Simple PCI Software (Compix, Irvine, CA) was utilized. For all other experiments (including live-cell imaging and vesicle tracking), samples were processed using a confocal laser scanning microscope TCS SP8-HCS (Leica, Mannheim, Germany) equipped with a Plan APO CS2 63×/1.3 NA water immersion objective. Images were acquired with a 90% open pinhole to compensate for thermal drift and active hardware autofocus. Cells were imaged in a K-Frame OKO-LABS stage-top incubator with 5% CO₂ at 37°C.

BON cells treated with scrambled shRNA or p0071 shRNA were seeded on glow discharged, carbon-coated sapphire discs (3 mm in diameter, 160 µm thick, Engineering Office M. Wohlwend GmbH, Sennwald, Switzerland) for 24 h (37°C, 5% CO₂). Cells were then high-pressure frozen using a Wohlwend HPF Compact 01 high-pressure freezer (Engineering Office M. Wohlwend GmbH) as described previously (Buser and Walther, 2008). Subsequently, samples were freeze-substituted by replacing the ice in the samples with substitution medium containing glutaraldehyde (3%), uranyl acetate (0.1%) and water (1.2%) in acetone. During freeze substitution, the temperature was slowly increased from 183 K to 273 K over a time period of 18 h. Afterwards the samples were kept at room temperature for 30 min before they were washed twice with acetone and embedded in epon (72 h, polymerization at 333 K), before sections were cut with an ultramicrotome (Leica Ultracut UCT ultramicrotome) using a diamond knife (Diatome, Biel, Switzerland). Imaging was performed with a JEOL JEM2100F.

For western blot analysis, cells were lysed in 10 mM Tris-HCl, 5 mM EDTA, 50 mM NaCl, 50 mM NaF and 1% Triton X-100 supplemented with Complete Protease inhibitor Cocktail and PhosStop tablets (Roche, Basel, Switzerland) and, when indicated, subsequently subjected to immunoprecipitation using 2 µg of antibody and Protein-A–Sepharose (GE Healthcare, Chicago, IL). For immunoblotting, proteins were transferred to PVDF (Millipore, Billerica, MA) and nitrocellulose (GE Healthcare) membranes, respectively, after SDS-PAGE. Membranes were blocked with 5% non-fat dry milk in phosphate-buffered saline (PBS) containing 0.2% Tween 20 and incubated overnight at 4°C with specific antibodies. Antibodies used for immunoprecipitation and western blot detection include antibodies against p0071, KIF3B, PKD2, KAP3, Flag-M2, α-tubulin, acetylated tubulin, V5 epitope and β-actin. A full list of primary antibodies including dilution, catalog numbers vendors and literature references is available in Table S1. After incubation with primary antibody, membranes were extensively washed, incubated with secondary horseradish peroxidase (HRP)-conjugated antibodies (GE Healthcare) and developed using enhanced chemoluminescence substrate (Super Signal West^® extended duration substrate, Thermo Scientific, Carlsbad, CA). Where indicated, western blot band intensities were quantified by using ImageJ software (Schneider et al., 2012).

p0071 deletion constructs [p0071-R(510-993)-pGex5.1 (armadillo repeat domain), p0071-head(1-509)-pGex5.1 (N-terminal head domain), p0071(tail)-pGex5.1 (C-terminal tail domain)] and KAP3-pGEX-6P3 were expressed in BL21DE3 bacteria (New England Biolabs) followed by induction at A₆₀₀=0.6 with 1 mM isopropyl 1-thio-β-D-galactopyranoside (Thermo Scientific, Waltham, MA) for 5 h at room temperature. After lysis of bacteria, proteins were purified using glutathione–Sepharose 4B™ (GE Healthcare). For pulldown of p0071-domain-binding proteins from total cell lysate, 20 µl of beads were incubated with 1.5 mg of cell lysate from HEK293T cells transfected with overexpression constructs and empty vector, respectively. GST-loaded beads were used as negative controls. After several washing steps, beads were boiled in SDS sample buffer at 95°C and eluates subjected to SDS-PAGE and western blotting. Nitrocellulose membranes were stained for protein by incubating in Ponceau S solution (AppliChem, Darmstadt, Germany), for 5 min and destained with double-distilled H₂O. For in vitro binding studies of isolated proteins in a completely cell-free system, KAP3–GST was bacterially expressed in induced BL21 cells and, after bacterial lysis and purification using glutathione–Sepharose 4B™ (GE Healthcare), the purified protein was eluted from the beads using 20 mM reduced glutathione in 50 mM Tris-HCl pH 8.0. Eluates were concentrated and re-buffered to 25 mM Tris-HCl pH 8.0 using Vivaspin2 concentrators (10,000 MWCO, Satorius Stedim Biotech). The GST-tag was removed by adding 1.5 units of HRV3C protease (MoBiTec, Goettingen, Germany) per 100 µg of protein for 8 h. The protein was then incubated with glutathione beads three times for 1 hour to remove the cut tag and residual uncut GST fusion protein as well as the protease. Proper removal of the GST tag from the fusion protein was verified by SDS-PAGE and Coomassie staining using PlusOne Coomassie tablets (PhastGel Blue R-350, GE Healthcare) according to the manufacturer's instructions. p0071-head domain protein, KAP3 and GST alone were bound to glutathione–Sepharose 4B™ beads (GE Healthcare) and eluted using L-glutathione (Sigma-Aldrich). Eluates were concentrated and re-buffered to PBS using Vivaspin2 concentrators (10,000 MWCO, Satorius Stedim Biotech). Purified proteins were then mixed as indicated, incubated overnight with specific antibody against KAP3 and pulled down with protein A–Sepharose beads followed by western blot analysis.

HeLa cells were stably transduced to express control shRNA and p0071 shRNA, respectively. Cells were transfected with a construct encoding a temperature-sensitive eGFP-tagged variant of VSV-Glycoprotein (Pusapati et al., 2010) and cultured at 39.5°C overnight. Cells were then incubated at 20°C for 2 h and returned to the permissive temperature (32°C) in the presence of 20 µM cycloheximide for live-cell imaging (one frame every 4 min 45 s). Fluorescence intensities over time were quantified in 10 cells per condition by measuring eGFP signals in a defined region of interest drawn around the visibly accumulated protein in the trans-Golgi area using the ImageJ software (Schneider et al., 2012).

Stably transduced BON cells were seeded into six-well plates and starved of fetal calf serum overnight, before fresh culture medium was added to monitor secretion for 4 h. Supernatants were subjected to an ELISA using NUNC MaxiSorp™ high protein-binding capacity polystyrene 96 well ELISA plates and anti-CgA mouse monoclonal antibody (DAKO, St Clara, CA). Absorption values were determined using a microplate reader (Tecan Infinite M200 Pro, Tecan, Maennedorf, Switzerland) measuring excitation at 450 nm.

Panc1 cells transduced with shRNA-expressing constructs were seeded in six-well dishes and growth medium was replaced with serum-free medium the following day. After 48 h of secretion, supernatants were collected and subjected to ELISA using a commercial kit (MMP9 Human ELISA Kit; Life Technologies, Carlsbad, CA) according to the manufacturer's instructions.

BON cells expressing the respective shRNAs were transiently transfected with CgA–eGFP and cultivated in serum free medium at 18°C overnight to induce cargo retention at the TGN. Secretion was re-initiated by further cultivating the cells at 37°C in a live microscopy chamber. Images were acquired at time points t=0, 5, 10, 15, 20 and 25 min. Subsequently, intensities were analyzed by defining a rectangular region of interest (ROI) including the entire cell. Mean of ROI intensities were quantified with NIH ImageJ software (Schneider et al., 2012).

To track intracellular vesicle movements, BON (or Panc1) cells expressing control shRNA or shRNA directed against p0071 and KIF3B, respectively, were transiently transfected with a construct encoding CgA–eGFP and treated with 100 nM SiR–tubulin (Spirochrome, Stein am Rhein, Germany) to stain microtubules. Cargo retention was induced at 18°C, after which cells were subjected to live imaging at 37°C (one frame every 1.29265 s) and vesicles were tracked in the cytoplasmic area between Golgi membranes and the plasma membrane by using the ImageJ manual Tracking tool (Schindelin et al., 2015). Vesicles were tracked for up to 100 frames in registered image series, dependent on the time each tracked vesicle was continuously moving within the focused layer (and thus observable). Tracking results were analyzed with help of the ibidi chemotaxis and migration tool v2.0 to calculate and visualize vesicle directional persistence.

End-binding-protein mEOS–EB3 was transiently expressed in HeLa cells to visualize microtubule polymerization at plus-ends. Video images were acquired at one frame per second. Fifty particles per condition were tracked by using the ImageJ manual tracking tool (Schindelin et al., 2015) over a time span of 5 s and velocities were quantified.

In vitro binding studies with purified microtubules were performed using the microtubule spin-down kit (#BK029, Cytoskeleton Inc., Denver, CO) according to the manufacturer's instructions. KIF3B–Flag protein for assays was affinity purified from lysate of 25 10-cm dishes of transfected HEK293 cells using FlagM2 Sepharose (Sigma-Aldrich) after previous treatment of cells with control shRNA and p0071-directed shRNA, respectively. Protein was eluted from beads by pH shift (glycine buffer pH 2.6) and concentrated as well as re-buffered in HEPES with Vivaspin2 concentrators (10,000 MWCO, Satorius Stedim Biotech). We used 2 µg of KIF3B–Flag protein per sample for microtubule-binding assays. Tubulin (100 µg) with added GTP was allowed to form polymers of ∼10 µm in length (concentration 5×10¹¹ microtubules/ml) and stabilized with taxol according to the manufacturer’s instructions. After incubation with KIF3B–Flag from p0071-depleted, control cells or microtubule-binding control proteins (BSA and MAP2, respectively), samples were loaded onto cushion buffer (80 mM PIPES, 1 mM MgCl₂, 1 mM EGTA, 60% glycerol) and subjected to a 100,000 g ultracentrifugation. Fractions were then analyzed by western blotting and quantified as stated in the respective figure legend.

Statistical analysis was performed using Prism software, version 6.00, for Windows (GraphPad, San Diego, CA). Graphs depict mean±s.e.m. for all conditions. Statistical significance: ns, not significant, *P=0.05–0.01, **P=0.01–0.001, ***P<0.001, ****P<0.0001.

The authors want to acknowledge use of the Imaging Core Facility for ‘Multiphoton and Confocal Microscopy’ at Ulm University. We thank Franz Oswald (Ulm University) for providing valuable research materials and Ninel Azoitei for critically reading the manuscript.