The phosphatase of regenerating liver 3 (PRL-3) promotes cell migration through Arf-activity-dependent stimulation of integrin α5 recycling

Protein tyrosine phosphatases (PTPs) are key regulatory enzymes in various signal transduction pathways. Defective or inappropriate regulation of PTP activity contributes to the development of many human diseases, including cancer. The PRL phosphatases represent a subfamily of PTPs, which is comprised of three members (PRL-1, -2 and -3) sharing a high degree (>75%) of homology. PRL-3 was found overexpressed in metastases of colorectal carcinoma, whereas its expression in primary tumors and normal epithelium was undetectable (Saha et al., 2001). Elevated PRL-3 expression was also detected in gastric, liver, ovarian and breast cancer and correlated with progression (Bessette et al., 2008). Moreover, cells stably transfected with PRL-3 exhibited enhanced motility and invasive activity (Zeng et al., 2003). Recent findings indicate that PRL-3 may function in regulating cell adhesion structures to affect epithelial–mesenchymal transition. More recently it was shown that expression of PRL-3 and Akt activity is mediated by the RNA-binding protein PCBP1 (Wang et al., 2010).

Although considerable evidence has accumulated suggesting that PRL-3 may play a major role in tumorigenesis and metastasis, little is known about the underlying mechanism(s). Enhanced migration is a fundamental characteristic of tumor cells. As cells migrate, intracellular signaling cascades are activated to promote remodeling of the actin cytoskeleton, to form membrane protrusions and to transport proteins to the areas where they are needed. Small GTPases belonging to the Rho and ADP-ribosylation factor (Arf) families have been characterized as key players regulating these processes (D'Souza-Schorey and Chavrier, 2006). Based on their sequence homology, human Arf proteins are classified into three classes: class I (Arf1 and Arf3), class II (Arf4 and Arf5) and class III (Arf6). Arf-GTPases cycle between an inactive GDP-bound form and an active GTP-bound form through the action of guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs). The large number of these regulatory proteins compared with the relatively small number of Arfs suggests that activation of these GTPases is under extensive regulatory control. Arf1, as a class I Arf, is classically associated with the Golgi apparatus where it regulates vesicle trafficking and interacts with a number of proteins including Arf GEFs and GAPs, enzymes involved in lipid metabolism and protein adaptors that aid in the recruitment of specific cargos (D'Souza-Schorey and Chavrier, 2006). However, a role for Arf1 in the control of integrin-mediated cell adhesion was indicated by studies on Brefeldin A (BFA)-resistant Arf GEFs and it was shown that Arf1 mediates paxillin recruitment to the focal adhesions and potentiates Rho-stimulated stress fiber formation in fibroblasts (Norman et al., 1998). Furthermore, inhibition of Arf1 activation by targeting ARNO, a BFA-resistant Arf 1-3/Arf6 GEF, impaired ARNO-dependent migration of MDCK cells (Viaud et al., 2007). Recently, Arf1 was detected in membrane ruffles of invasive breast cancer cells and it was shown that Arf1 controls epidermal growth factor (EGF)-dependent growth and migration of breast cancer cells (Boulay et al., 2008).

Here we show that PRL-3 interacts with the small GTPase Arf1, in an Arf-activity-dependent fashion. In addition, PRL-3 and Arf1 colocalize on intracellular membranes and PRL-3 induces a BFA-sensitive activation of Arf1. Furthermore, the promigratory effect of PRL-3 depends on class-I Arf expression and activation, as constitutive activation of Arf1 leads to a significant increase in cell velocity and stimulates the trafficking of membrane receptors such as integrins. Our data suggest that PRL-3 exerts a promigratory role through activation of Arf1, inducing faster trafficking of integrins.

PRL-3-dependent induction of migration has been described as the hallmark of PRL-3 activation. Therefore, we performed cell migration assays to verify the proposed phosphatase- and prenylation-dependent promigratory role of PRL-3 using HeLa cells and a stable cell line with tetracycline-inducible PRL-3 expression derived from HeLa cells (T-REx-HeLa-PRL-3 cells; supplementary material Fig. S1, Table S1). Given that only few potential PRL-3 interaction partners have been reported and that direct substrates are still unknown, we used a bacterial two-hybrid assay to identify potential PRL-3 interaction partners. A number of putative PRL-3-interacting proteins were identified, including the ADP-ribosylation factor 5 and the coatomer subunit zeta-1 (supplementary material Table S2). Both proteins share localization and function by being constituents of membrane trafficking pathways. Putative interactions were further validated by GST pull-down, using GST-fused Arf5 and zeta-COP proteins (Fig. 1A). Further GST pull-down assays revealed binding of GST-tagged PRL-3 to endogenous Arf1 (Fig. 1B). Supporting this, GST-tagged Arf1 was binding to PRL-3 from Tet-induced T-REx-HeLa-PRL-3 cells (Fig. 1C). In addition, using a GFP-tagged version of PRL-3 interaction with endogenous Arf1 was confirmed by co-immunoprecipitation assays (Fig. 1D). Much weaker binding of GST–PRL-3 to Arf6 was observed when compared to Arf1 (Fig. 1E) and no co-immunoprecipitation of GFP–PRL-3 with endogenous Arf6 was detectable (data not shown), suggesting primary interaction with class I and class II Arfs. Since the data available for class I Arfs are much more advantageous we decided to focus on the interaction of PRL-3 with Arf1.

Subcellular localization of PRL-3 and its association with Arf1 in HeLa cells were further defined by co-transfection experiments with EGFP–PRL-3 and Arf1–mRuby constructs. Fluorescence microscopy revealed that PRL-3 and Arf1 finely colocalize in the perinuclear (Golgi) region, as well as in some vesicular structures outside of the Golgi (Fig. 1F); for quantitative analysis see supplementary material Fig. S2. In good agreement with previous reports some of these PRL-3- and Arf1-positive structures were identified as early endosomes (supplementary material Fig. S2). Upon incubation of the cells with transferrin receptor (TfR) antibodies, EGFP–PRL-3 and Arf1–mRuby colocalized with internalized TfR in perinuclear vesicular compartments, representing recycling endosomes (Fig. 1G); for quantitative analysis see supplementary material Fig. S2. These data provide first evidence that PRL-3 interacts with members of Arf GTPase family. Moreover, PRL-3 colocalizes with Arf1 in the Golgi region, and in a perinuclear compartment consisting of early and recycling endosomes.

GTP-bound, active Arf1 has an increased membrane affinity and recruits various coat and adaptor proteins, regulating the secretory membrane transport at the Golgi. We performed GST pull-down assays using GST–Arf1 pre-loaded with either guanosine 5′-O-(γ-thio)triphosphate (GTPγS), a nonhydrolyzable GTP analogue or guanosine diphosphate (GDP), thus activating or inactivating the Arf1 GTPase, respectively. GTPγS-loaded GST-Arf1 bound significantly more PRL-3 protein, compared to untreated or GDP-treated GST–Arf1 (Fig. 2A). Furthermore, when EGFP–PRL-3 was co-expressed with the constitutively active Arf1(Q71L) mutant, PRL-3 was more concentrated at the Golgi (Fig. 2B, middle) compared to wild-type (wt) Arf1 or the dominant-negative Arf1(T31N) (Fig. 2B). These data are in good agreement with observations showing alterations of affinities for target membranes of activated Arfs and Arf-binding proteins (Szul et al., 2005). Quantification of fluorescence intensities revealed that Arf1(Q71L) expression causes a significant increase in PRL-3 accumulation in the perinuclear region; for quantitative analysis see supplementary material Fig. S2. Accordingly, the ratio of perinuclear and whole cell area of PRL-3 fluorescence is significantly higher in cells co-expressing PRL-3 and Arf1(Q71L), compared to cells transfected with wt Arf1 or Arf1(T31N) (Fig. 2C). Thus, PRL-3 exhibits a higher affinity toward an activated Arf1 in in vitro binding assays, as well as in colocalization assays where PRL-3 seems to be retained in the perinuclear region upon expression of constitutively active Arf1.

The functional consequences of PRL-3 expression and Arf expression and/or activity were analyzed by siRNA-mediated knock-down of Arf1 and Arf3 (supplementary material Fig. S3). Due to the high degree of homology and compensatory functions between class I Arfs (Volpicelli-Daley et al., 2005) the single isoform-selective (Arf1 or Arf3) knockdown did not abolish the promigratory effect of PRL-3 in T-REx-HeLa-PRL-3 cells. However, double knockdown of Arf1 and Arf3 completely abolished the promigratory effect of PRL-3 expression in these cells (Fig. 2D; supplementary material Table S3). To further determine how Arf1 activity affects cell migration in conjunction with PRL-3 expression, cells were transfected with dominant Arf1 mutants and/or treated with Brefeldin A (BFA), a potent inhibitor of class I Arfs. Expression of the dominant-negative Arf1(T31N) and class-I Arf inhibition through BFA treatment completely abrogated the promigratory effect of PRL-3. In contrast, the constitutively active Arf1(Q71L) yielded a significant increase in cell velocity, and co-expression with PRL-3 did not result in a statistically significant further increase (Fig. 2E; supplementary material Tables S4, S5). The effects of Arf1 mutants on cell migration were also tested using parental HeLa cells. The results were consistent with the data obtained with T-REx-HeLa-PRL-3 cells (Fig. 2F; supplementary material Table S6). Thus, the PRL-3-mediated promigratory effects require expression and activation of Arf1.

Our data suggest that subcellular localization of PRL-3 is critical for its function. PRL-3 was scanned for relevant functional motifs using the Minimotif Miner (Balla et al., 2006). We found that PRL-3 contains a MXXE motif, which has been described as a Golgi-targeting motif within Arf1 (Honda et al., 2005). In addition, this motif is present in other members of the PRL family and is evolutionally conserved in PRL-3 (Fig. 3A). Therefore, a PRL-3 mutant with a MXXE-deletion was generated [PRL-3(ΔMXXE)]. In pull-down assays, Arf1-binding to GST–PRL-3(ΔMXXE) was diminished (Fig. 3B), implying a significant contribution to the interaction. In addition, PRL-3(ΔMXXE) was absent from the Golgi and the cell membrane, and the colocalization with TfR and Arf1 was lost (Fig. 4C). However, the prenylation defective PRL-3 mutant [PRL-3(ΔCAAX)] and the phosphatase dead PRL-3 [PRL-3(C104S)] were able to bind Arf1 in vitro (Fig. 3B; supplementary material Fig. S4) suggesting that phosphatase activity, interaction of PRL-3 with Arf1 and subcellular localization might be crucial for PRL-3 function. In good agreement, expression of the PRL-3 mutants did not increase cell velocity in migration assays (Fig. 3D; supplementary material Table S1).

Next we asked whether PRL-3 affects Arf1 activity. The GAT domains of the clathrin adaptors GGA1, GGA2 and GGA3 were shown to interact with GTP-bound Arf (D'Souza-Schorey and Chavrier, 2006). Indeed, in pull-down assays GST–GGA1-GAT preferentially bound Arf1 from the GTPγS-loaded cell lysates (supplementary material Fig. S5). Using GGA1–GAT as bait, we determined the level of activated, GTP-bound Arf1 in T-REx-HeLa-PRL-3 cells or transfected HeLa cells. In T-REx-HeLa-PRL-3 cells the induction of PRL-3 expression led to a significant increase in GGA1-bound Arf1–GTP (Fig. 4A). Transient expression of wild-type PRL-3 in HeLa cells also resulted in an increase in Arf1 activity compared to controls: PRL-3 mutants [PRL-3(ΔCAAX), PRL-3(C104S) and PRL-3(ΔMXXE)] did not promote Arf1 activation. BFA treatment caused a significant drop in Arf1 activity and abolished the PRL-3-mediated increase of Arf activity (Fig. 4B,C). In a control experiment, BFA treatment expectedly resulted in dissociation of Golgi markers and Arf1–mRuby from the Golgi (Fig. 4D). Interestingly, after BFA treatment, EGFP–PRL-3 remained at least in part in a perinuclear compartment suggesting additional binding sites other than Arf1 (supplementary material Fig. S6). Taken together, PRL-3 promoted activation of class I Arfs in a phosphatase activity and prenylation-dependent manner. Moreover, the effect was BFA sensitive and required the interaction with Arf1.

There is increasing evidence that the membrane trafficking pathways that recycle integrins are essential for cell migration, and that perturbed integrin signaling and trafficking contributes to cell invasion and migration during cancer metastasis (Caswell et al., 2008). Here we performed recycling assays to determine whether PRL-3 mediates its promigratory effect by influencing the recycling of membrane receptors through Arf1. After internalization, fluorescently labeled transferrin (Tf-546) and antibody–α5-integrin complexes displayed a high level of colocalization in perinuclear vesicular structures, presumably representing recycling endosomes (Fig. 5A). Interestingly, the gradual decrease in the amount of perinuclear Tf-546 and antibody–α5 integrin complexes in these assays was absent in BFA-treated cells suggesting dependence on BFA-sensitive ARF-GEFs (Fig. 5B).

Quantification of α5 integrin and TfR recycling was performed by a flow-cytometry-based assay. While the overall surface expression level of α5 integrins was not changed after PRL-3 expression (supplementary material Fig. S7), the levels of recycled TfR and α5 integrin complexes in cells expressing PRL-3 were significantly higher than in mock-transfected cells (Fig. 5C,E). The mean fluorescence intensities from repeated experiments were averaged and presented relative to control conditions (mock). As shown in Fig. 5D,F, wild-type PRL-3-expressing cells displayed a significant increase (∼50%) in the level of recycled TfR and α5 integrin, which was completely abolished by BFA treatment and required phosphatase activity (supplementary material Tables S7, S8).

Next, we studied the trafficking of integrins in greater detail by fluorescence uptake after photoconversion (FUAP) microscopy. To follow the trafficking of α5 integrins, we fused α5-integrin cDNA to a photoconvertible protein, EosFP (Wiedenmann et al., 2004). Upon transfection in HeLa cells, α5–EosFP localized to focal adhesions together with endogenous paxillin (supplementary material Fig. S7). Spreading and migration of α5–EosFP transfected cells was unchanged in control cells (data not shown) and subcellular localization of α5 integrins showed no obvious changes (supplementary material Fig. S8).

After conversion of α5–EosFP with a short pulse of UV light, uptake of the photoconverted α5–EosFP (red) in unconverted focal adhesions was followed by time-lapse fluorescence microscopy (Fig. 6A). The mean curves describing α5–EosFP uptake in T-REx-HeLa-PRL-3 cells had sharper gradients for PRL-3-(Tet+) and Arf1(Q71L)-expressing cells, compared to the control cells (Tet–), indicating a faster uptake; the curves of Arf1(T31N)-transfected cells and BFA-treated PRL-3-expressing cells were overlapping with the control curves (Fig. 6B). In HeLa cells expressing wild-type PRL-3, the curve had a steeper gradient compared to curves recorded in cells co-expressing α5–EosFP together with PRL-3(C104S), PRL-3(ΔCAAX) or PRL-3(ΔMXXE), respectively (Fig. 6C). Wild-type PRL-3 and the constitutively active Arf1(Q71L) significantly decreased the halftime (t_1/2) of α5–EosFP uptake in T-REx-HeLa-PRL-3 cells; BFA treatment and expression of dominant-negative Arf1(T31N) abolished the PRL-3 effect (Fig. 6D; supplementary material Table S9). The same results were obtained using transient transfection of Arf1 mutants in parental HeLa cells (Fig. 6E–F; supplementary material Table S10). In conclusion, our data suggest that PRL-3 accelerates the recycling of α5 integrins through activation of Arf1 which enhances migration and might facilitate metastasis.

There is increasing evidence, that aberrant expression and activation of PRL-3 is implicated in the formation and progression of human cancer (Saha et al., 2001; Stephens et al., 2005). Here we show that PRL-3-mediated activation of Arf1 enhances migration through an increased turnover of α5 integrins. Our data suggest that PRL-3 activity and subcellular localization are critical for the proposed phenotype. Aside from phosphatase activity two motifs seem to be critical for PRL-3 function – the CAAX motif and the MXXE motif. The CAAX motif facilitates membrane localization of PRL-3. Previous reports have suggested that PRL-3 localized to the plasma membrane and to the early endosome through CAAX-mediated prenylation (Fiordalisi et al., 2006; Zeng et al., 2000). Our data show that PRL-3 is present at the Golgi apparatus and the recycling endosome, colocalizing there with Arf1. The association of PRL-3 with membranes was critical since PRL-3(ΔCAAX) was neither able to induce migration nor increase turnover of matrix receptors (Zeng et al., 2000). After identification of Arf1 as a PRL-3 interacting protein, characterization of the MXXE motif underlined the importance of subcellular compartmentalization for PRL-3 function. Originally described as a Golgi-targeting motif within Arf1 (Honda et al., 2005) we found the MKYE stretch to be present and evolutionary conserved in members of the PRL family. Moreover, in addition to class I Arfs and Arf5, the MXXE motif is present in 6 copies in BFA-sensitive Arf GEFs such as GBF1, BIG1 and BIG2, and in 1 copy in ARNO. This suggests that not only the subcellular localization but also the protein–protein interactions within the right compartment are critical. Since the effects of PRL-3 depended on phosphatase activity it is attractive to speculate that dephosphorylation within this compartment controls Arf1 activation. Interestingly, dephosphorylation of BIG1 and BIG2 was identified as a regulatory mechanism for Arf activation (Kuroda et al., 2007). Interestingly, our data suggest a positive feedback loop between PRL-3 and Arf1. However, in the presence of dominant-negative Arf1, there is still some PRL-3 in the perinuclear region suggesting additional and/or activity-independent binding sites for PRL-3. Thus, further studies are needed to identify the precise mechanism of Arf activation by PRL-3 and to characterize the composition of the signaling complex containing PRL-3 and Arf1 in this compartment.

A promigratory function has been previously suggested for Arf1. The recruitment of adhesion components such as paxillin required Arf1 (Liu et al., 2005). Based on our results PRL-3 might enhance the turnover of α5 integrins through Arf1. A coordinated internalization of integrins allows local detachment and the subsequent recycling of integrins facilitates new attachments. Clathrin-mediated endocytosis (CME) has been described to regulate integrin recycling. However, with the exception of the αvβ5 integrin, the evidence that clathrin contributes to continuous internalization of integrin heterodimers is weak (Pellinen and Ivaska, 2006). By contrast, α5β1 integrin is rapidly internalized despite inhibition of CME, and it is clear that the β3 integrins, αIIβ3 and αvβ3, are efficiently and rapidly internalized after disruption or deletion of clathrin-dependent motifs in the β3 cytodomain (Woods et al., 2004). On the other hand, clathrin-independent endocytosis (CIE) is regulated by the Rho family of small GTPases which are classical housekeepers of cell migration. Cdc42 and other important CIE proteins are crucially linked to migration (Simpson et al., 2008). More recently, proteomic analysis provided direct links between CIE, adhesion turnover and migration. There, CIE mediated endocytosis of key cargo proteins such as CD44, Thy-1 and β1 integrins was polarized at the leading edge of migrating fibroblasts, while transient ablation of CIE impaired the ability to migrate (Howes et al., 2010). Parameters currently identified for the CIE pathway include dynamin-independent plasma-membrane scission, enrichment in GPI-anchored proteins and most importantly reliance on Arf1 activity (Kumari and Mayor, 2008). Since Arf1 is also a major secretory GTPase (Islam et al., 2007), it is an interesting possibility that Arf1 could help in the coupling of endocytosis with exocytosis. Our data, suggesting the PRL-3-dependent activation of Arf1 and subsequent stimulation of migration through enhanced turnover of integrins, certainly support this theory.

Cycling receptors such as the transferrin receptor (TfR) or integrins are trafficked by at least two temporally and spatially distinct mechanisms: a short loop and a long loop (Caswell et al., 2008). Following internalization, receptors are delivered to early endosomes that are distributed largely in the peripheral cytoplasm. Although the mechanisms that internalize integrins differ from those that endocytose the TfR, it is clear that the post-internalization trafficking routes of integrins and the TfR are closely related. In the early endosome, key decisions are made concerning the fate of internalized receptors. Those selected for short-loop recycling are sorted to particular subdomains of the early endosome and then rapidly returned to the plasma membrane. Alternatively, recycling receptors may pass from early endosome to the perinuclear recycling compartment and return to the plasma membrane, thus completing a long loop of internalization and recycling (Pellinen and Ivaska, 2006). Recent observations suggest that both class I and class II Arfs function in transferrin recycling (Volpicelli-Daley et al., 2005). This result is further supported by the presence of BFA sensitive ARF–GEFs on the recycling endosome. The presence of BIG2 on recycling endosomes is consistent with the sensitivity of this compartment to BFA. Thus it seems likely that PRL-3-dependent activation of Arf1 also stimulates exit of recycled integrins on the level of the early endosome and the perinuclear recycling compartment.

It is interesting to note that basal migration remained unchanged or was only minimally reduced upon inhibition of class I Arfs by BFA, knockdown of Arf1/3 or transfection of dominat negative Arf1 constructs. Nevertheless, PRL-3 strongly depended of the presence and activity of Arf1. These data indicate that de novo expression of PRL-3, as seen in many forms of cancer, stimulates turnover by functionally hijacking BFA-sensitive Arf1 activation to the turnover machinery of integrins. It is increasingly appreciated that proteins that are expressed in cancer can act in signaling cascades otherwise not influenced by their presence due to subcellular sequestration (Belov et al., 2007). Our data showing that PRL-3 and activated Arf1 are recruited to intracellular membranes strongly support this hypothesis.

Taken together, we have shown here for the first time that PRL-3 and Arf1 interact in order to increase integrin turnover. In addition, these data suggest that control of endosomal structures and dynamics could be an important aspect in cancer cell behavior and specific targeting of this compartment might offer exciting new options in cancer therapy.

HeLa cells (CCL-2; LGC Standards) and T-REx-HeLa cells (Invitrogen, Karlsruhe, Germany) were maintained in DMEM (Invitrogen), with 10% (v/v) fetal calf serum (Invitrogen) and 1% (v/v) penicillin–streptomycin (Invitrogen) in a humidified atmosphere of 5% CO₂: 95% air at 37°C and passaged every 4 days. The medium for T-REx-HeLa cells was supplemented with 5 µg/ml Blasticidine (Invivogen, San Diego, CA). Fugene 6 (Roche, Mannheim, Germany) was used for transfection. T-REx-HeLa cells were stably transfected with pcDNA4/TO-PRL-3 construct followed by selection of positive clones in medium supplemented with 200 µg/ml Zeocin (Invivogen).

All synthetic oligonucleotides were purchased from Thermo Scientific (Ulm, Germany). DNA coding for human PRL-3 isoform 1 was PCR-amplified from a pET-15B-PRL-3 template (a kind gift from Irena Ekiel, NRC-BRI, Canada). The following primers were used: Forward 5′-CCGGTACCATGGCTCGGATGAACCGCC-3′; Reverse 5′-GCGGATCCTACATAACGCAGCACCGGG-3′. The PCR product was inserted into pcDNA3 (Invitrogen) to produce pcDNA3-PRL3. EGFP-PRL-3 was derived from the same template, using a forward primer (5′-CCGGTACCGCTCGGATGAACCGCCC-3′), and a reverse primer (5′-GCGGATCCTACATAACGCAGCACCGGG-3′). The PCR fragment was then cloned into pEGFP-C1 (Clontech). pBT-PRL-3 was derived from the same template by PCR, with a forward primer (5′-GCGAATTCAGCTCGGATGAACCGCCC-3′), and a reverse primer (5′-GCGGATCCCTACATAACGCAGCACCGG-3′). To create the pcDNA4/TO-PRL-3 construct, PRL-3 coding sequence was inserted into the pcDNA4/TO vector (Invitrogen). The GST-PRL-3 construct was created using a PCR template, with a forward primer (5′-CGGGATCCGCTCGGATGAACCGCCC-3′), and a reverse primer (5′-GCCTCGAGCTACATAACGCAGCACCGGG-3′) which was cloned into the corresponding sites of the pGEX-6P-1 vector (GE Healthcare). The C104S and ΔMXXE PRL-3 mutations were introduced in the pcDNA3-PRL-3 and GST-PRL-3 vectors using the Quikchange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA). pcDNA3-PRL-3(ΔCAAX) was generated by PCR using a forward primer (5′-GCGGATCCATGGCTCGGATGAACCGC-3′) and a reverse mutagenic primer (5′-GCCTCGAGCTACCGGGTCTTGTGCGTG-3′). pEGFP-PRL-3(ΔCAAX) was generated by insertion of the fragment into pEGFP-C1. pTRG-Arf5-clone 39 and pTRG-COPZ1-clone 23 were extracted from the two-hybrid assay according to the manufacturer's protocol (Stratagene). pTRG-Arf5 was generated by amplification of full-length human Arf5 cDNA from a fetal brain cDNA library (Stratagene) using a forward primer (5′-GCGGATCCGGCCTCACCGTG-3′), and a reverse primer (5′-GCGAATTCCTAGCGCTTTGACAGCTC-3′). GST-Arf5 was created by PCR, using pTRG-Arf5 as a template, with a forward primer (5′-GCGGATCCGGCCTCACCGTG-3′), and a reverse primer (5′-GCCTCGAGCTAGCGCTTTGACAGCTC-3′). The fragment was subsequently inserted into the pGEX-6P-1 vector. To generate the GST–COPZ1, COPZ1 coding sequence was inserted into the pGEX-6P-1. The GST–Arf1 construct has been described previously (Pusapati et al., 2010). Arf1–mRuby was generated by PCR, using the GST–Arf1 as a template with a forward primer (5′-GCGGTACCATGGGGAACATCTTCGCC-3′), and a reverse primer (5′-GCCTCGAGCTTCTGGTTCCGGAGCTG-3′). The PCR fragment was inserted into pcDNA3-mRuby. Arf1(T31N)–mRuby was created using the above set of primers and pXS-Arf1(T31N)-HA (a kind gift from Julie Donaldson, NHLBI, USA) as a PCR template. Arf1(Q71L)–mRuby was generated using the Quikchange site-directed mutagenesis kit and Arf1-mRuby as template. The untagged pcDNA3-Arf1(T31N) and pcDNA3-Arf1(Q71L) were created by PCR, using the same forward primer as above and a reverse primer (5′-GCGGATCCTCACTTCTGGTTCCGGAG-3′). The PCR fragments were inserted into the corresponding sites of pcDNA3 vector. The GST–GGA1-GAT construct was a kind gift from Juan S. Bonifacino (NICHD, USA). α5–EosFP was generated from pEGFP-N3-α5 (a kind gift from Alan Horwitz, PSU, USA) and pcDNA3-EosFP(T158H). EosFP coding sequence was retrieved by PCR, using a forward primer (5′-GCCCGCGGGCCCGGGATCCATCGCCACCAGTGCGATTAAGCCAG-3′), incorporating a 10-amino-acid linker (GTAGPGSIAT), and a reverse primer (5′-GCGCGGCCGCTTATCGTCTGGCATTG-3′). The PCR product was then cloned into pEGFP-N3-α5 after removal of the EGFP tag. All DNA constructs were verified by DNA sequencing.

Polyclonal anti-PRL-3 and anti-β-actin were purchased from Sigma (Steinheim, Germany), monoclonal anti-PRL-3 (clone 318) and anti-Arf1 (clone ARFS 1A9/5) from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Anti-α5 integrin (clone IIA1), anti-Arf3 (clone 41), anti-EEA1 (clone 14), anti-paxillin (clone 349), and isotype-specific IgG1 antibody (clone 107.3) were purchased from BD Biosciences (Franklin Lakes, NJ). Anti-Arf6 (clone 6ARF01) was purchased from Millipore, (Billerica, MA), anti-βCOP and anti-transferrin receptor (clone B-G24) from Abcam (Cambridge, UK), and anti-GFP from Roche (Mannheim, Germany). Enhanced chemiluminescence (ECL) detection reagents were purchased from GE Healthcare (Buckingamshire, UK). Anti-Golgin-97, anti-Rab11, Alexa-Fluor-488-, Alexa-Fluor-568- and Alexa-Fluor-647-labelled anti-mouse or anti-rabbit IgG were purchased from Invitrogen. Brefeldin A, GTPγS and GDP were purchased from Sigma. Human plasma fibronectin was from Roche. All other reagents were of the highest grade available.

After detachment, cells were incubated for 1 h in serum-free DMEM, supplemented with 0.5% BSA at 37°C. Fibronectin (50 µg/ml) was adsorbed onto coverglass and cells were allowed to spread for 60 min. Images were acquired every 10 min for a period of 15 h, using an inverted microscope (Olympus, IX71, 10×) (Olympus, Hamburg, Germany) connected to a CCD camera (Orca-HR, Hamamatsu, Japan). Individual cell velocities and migratory paths were obtained using ImageJ. Statistical analyses were performed using Student's t-test or one-way ANOVA, followed by a Newman–Keuls post-hoc test.

Bacterial two-hybrid analysis was carried out according to the manufacturer's instructions (Bacteriomatch II Two-Hybrid System, Stratagene). A fetal brain cDNA library (Stratagene) was used as a source of target vectors (pTRG). Positive cotransformants were selected on selective screening medium plates containing 5 mM 3-amino-1,2,4-triazole (3-AT; Sigma). Secondary screening was done on dual selective medium plates containing 5 mM 3-AT and streptomycin (12.5 µg/ml). DNA sequencing was employed to analyze the positive clones.

HeLa cells were transfected with EGFP-PRL-3 or EGFP empty vector, scraped and lysed with the CHAPS lysis buffer (30 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% CHAPS) supplemented with protease and phosphatase inhibitor cocktails (Roche). Affinity-purified protein complexes were eluted with Laemmli sample buffer and resolved by SDS-PAGE and further analyzed using standard methods.

GST fusion proteins were purified from transformed BL21 cells induced with 1 mM IPTG for 2 h at 37°C. Bacteria were lysed with bacterial lysis buffer [20 mM Tris, 250 mM NaCl, 2.5 mM MgCl₂, lysozyme (Sigma) 1 mg/ml] and centrifuged, supernatants were incubated with glutathione–Sepharose beads (GE Healthcare) for 2 h at 4°C. In some pull-down assays, GST-Arf1 was pretreated with GTPγS or GDP, final concentration of 1 mM, for 10 min at 30°C.

Cells were lysed with RIPA buffer and the extracts were incubated with equal amounts of the GST fusion proteins immobilized on Sepharose beads for 2 h at 4°C. After extensive washing, bound proteins were eluted with Laemmli sample buffer, resolved by SDS-PAGE, and analyzed by immunoblotting. For active-Arf pull-down assays, the GST-fused GGA1-GAT construct (Mattera et al., 2004) was used to retrieve the active Arf from cell lysates. Where required, the cells were pretreated with BFA (final concentration 5 µg/ml) for 2 h prior to experiment (quality controls for GST proteins are shown in supplementary material Fig. S4).

Cells were seeded on fibronectin-coated coverglasses. After fixation and permeabilization, the coverslips were incubated with primary and secondary antibodies for 1 h at RT. Image acquisition was performed as described above or were analyzed by confocal microscopy on a Leica TCS SP2 confocal system (Leica Microsystems, Mannheim, Germany). Colocalization was analyzed using the method described by Jaskolski et al. (Jaskolski et al., 2005). In brief, this approach calculates a correlation index and generates an image of the correlated signals from two original images. The degree of colocalization is given for every single pixel and calculated as the normalized mean deviation product (nMDP). Positive nMDPs are in accordance with positive colocalization. Picture analysis and calculation of nMDPs was done in the program MATLAB, available at http://www.synapse.u-bordeaux2.fr/nMDPmethod.m.

siRNA targeting Arf1 (5′-GUGGAAACCGUGGAGUACA-3′), the 3′UTR of Arf3 (5′-CCUAUAUGACCAAUCCCUA-3′), and a negative control siRNA were purchased from Qiagen (20 nM final) (Qiagen, Hamburg, Germany). siRNA transfections were performed using the HiPerfect reagent (Qiagen), according to the manufacturer's instructions. The efficiency of Arf1/Arf3 knockdown was validated by western blotting and quantitative real-time PCR.

RNAs were extracted from cells using QiaZol (Qiagen, Hilden, Germany), treated with DNase I and purified with RNeasy kit (Qiagen), all according to the manufacturer's instructions. cDNAs were prepared from total RNAs using Superscript reverse transcriptase (Invitrogen). The quantitative RT-PCR (qRT-PCR) for detection of PRL-3 was performed using a QuantiTect Primer Assay–Hs_PTP4A3_1_SG A (Qiagen), according to the manufacturer's protocol. qRT-PCR for detection of Arf1 and Arf3 was performed using the QuantiTect Primer Assays (Qiagen), respectively.

The pulse–chase recycling assays were based on a modified protocol of Hsu and co-workers (Powelka et al., 2004). Briefly, the cells were pulsed with anti-α5 integrin (1 µg/ml) or anti-TfR (1 µg/ml), or transferrin from human serum conjugated with Alexa Fluor 546 (Invitrogen) (5 µg/ml). Isotype-specific IgG1 antibody (1 µg/ml) served as a control for FACS analysis. Cells were treated with BFA (5 µg/ml). Surface-bound antibodies were removed by acid wash (0.5% acetic acid, 0.5 M NaCl, pH 3.0) followed by washing with serum-free DMEM. The chase was performed in serum-free DMEM at 37°C.

After the chase period cells were fixed, immunostained and further analyzed by fluorescence microscopy.

After 30 min chase, cells were washed and pelleted. The cell pellets were incubated with Alexa-Fluor-647-conjugated anti-mouse antibody (Invitrogen) (4 µg/ml) and fixed with 3% paraformaldehyde. The flow cytometry was performed on LSRII cell analyzer (BD Biosciences, Heidelberg, Germany), and the results were analyzed using the WinMDI program.

FUAP is a new method for analysis of protein turnover and trafficking by live-cell imaging, using a photoconvertible fluorescent protein (EosFP) (Wiedenmann et al., 2004). HeLa cells were transfected with α5–EosFP plasmid and plated on fibronectin-coated coverglass. Prior to photoconversion, one image in both green (filter: bp 470±40) and red (filter: bp 560±40) channel was acquired. Then, one half of the cell was photoconverted, using a special filter set (bp 400±20). Images were taken in regular time intervals (30 sec) for 1 h which was found to be the sufficient time to reach a ‘FUAP’ plateau. In order to prevent changes in size and localization of focal contacts only stationary cells were analyzed. To follow the uptake of red (converted) EosFP in the region of interest (ROI), we used the RectTrack plug-in for ImageJ (available upon request) which gave light intensity values in the ROI for each frame. Then a single exponential function with two parameters [y = a(1−e^−bx)] was fitted into the experimental values and the t_1/2 of α5-integrin uptake was calculated from the following formula: t_1/2 = −1/b ln [1−(50/a)] (SigmaPlot; Systat Software Inc, USA). Mean kinetic curves were obtained by introducing the mean values of a and b parameters acquired from individual curves into the formula above. Statistical analysis was performed using ANOVA followed by a Newman–Keuls post-hoc test.

We thank Silke Rosinger and Andreas Spyrantis for their support with flow cytometry, and Ganesh Varma Pusapati for helpful discussions and provided materials.