Skip to main content
PMC home page
eLife logoLink to eLife
. 2017 Aug 29;6:e23649. doi: 10.7554/eLife.23649

Negative regulation of urokinase receptor activity by a GPI-specific phospholipase C in breast cancer cells

Michiel van Veen 1,†, Elisa Matas-Rico 1,†, Koen van de Wetering 2,‡, Daniela Leyton-Puig 1, Katarzyna M Kedziora 1,§, Valentina De Lorenzi 3, Yvette Stijf-Bultsma 4, Bram van den Broek 1, Kees Jalink 1, Nicolai Sidenius 3, Anastassis Perrakis 4, Wouter H Moolenaar 1,✉
Editor: Jonathan A Cooper5
PMCID: PMC5576486  PMID: 28849762

Abstract

The urokinase receptor (uPAR) is a glycosylphosphatidylinositol (GPI)-anchored protein that promotes tissue remodeling, tumor cell adhesion, migration and invasion. uPAR mediates degradation of the extracellular matrix through protease recruitment and enhances cell adhesion, migration and signaling through vitronectin binding and interactions with integrins. Full-length uPAR is released from the cell surface, but the mechanism and significance of uPAR shedding remain obscure. Here we identify transmembrane glycerophosphodiesterase GDE3 as a GPI-specific phospholipase C that cleaves and releases uPAR with consequent loss of function, whereas its homologue GDE2 fails to attack uPAR. GDE3 overexpression depletes uPAR from distinct basolateral membrane domains in breast cancer cells, resulting in a less transformed phenotype, it slows tumor growth in a xenograft model and correlates with prolonged survival in patients. Our results establish GDE3 as a negative regulator of the uPAR signaling network and, furthermore, highlight GPI-anchor hydrolysis as a cell-intrinsic mechanism to alter cell behavior.

Research organism: Human

eLife digest

Every process in the body, from how cells divide to how they move around, is tightly regulated. For example, cells only migrate when they receive the correct signals from their environment. These signals are recognised by receptor proteins that sit on the cell surface and connect the outside signal with the cell’s response. However, in cancer cells, these processes are out of control, which is why cancer cells can grow very quickly or spread to many different parts of the body.

One important receptor protein is the urokinase receptor, which helps to reorganize the tissue, for example, when wounds heal, but also enables cancer cells to grow and spread. A special feature of urokinase receptor is the way it is connected to the cell surface, namely through a molecule that acts as an anchor, called the GPI anchor. The urokinase receptor and some other GPI-anchored proteins can be released from their anchor. However, until now it was not clear why and how the urokinase receptor is released from cells, or how losing the receptor affects the cell.

Now, van Veen, Matas-Rico et al. studied breast cancer cells, and discovered that an enzyme called GDE3 cuts the urokinase receptor off its GPI anchor to release the receptor from the cells. However, when breast cancer cells shed the urokinase receptor, they also lost the receptor from the cell surface in specific areas. As a result, the receptor could not work anymore. When breast cancer cells were experimentally modified to produce high levels of GDE3, the cancer cells became less mobile and aggressive.

Van Veen, Matas-Rico et al. then implanted ‘normal’ breast cancer cells, and breast cancer cells with extra GDE3 into mice, and observed that the tumors of mice with additional GDE3 grew less quickly. Moreover, breast cancer patients with high levels of GDE3 tend to live longer than patients with low levels of GDE3. These results suggest that the enzyme GDE3 can suppress tumor growth.

These findings uncover a new way how cells can alter their behavior, namely by cleaving GPI anchors at the cell surface. Future experiments will need to address how GDE3 itself is controlled, and if it releases other GPI-anchored proteins from cells. Once we know how to increase GDE3 activity in tumor cells, the new knowledge could one day lead to therapies to help patients with cancer.

Introduction

The urokinase-type plasminogen activator receptor (uPAR) is a central player in a complex signaling network implicated in a variety of remodeling processes, both physiological and pathological, ranging from embryo implantation to wound healing and tumor progression (Boonstra et al., 2011; Ferraris and Sidenius, 2013; Smith and Marshall, 2010). uPAR is a glycosylphosphatidylinositol (GPI)-anchored protein and hence lacks intrinsic signaling capacity. Instead, uPAR acts by binding two major ligands, namely the protease urokinase plasminogen activator (uPA) and the extracellular matrix (ECM) protein vitronectin (Ferraris and Sidenius, 2013; Madsen et al., 2007; Smith and Marshall, 2010). Through uPA binding, uPAR localizes plasmin generation to the cell surface and thereby promotes pericellular proteolysis and ECM degradation (Ferraris and Sidenius, 2013; Smith and Marshall, 2010). In addition, through vitronectin binding and functional interactions with integrins and growth factor receptors, uPAR activates intracellular signaling pathways leading to cytoskeletal reorganization, enhanced cell adhesion and motility and other features of tissue remodeling and cell transformation (Ferraris et al., 2014; Madsen et al., 2007; Smith and Marshall, 2010). As such, uPAR is a master regulator of extracellular proteolysis, cell motility and invasion. uPAR expression is elevated during inflammation and in many human cancers, where it often correlates with poor prognosis, supporting the view that tumor cells hijack the uPAR signaling system to enhance malignancy (Boonstra et al., 2011; Ferraris and Sidenius, 2013; Smith and Marshall, 2010). Increased uPAR expression in solid tumors and the corresponding activated stroma is being evaluated by PET-imaging for patient stratification (Persson et al., 2015).

It has long been known that full-length uPAR is released from the plasma membrane resulting in a soluble form (suPAR) (Pedersen et al., 1993; Ploug et al., 1992), which is detectable in body fluids and considered a marker of disease severity in cancer and other life-threatening disorders (Haupt et al., 2012; Hayek et al., 2016; Shariat et al., 2007; Sidenius et al., 2000; Stephens et al., 1999). Circulating suPAR is derived from activated immune and inflammatory cells (Ferraris and Sidenius, 2013; Smith and Marshall, 2010), and also from circulating tumor cells (Mustjoki et al., 2000).

Locally produced suPAR might function as a ligand scavenger to confer negative feedback on uPAR (Smith and Marshall, 2010). In addition, both uPAR and suPAR can undergo proteolytic fragmentation by uPA and other proteases, possibly leading to new signaling activities (Montuori and Ragno, 2009). Yet, despite decades of research, the mechanism of uPAR release and its physiological implications have been elusive. A GPI-specific phospholipase D (GPI-PLD) (Scallon et al., 1991) has often been assumed to mediate the shedding of GPI-anchored proteins, but this unique PLD does not function on native membranes (Mann et al., 2004).

A possible clue to the mechanism of uPAR release comes from recent studies showing that a member of the glycerophosphodiester phosphodiesterase (GDPD/GDE) family (Corda et al., 2014), termed GDE2, promotes neuronal differentiation by cleaving select GPI-anchored proteins, notably a Notch ligand regulator and heparan sulfate proteoglycans (glypicans) (Matas-Rico et al., 2016; Matas-Rico et al., 2017; Park et al., 2013). GDE2, along with GDE3 and GDE6, belongs to a GDE subfamily characterized by six-transmembrane-domain proteins with a conserved catalytic ectodomain (Figure 1A) (Corda et al., 2009; Matas-Rico et al., 2016). GDE2’s close relative, GDE3, accelerates osteoblast differentiation through an unidentified mechanism (Corda et al., 2009; Yanaka et al., 2003), while the function of GDE6 is unknown.

Figure 1. GDE3, but not GDE2, sheds uPAR from HEK293 cells.

(A) Domain structure of GDE2, GDE3 and GDE6 (left panel), and the transmembrane scheme of GDE3 (right panel). GDPD denotes catalytic glycerophosphodiesterase domain. Asterisk in GDPD domain depicts catalytic His residue in both GDE2 and GDE3. (B) Immunoblot analysis of uPAR release into the medium. HEK-uPAR cells transfected with empty vector (control), GDE2 or GDE3. PI-PLC served as positive control. (C) Mutant GDE3(H229A) fails to release uPAR. (D) Partial loss of uPAR from the plasma membrane by GDE3, as measured by flow cytometry. (E) TIRF microscopy reveals loss of uPAR from the basolateral plasma membrane. Box plot shows uPAR-GFP intensity at the ventral membrane (n = 3, mean ±SEM ****p<0001). (F) Homology modeling of the GDE2 and GDE3 catalytic domains showing surface charge distributions (blue, positive; red, negative; green line, putative GPI-binding groove; yellow line, proposed substrate-binding surface). The active site is indicated by glycerol-3-phosphate located at the template structure.

Figure 1.

Figure 1—figure supplement 1. GDE subcellular localization and induction of uPAR release from HEK293-uPAR cells.

Figure 1—figure supplement 1.

(A) Confocal images of HEK293 cells expressing human GDE2-HA, GDE3-HA or GDE6-HA, as indicated; bar, 10 μm. (B) GDE3 localizes to distinct microdomains (yellow square, open arrows) and filopodia-like extensions (orange square, solid arrow), as visualized by confocal and super-resolution microscopy. Bars, 10 μm and 1 μm respectively. (C) Immunoblot analysis showing that GDE3 reduces the membrane-anchored uPAR pool and competes with PI-PLC. The medium and lysates of HEK-uPAR cells expressing GDE3 or empty vector (EV) were analyzed without or with PI-PLC treatment (45 min.) of the cells, as indicated (left and right panels, respectively). Right lanes refer to wild-type (WT) HEK293 cells. (D) Expression and localization of GDE3 and mutant GDE3(H229A), as measured by immunoblotting and confocal microscopy, respectively, at 24 hr after transfection. Actin was used as a loading control. (E) uPAR-TM containing the transmembrane domain of EGFR is not released by GDE3, as shown by immunoblotting using anti-uPAR antibody, while uPAR-TM is properly expressed at the plasma membrane (bar, 10 μm).
Open in a new tab

Here we identify GDE3 as the first mammalian GPI-specific phospholipase C (GPI-PLC) that cleaves and sheds uPAR with consequent loss of uPAR activities in both HEK293 and breast cancer cells.

Results

GDE3, but not GDE2, sheds uPAR from HEK293 cells

We set out to determine whether uPAR can be released by any of the three related GDE family members, GDE2, GDE3 and GDE6. When expressed at relatively low levels in HEK293 cells, human GDE2 and GDE3 (HA-tagged) localized to distinct microdomains at the plasma membrane, possibly representing clustered lipid rafts where GPI-anchored proteins normally reside (Maeda and Kinoshita, 2011), as well as to filopodia-like extensions (Matas-Rico et al., 2016) (Figure 1—figure supplement 1A,B). By contrast, human GDE6 was mainly detected in intracellular compartments and therefore was not further tested (Figure 1—figure supplement 1A).

To assess GDE activity, we generated stable uPAR-expressing HEK293 cells (HEK-uPAR cells), expressed GDE2 and GDE3 and examined the appearance of suPAR in the medium, using bacterial phospholipase C (PI-PLC) as a positive control (Matas-Rico et al., 2016). Strikingly, uPAR was readily released into the medium by GDE3 and PI-PLC, but not by GDE2 (Figure 1B). GDE3 competed with exogenous PI-PLC to deplete uPAR, since PI-PLC was much less efficient in GDE3-overexpressing than in control HEK-uPAR cells (Figure 1—figure supplement 1C).

Mutating putative active-site residue H229, corresponding to H233 in GDE2 (Matas-Rico et al., 2016), abolished GDE3 activity without affecting its membrane localization (Figure 1C; Figure 1—figure supplement 1D). Furthermore, a transmembrane version of uPAR (uPAR-TM) lacking the GPI moiety (Cunningham et al., 2003) was resistant to GDE3 attack, consistent with GDE3 acting through GPI-anchor hydrolysis (Figure 1—figure supplement 1E). Flow cytometry analysis of GDE3-overexpressing cells confirmed decreased uPAR levels in the plasma membrane (Figure 1D), while TIRF microscopy revealed substantial uPAR loss from the ventral cell surface (Figure 1E).

GDE2 versus GDE3: homology modeling

The selectivity of GDE3 versus GDE2 towards uPAR cleavage is striking. In an attempt to understand the structural basis of this selectivity, we constructed homology-based models of the globular α/β barrel GDPD domains, using I-TASSER (Yang et al., 2015). We reasoned that the catalytic domains must recognize not only the PI lipid moiety at the membrane-water interface, but also the attached protein. While GDE2 and GDE3 have a similar putative GPI-binding groove leading to the active site, they show striking differences in their surface charge distribution, particularly at the putative substrate interaction surface (Figure 1F). It therefore seems likely that surface properties are a major determinant of selective substrate recognition by GDEs, a notion that should be validated by future structural studies.

GDE3 is a GPI-specific phospholipase C that cleaves uPAR in cis

We asked whether GDE3 attacks uPAR in cis (same cell) or in trans (adjacent cell), or both. By mixing GDE3-expressing cells (lacking uPAR) with uPAR-expressing cells (lacking GDE3), GDE3-expressing cells failed to shed uPAR from the GDE3-deficient cell population (Figure 2A). Thus, GDE3 acts in cis, attacking uPAR on the same plasma membrane, not on adjacent cells. To determine whether GDE3 acts as a phospholipase in GPI-anchor cleavage, we used Triton X-114 partitioning and liquid chromatography-mass spectrometry (LC-MS). Triton X-114 partitioning revealed that suPAR did not contain lipid moieties, as it was not detected in the detergent phase (Figure 2B). Next, suPAR was immunoprecipitated from the medium of GDE3-expressing cells and treated with nitrous acid (HONO) to cleave the glucosamine-inositol linkage in the GPI core (Figure 2C). Subsequent analysis by LC-MS revealed that the acid-treated suPAR samples contained inositol 1-phosphate (Figure 2D). This result defines GDE3 as the first mammalian GPI-specific phospholipase C (GPI-PLC).

Figure 2. GDE3 is a GPI-specific PLC that cleaves uPAR in cis.

Figure 2.

Open in a new tab

(A) (Left) Scheme showing uPAR cleavage in cis or trans. (Right) GDE3-expressing HEK-uPAR cells were mixed with a GDE3-deficient cell population, as indicated. Immunoblot analysis of uPAR in medium and cell lysates indicates that GDE3 acts in cis; mock refers to empty vector-transfected cells. (B) GDE3 expression leads to increased GPI-free suPAR. Conditioned medium from HEK-uPAR cells expressing GDE2 or GDE3 was subjected to Triton X-114 phase separation. suPAR in the aqueous (A) and detergent (D) fractions was analyzed by immunoblotting. (C) GPI-anchor with phospholipase cleavage sites indicated; HONO, nitrous acid. (D) Representative LC-MS ion chromatograms (m/z 259.02–259.03); inositol 1-phosphate peaks in red. HONO-treated suPAR contains inositol 1-phosphate (n = 3, mean ±SEM; *p<0.05).

GDE3 suppresses the vitronectin-dependent activities of uPAR

When compared to uPAR-deficient cells, HEK-uPAR cells showed markedly increased cell adhesion, loss of intercellular contacts and enhanced spreading with prominent lamellipodia formation on vitronectin, but not on fibronectin (Figure 3A), typical features of a Rac-driven motile phenotype, in agreement with previous studies (Kjøller and Hall, 2001; Madsen et al., 2007). Cell spreading coincided with activation of focal adhesion kinase (FAK), indicative of integrin activation (Figure 3B). Strikingly, expression of GDE3 largely abolished the uPAR-induced phenotypes and cellular responses (Figure 3A–F). Catalytically dead GDE3(H229A) had no effect, neither had overexpressed GDE2 (Figure 3F and results not shown). Thus, by releasing uPAR from the plasma membrane, GDE3 suppresses the vitronectin-dependent activities of uPAR. Treatment of diverse cell types with suPAR-enriched conditioned media from HEK293 cells did not evoke detectable cellular responses, supporting the notion that suPAR is biologically inactive, at least under cell culture conditions.

Figure 3. GDE3 suppresses uPAR activity in HEK-uPAR cells.

Figure 3.

Open in a new tab

(A) uPAR confers increased adhesion to vitronectin, but not fibronectin, which is prevented by GDE3 expression (n = 3, mean ±SEM), ****p<0.0001; n.s., not significant. (B) GDE3 inhibits FAK activation during cell adhesion. FAK activity was assayed at the indicated times after plating. (C) GDE3 inhibits uPAR-induced cell spreading, scattering and lamellipodia formation on vitronectin (bar, 10 μm). (D) Cell-matrix contact area of cells expressing the indicated constructs plated on either fibronectin or vitronectin; (-) denotes cells on uncoated cover slips. Box plots show the mean of three independent experiments. Dotted line represents the mean of control cells (n = 3, mean ±SEM, ****p<0.0001; n.s., not significant). (E) uPAR-induced lamellipodia on vitronectin disappear upon GDE3 expression, as shown by super-resolution microscopy (bar, 5 μm). (F) Quantification of cell-matrix contact area in GDE3- and GDE3(H229A)-expressing cells. GDE3(H229A) fails to affect lamellipodia formation (n = 3, mean ±SEM ***p<0.001; ****p<0.0001).

GDE3 locally depletes uPAR and suppresses its activities in breast cancer cells

We next assessed the impact of GDE3 on endogenous uPAR activity in MDA-MB-231 triple-negative breast cancer cells. These cells express relatively high levels of uPAR (Figure 4A) and its ligand uPA (LeBeau et al., 2013), thus forming an autocrine signaling loop. Expression of GDE3 (encoded by GDPD2) is relatively low in breast cancer lines (n = 51), including MDA-MB-231 cells (Barretina et al., 2012) (Figure 4B). GDE3 expression in MDA-MB-231 cells led to a modest loss of uPAR from the plasma membrane as shown by flow cytometry (Figure 4C). To determine how GDE3 expression affects localized uPAR levels at the basolateral plasma membrane, we used confocal and dual-color super-resolution microscopy in TIRF mode. Strikingly, basolateral membrane microdomains containing wild-type GDE3 showed little or no colocalization with endogenous uPAR. In marked contrast, catalytically dead GDE3(H229A) clearly colocalized with uPAR in those membrane domains (Figure 4D). Quantification of co-localization was performed on multiple cells (Figure 4D, right panel). These results strongly suggest that active GDE3 depletes uPAR levels from distinct domains at the basolateral membrane. Consistent with this, CRISPR-based knockout of GDE3 resulted in increased basolateral uPAR levels when the cells were plated on vitronectin (Figure 4E,F; Figure 4—figure supplement 1).

Figure 4. GDE3 depletes uPAR from distinct basolateral membrane domains in MDA-MB-231 breast cancer cells.

(A) Endogenous uPAR expression in MDA-MB-231 versus HEK293 cells, as determined by immunoblot. (B) Endogenous GDPD2 expression, as determined by qPCR analysis. (C) (left) Cell-surface expression of GDE3-mCherry of MDA-MB-231 cells expressing GDE3, as detected by flow cytometry. (Right) Cell-surface expression of uPAR in control (grey) and GDE3-expressing MDA-MB-231 cells (red), as detected by flow cytometry. (D) Confocal (top) and dual-color super-resolution microscopy images (bottom) of MDA-MB-231 cells expressing GDE3-GFP or catalytically dead GDE3(H229A)-GFP. Endogenous uPAR was immunostained in red. Merged images show colocalization of uPAR with GDE3(H229A) but not with wild-type GDE3 and uPAR. Scale bars, 10 μm (confocal) and 1 μm (super-resolution). Co-localization analysis (Mander's coefficient) on peripheral uPAR patches in confocal images was done using ImageJ software (n = 30 cells, three independent experiments). (E) Endogenous uPAR staining in control, GDE3-overexpressing and GDE3 knockout MDA-MB-231 cells plated on vitronectin. Two distinct GDE3 knockout clones (KO1 and KO2) were examined, as indicated. Scale bar,10 μm. (F) Quantification of basolateral uPAR-containing membrane domains referring to the cells in panel (E) (n = 3, mean ±SEM, ****p<0.0001). GDE3 suppresses the vitronectin- and uPAR-dependent phenotype of MDA-MB-231 breast cancer cells.

Figure 4.

Figure 4—figure supplement 1. GDE3 knockout validation.

Figure 4—figure supplement 1.

GDE3 knockout in MDA-MB231 cells was achieved using CRISPR/Cas9 genome editing. Surviving colonies were screened for cassette integration and indels into the query gene by PCR. Two genetically individual clones (KO1 and KO2) were selected, and knockout was confirmed using Sanger sequencing followed by TIDE deconvolution (Brinkman et al., 2014).
Open in a new tab

Wild-type MDA-MB-231 cells adopted a motile phenotype on vitronectin, as evidenced by increased cell spreading with marked lamellipodia formation (Figure 5A–C), strongly reminiscent of a uPAR-regulated phenotype. Overexpressed GDE3 abolished the vitronectin-dependent phenotype of MBD-MB-231 cells (Figure 5A–C). Very similar effects of GDE3 overexpression were observed in another uPAR-positive breast cancer cell line (triple-negative Hs578T cells) (Figure 5—figure supplement 1). Of note, no effects were observed upon GDE2 overexpression in these cells (data not shown).

Figure 5. GDE3 suppressess the vitronectin- and uPAR-dependent transformed phenotype of MDA-MB-231 breast cancer cells.

A) Confocal images showing that GDE3 prevents cell spreading and lamellipodia formation on vitronectin (VN) but not on uncoated cover slips (-). Bar, 10 μm. (B) Quantification of reduced cell spreading on vitronectin by GDE3. Non-cleavable uPAR-TM prevents GDE3 attack. ****p<0.0001; n.s., not significant. (C) Quantification of lamellipodia formation on vitronectin. ****p<0.0001. (D) Immunoblot analysis of shRNA-mediated uPAR knockdown; maximum knockdown was achieved by small hairpins #1 and #3. The upper protein band represents full-length uPAR, the lower band proteolytically cleaved uPAR(D2 +D3) (Høyer-Hansen et al., 1992). (E) GDE3 overexpression mimics the uPAR knockdown phenotype in cells plated on vitronectin (VN); bar, 10 μm. (-) denotes cells on non-coated cover slips. (F,G) Quantification of cell adhesion (F) n = 3; mean ±SEM) and cell spreading (G) induced by GDE3 and uPAR knockdown on the indicated substrates. *p<0.05 **p<0.01; ****p<0.0001. GDE3 overexpression attenuates the uPAR-dependent transformed phenotype of breast cancer cells.

Figure 5.

Figure 5—figure supplement 1. GDE3 overexpression suppresses the uPAR-vitronectin-dependent phenotype in Hs578T breast cancer cells.

Figure 5—figure supplement 1.

(A) Relative expression of uPAR (encoded by PLAUR)) in the indicated cell lines, as determined by qPCR. (B) GDE3-GFP expression in Hs578T cells, using anti-GFP antibody. (C) Confocal images showing cell morphology (F-actin staining) of GDE3-expressing versus control cells plated on vitronectin (Bar, 10 μm). (D) Quantification of cell area of GDE3-expressing versus control Hs578Tcells, plated on vitronectin (n = 3, mean ±SEM, ****p<0.0001). (E) Hs578T cells with or without stable GDE3-GFP expression were plated on vitronectin and stained for endogenous uPAR. Representative images of the basolateral membrane are depicted. Bar, 10 μm. (F) uPAR in the basolateral membrane was quantified in control and GDE3-expressing Hs578T cells plated on vitronectin (n = 3, mean ±SEM, ****p<0.0001).
Open in a new tab

To confirm that GDE3 acts through uPAR attack, we expressed non-cleavable uPAR-TM to compete out endogenous uPAR, and found that the GDE3-induced reduction of cell spreading on vitronectin was largely inhibited (Figure 5B). Furthermore, shRNA-mediated knockdown uPAR (Figure 5D) gave rise to the same phenotype as GDE3 overexpression, namely reduced cell adhesion, spreading and lamellipodia formation on vitronectin (Figure 5E,F,G).

GDE3 overexpression attenuates the transformed phenotype, slows tumor growth in vivo and correlates with improved survival in breast cancer

In long-term assays, wild-type MDA-MB-231 cells showed marked scattering, indicative of increased cell motility with loss of intercellular contacts (Figure 6A). Again, GDE3 overexpression mimicked uPAR depletion either in greatly reducing both cell motility and clonogenic potential, using either shRNA-mediated knockdown (Figure 6A,B) or CRISPR-mediated knockout of uPAR (Figure 6—figure supplement 1).

Figure 6. GDE3 attenuates the uPAR-dependent transformed phenotype of breast cancer cells.

(A) GDE3 overexpression in MDA-MB-231 cells mimics uPAR knockdown in inhibiting cell scattering. MDA-MB-231 cells grow out as scattered colonies; bar, 100 μm (n = 3, mean ±SEM; ***p<0.001; ****p<0.0001). (B) GDE3 overexpression mimics uPAR knockdown in suppressing colony formation. colonies were counted after 14 days (n = 3, mean ±SEM; ***p<0.001; ****p<0.0001. (C) Degradation of a gelatin matrix mixed with vitronectin by MDA-MB-231 cells in the presence of serum (bar, 10 μm). Arrows point to black spots where gelatin was degraded. Metalloprotease inhibitor GM6001 was used as a control. (D) Quantification of matrix degradation at 20 hr after plating (mean ±SEM; *p<0.05; **p<0.01).

Figure 6.

Figure 6—figure supplement 1. uPAR knockout in MDA-MB-231 cells phenocopies GDE3 overexpression.

Figure 6—figure supplement 1.

(A) uPAR knockout in MDA-MB-231 cells was achieved using CRISPR/Cas9 genome editing. uPAR knockout was assayed by immunoblot analysis using anti-uPAR antibody. Individual clones were picked and plated on vitronectin. (B) uPAR-deficient MDA-MB-231 cells show GDE3 overexpression phenotype. uPAR KO cells show decreased cell spreading and lamellipodia formation on vitronectin. scattering (bar, 100 μm). (C) uPAR-deficient MDA-MB-231 cells show reduced cell motility (n = 3, mean ±SEM; ****p<0.0001). (D) MDA-MB-231 wild-type or uPAR knockout colonies were counted after 14 days. ****p<0.001. (E) Gelatin matrix degradation by wild-type and uPAR-deficient MDA-MB-231. Degradation was quantified at 20 hr after plating (mean ±SEM; ****p<0.0001).
Open in a new tab

Having shown that GDE3 suppresses the non-proteolytic activities of uPAR, we next examined how GDE3 affects uPAR-driven proteolytic matrix degradation by MDA-MB-231 cells. Also in this cell system, GDE3 overexpression mimicked uPAR silicencing in inhibiting the degradation of a gelatin matrix (mixed with vitronectin) in the presence of serum (Figure 6C,D) (Figure 6—figure supplement 1). On the basis of these results, we conclude that GDE3 attenuates the transformed phenotype of uPAR-positive breast cancer cells through loss of functional uPAR.

When injected into the mammary fat pads of female nude mice, GDE3-overexpressing MDA-MB-231 cells showed diminished tumor growth over time, when compared to empty vector-expressing cells (Figure 7A), consistent with the cell-based data. However, the full implications of GDE3 expression on uPAR-dependent tumor growth remain to explored in further detail. Finally, in patients, high expression of GDPD2 was found to correlate with prolonged relapse-free survival in breast cancer, particularly in triple-negative (basal-like) subtype patients (N = 618) (Figure 7B). No such correlation was found for GDE2 (encoded by GDPD5; not shown). This suggests that GDE3/GDPD5 may serve as a marker of clinical outcome in breast cancer.

Figure 7. GDE3 overexpression in breast cancer cells slows tumor growth in mice, correlating with higher survival probability in patients.

Figure 7.

Open in a new tab

(A) Female nude mice (n = 16) were injected with either wild-type or GDE3-overexpressing MDA-MB-231 cells into the mammary fat pad, as described under Materials and methods. Tumor volume was measured every three days after injection for 9 weeks until the tumor had grown to the appropriated size (data represent the mean ±SEM; **p<0.01; ***p<0.001, t-test corrected for multiple comparison). (B) High GDPD2 expression significantly correlates with higher survival rate in triple-negative (basal-like) subtype breast cancer patients (N = 618). HR, hazard ratio with 95% confidence interval indicated. Analysis based on microarray data (www.kmplot.com). (C) Schematic model. GDE3 functions as a GPI-PLC that cleaves and sheds uPAR, leading to loss of uPAR function. VN, vitronectin, FAK, focal adhesion kinase, suPAR, soluble uPAR.

Discussion

GPI-anchoring is a compIex post-translational modification that anchors select proteins in the outer leaflet of the plasma membrane. Despite decades of research, the biological significance of GPI anchors has long remained a mystery (Kinoshita and Fujita, 2016; Paulick and Bertozzi, 2008). Some GPI-anchored proteins are released from their anchor and detected in body fluids, implying involvement of one or more endogenous GPI-specific hydrolases. Recent studies have advanced the field by showing that cleavage and shedding of certain GPI-anchored proteins is mediated by a cell-intrinsic transmembrane glycerophosphodiesterase, termed GDE2 (or GDPD5), thereby promoting neuronal differentiation through multiple signaling pathways (Matas-Rico et al., 2016; Matas-Rico et al., 2017; Park et al., 2013).

In this study, we focused on the shedding of GPI-anchored uPAR because of its regulatory role in multiple cellular and (patho)physiological activities, while soluble uPAR is considered a biomarker of various human pathologies. Here we report that GDE3 functions as a long-sought GPI-specific PLC that releases uPAR from its anchor. By contrast, its homologue GDE2 failed to release uPAR. As a consequence of GDE3 action, uPAR loses its vitronectin-dependent and matrix-degrading activities, when assayed in HEK293-uPAR and triple-negative breast cancer cells that express both uPAR and uPA. Importantly, loss of uPAR expression by GDE3 was found to be restricted to certain microdomains at the basolateral plasma membrane, where signal transduction is likely to take place. Thus, by acting as a GPI-specific PLC towards uPAR, GDE3 is a negative regulator of the uPAR signaling network (Figure 7C) that includes uPAR’s proteolytic and non-proteolytic activities. Consistent with this, GDE3 overexpression in uPA/uPAR-positive MDA-MB-231 breast cancer cells slowed tumor progression in a xenograft mouse model. Although statistically significant, the inhibitory effect of GDE3 overexpression on tumor growth was not dramatic, which should not come as a surprise since MDA-MBA-31 cells express the strongly oncogenic mutant K-RAS protein, which tends to override the regulation of numerous signaling pathways. Yet, this finding adds to the relevance of GPI-specific phospholipases in slowing tumor progression. Furthermore, high GDE3 expression was found to correlate with increased survival probability in triple-negative breast cancer patients. Interestingly, our previous work revealed a similar association between overexpression of GDE2 and positive clinical outcome in neuroblastoma patients, which appears attributable to GDE2-induced glypican shedding(Matas-Rico et al., 2016). The present patient survival analysis should be interpreted with caution, however, since involvement of uPAR release remains to be formally proven. Furthermore, we cannot rule out that GDE3 may cleave additional GPI-anchored substrates whose functional loss could contribute to positive clinical outcome.

The present results predict that, depending on its expression levels, GDE3 may downregulate normal uPAR-dependent remodeling processes. Indeed, upregulated GDE3 accelerates osteoblast differentiation (Corda et al., 2009; Yanaka et al., 2003) in a manner resembling the uPAR knockout phenotype (Furlan et al., 2007). Furthermore, a striking >200 fold upregulation of GDE3/GDPD2 is observed during blastocyst formation (Munch et al., 2016), implicating GDE3 in the invasion of pre-implantation embryos, a process in which the uPA/uPAR signaling network has been implicated (Multhaupt et al., 1994; Pierleoni et al., 1998). Although correlative, these results support the view that GDE3 is upregulated to downregulate uPAR activity in vivo. The present findings also suggest that circulating full-length suPAR should be regarded as a marker of GDE3 activity, not necessarily reflecting uPAR expression levels.

It will now be important to determine how GDE3 expression and activity are regulated and, furthermore, to explore the substrate selectivity of the respective GDEs in further detail. Homology modeling revealed striking differences in electrostatic surface properties of GDE2 versus GDE3, suggesting that protein-protein interactions may determine substrate recognition by these GDE family members. Specific GPI-anchor modifications (Kinoshita and Fujita, 2016; Paulick and Bertozzi, 2008) could also determine the sensitivity of GPI-anchored proteins to GDE attack.

Finally, when regarded in a broader context, the present and previous findings (Matas-Rico et al., 2016; Matas-Rico et al., 2017; Park et al., 2013) support the view that vertebrate GDEs, notably GDE2 and GDE3, have evolved to modulate key signaling pathways and alter cell behavior through selective GPI-anchor cleavage.

Materials and methods

Cell culture and materials

HEK293, MDA-MB-231 and Hs578T cells were obtained from the ATCC and grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and antibiotics at 37°C under 5% CO2. Original MDA-MB-231 cells were pathogen tested using the ImpactI test (Idexx Bioresearch, Westbrook, ME, USA) and were negative for all pathogens tested. All cell lines were routinely tested negative for mycoplasma contamination. Antibodies used: anti-mCh and anti-GFP, home-made; anti-Flag, M2, anti-Vinculin and β-Actin (AC-15) from Sigma; anti-uPAR (MAB807) from R&D systems; anti-uPAR (13F6) (Zhao et al., 2015); anti-FAK(pTyr397) from Thermo Fisher. Vitronectin, fibronectin, inositol 1-phosphate (dipotassium salt) and inositol were purchased from Sigma-Aldrich. B. cereus PI-PLC was from Molecular Probes. Phalloidin red (actin-stain 647 phalloidin) and green (actin-stain 488 phalloidin) were from Cytoskeleton. GM6001 was from Millipore. Research Source Identifiers: MDA-MB-231 cells RRID:CVCL_0062; Hs578T cells RRID:CVCL_0332; Antibodies: Flag M2 RRID:AB_259529; Anti Vinculin RRID:AB_10746313; Anti actin RRID:AB_2223210; uPAR RRID:AB_2165463.

Expression vectors

GDE2 cDNA was subcloned as described (Matas-Rico et al., 2016). GDE3 cDNA was amplified by PCR and subcloned into a pcDNA3-HA plasmid using AflII/HpaI (PCR product) and AflII/EcoRV (plasmid) restriction sites. GDE3 was recloned into a pcDNA3(-mCherry) construct by PCR amplification with restriction sites PmI1/Xba1, followed by vector digestion using EcoRV/Xba1. GDE6 transcript variant X7 (GDPD4; NCBI: XM_011544834.1) was cloned into pcDNA5_FRT_TO_puro (provided by dr. Geert Kops, University Medical Center Utrecht). All constructs were epitope- tagged on the C-terminus unless otherwise stated. Mutant GDE3(H229A) was generated by amplification with oligos containing the mutation, followed by Dpn1 digestion of the template. The viral plasmids (pBABE-GDE3-mCh, pBABE-uPAR-GFP) were constructed by subcloning the GDE3-pcDNA3 and uPAR-GFP-pEGFP-N1 into a pBABE plasmid. GDE3-mCherry-pcDNA3 was cut using PmeI followed by digestion of the pBABE backbone with SnaBI. uPAR-GFP-pEGFP-N1 was cut with BglII and HpaI and ligated into the pBABE vector, digested with BAMHI and SnaB1. Constructs uPAR-GFP, uPAR-FLAG and non-cleavable uPAR-TM were previously described (Caiolfa et al., 2007; Cunningham et al., 2003). Transmembrane-anchored uPAR-TM was constructed by substituting the GPI-anchoring sequence of uPAR (aa 274–313) with the transmembrane region (aa 614–653) of the human epidermal growth factor receptor (EGFR), as described (Cunningham et al., 2003).

Transfections and RNAi-mediated knockdown

Cells stably expressing uPAR-GFP or GDE3-mCherry were generated using retroviral transduction and subsequent selection with puromycin. Transient transfections were done using the calcium phosphate protocol or XtremeGene 9 agent (Roche). Stable uPAR knockdown in MDA-MB-231 cells was achieved using shRNAs in a lentiviral pLKO vector; five shRNAs from three RC human shRNA library were tested: TRCN0000052637, TRCN0000052636, TRCN0000052634, TRCN0000052633 and TRCN0000052635. The latter two were used for experiments; sequences: CCGGCCCATGAATCAATGTC TGGTACTCGAGTACCAGACATTGATTCATGGGTTTTTG and CGGGCTTGAAGA TCACCAGCCTTACTCGAGTAAGGCTGGTGATCTTCAAGCTTTTTG, respectively. For virus production, HEK293T cells were transiently transfected using calcium phosphate, and virus particles were collected 48 hr thereafter. uPAR knockdown cells were selected in medium containing 2 μg/ml puromycin.

CRISPR knockout

uPAR and GDE3 knockout cell lines were generated using CRISPR/Cas9 genome editing. CRISPR sequences were designed targeting uPAR (PLAUR; exon 2; 5’- CATGCAGTGTAAGACCAACG-3’ and 5’-CCAGGGCGCACTCTTCCACA-3’) or GDE3 (GDPD2; exon 2; 5’-AGGATGCAAACCAGCAAGG-3’) and cloned into pX330 (Cong et al., 2013). MDA-MB-231 cells were transfected with the exon-specific pX330 plasmids in addition to a plasmid containing a guide RNA to the Danio Rerio TIA gene (5’- GGTATGTCGGGAACCTCTCC -3’) and a cassette of a 2A sequence followed by a BlastR gene, flanked by two TIA target sites. Co-transfection results in infrequent integration of the BlastR gene at the targeted genomic locus by NHEJ, as previously described (Blomen et al., 2015). Successful integration of the cassette renders cells resistant to blasticidin. Three days following transfection, the culture medium was supplemented with blasticidin (25 μg/ml). Surviving colonies were clonally expanded, screened for cassette integration and indels into the query gene by PCR (GDE3; 5’- TATGAATCCTGCCCGAAAAG-3’ and 5’-AGAGCAGGCCAAACCAGATA-3’) or by western blot analysis of the target protein (uPAR).

Liquid chromatography-mass spectrometry (LC-MS)

To determine the inositol phosphate content of cleaved uPAR, suPAR was immuno-precipitated from HEK293 cell conditioned medium using anti-GFP beads (ChromoTek). To remove inositol phosphate from suPAR, the beads were treated with 0.1M acetate buffer (pH 3.5) and subsequently with 0.5M NaNO2 or 0.5M NaCl (Control) for 3 hr as previously described (Mehlert and Ferguson, 2009). Inositol phosphate-containing samples were preprocessed by adding methanol to a final concentration of 70% and shaken at 1000 RPM at room temperature for 10 min. Following centrifugation (20,400 x g at 4°C for 10 min), the supernatant was evaporated to dryness in a Speedvac at room temperature. The dried extracts were reconstituted in 50 mM ammonium acetate (pH 8.0), centrifuged (20,400 x g at 4°C for 10 min) and transferred to autosampler vials. Liquid chromatography (LC) was performed using a Dionex Ultimate 3000 RSLCnano system (Thermo Fisher Scientific). A volume of 5 μl was injected on a Zorbax HILIC PLUS column (150 × 0.5 mm, 3.0 μm particles) maintained at 30°C. Elution was performed using a gradient: (0–5 min, 20% B; 5–45 min 20–100% B; 45–50 min 100% B; 50–50.1 min 20% B; 50.1–60 min 20% B) of 100% acetonitrile (mobile phase A) and 50 mM ammonium acetate adjusted to pH 8.0 with ammonium hydroxide (mobile phase B) at a flow rate of 15 μl/min. Inositol 1-phosphate was detected with an LTQ-Orbitrap Discovery mass spectrometer (Thermo Fisher Scientific) operated in negative ionization mode scanning from m/z 258 to 260 with a resolution of 30,000 FWHM. Electrospray ionization was performed with a capillary temperature set at 300°C and the sheath, auxiliary and sweep gas flow set at 17, 13 and 1 arbitrary units (AU), respectively. Setting for Ion guiding optics were: Source voltage: 2.4 kV, capillary voltage: −18 V, Tube Lens: −83 V, Skimmer Offset: 0 V, Multipole 00 Offset: 5 V, Lens 0: 5 V, Multipole 0 Offset: 5.5 V, Lens 1: 11 V, Gate Lens Offset: 68 V, Multipole 1 Offset: 11.5 V, Front Lens: 5.5 V. Data acquisition was performed using Xcalibur software (Thermo Fisher Scientific). Reference inositol phosphate (Sigma) was used to determine the retention on the Zorbax HILIC plus column. After applying Xcalibur’s build-in smoothing algorithm (Boxcar, 7), extracted ion chromatograms (m/z 259.02–259.03) were used to semi-quantitatively determine inositol phosphate levels.

Cell adhesion and spreading

48-wells plates were coated overnight at 4°C with fibronectin (10 μg/ml) or vitronectin (5 μg/ml), or left untreated. Thereafter, plates were blocked for 2 hr at 37°C using 0.5% BSA in PBS. Cells were washed and harvested in serum-free DMEM supplemented with 0.1% BSA. Equal numbers of cells were seeded and allowed to adhere for 1 hr. Non-adherent cells were washed away using PBS. Attached cells were fixed with 4% paraformaldehyde (PFA) for 10 min, followed by washing and staining with Crystal violet (5 mg/ml in 2% ethanol) for 10 min. After extensive washing, cells were dried and lysed in 2% SDS for 30 min. Quantification was done by measuring absorbance at 570 nm using a plate reader. For cell-matrix contact area and lamellipodia measurements, coverslips were coated overnight with fibronectin or vitronectin and washed twice with PBS. Cells were trypsinized, washed and resuspended in DMEM and left to adhere and spread for 4 hr. After fixation (4% PFA) and F-actin staining with phalloidin, images were taken using confocal microscopy. Cell and lamellipodia area was quantified using an ImageJ macro.

Cell scattering and colony formation

Cell scattering was determined as described (LeBeau et al., 2013). In brief, single MDA-MB-231 cells were allowed to grow out as colonies, and the area covered by the scattered colonies (colony size) was measured at 6 days after plating. For measuring colony outgrowth, 500 cells were plated and colonies were counted after 14 days.

Matrix degradation

Coverslips were coated with OG-labelled gelatin (InVitrogen) supplemented with vitronectin (5 μg/ml). About 100.000 cells per coverslip were seeded in DMEM with 10% FCS. After 20 hr, cells were washed, fixed with 4% PFM and stained with phalloidin-Alexa647 (InVitrogen). Gelatin degradation was determined from confocal images of >15 randomly automatically chosen fields of view per coverslip (testing at least two coverslips/condition on two separate days: total four coverslips per condition). The images were randomized and the area of degradation was normalized to the total area of cells or to the number of cells.

RT-qPCRT

Total RNA was isolated using the GeneJET purification kit (Fermentas). cDNA was synthesized by reverse transcription from 2 μg RNA with oligodT 15 primers and SSII RT enzyme (Invitrogen). Relative qPCR was measured on a 7500 Fast System (Applied Biosystems) as follows: 95°C for 2 min followed by 40 cycles at 95°C for 15 s followed by 60°C for 1 min. 200 nM forward and reverse primers, 16 μl SYBR Green Supermix (Applied Biosystems) and diluted cDNA were used in the final reaction mixture. GAPDH was used as reference gene and milliQ was used as negative control. Normalized expression was calculated following the equation NE = 2(Ct target-Ct reference). Primers used: GDE3, forward TCAGCAGGACCACGAATGTA, reverse GCTGCAGCTTCCTCCAATAG; uPAR, forward AATGGCCGCCAGTGTTACAG, reverse CAGGAGACATCAATGTGGTTC; Cyclophilin, forward CATCTGCACTGCCAAGACTGA, reverse TTGCCAAACACCACATGCTT. For RT-PCR 25 ng cDNA was used in a RT-PCR reaction using GoTaq (Promega). Primer sequences GAPDH forward 5’-CCATGTTCGTCATGGGTGT-3’, GAPDH reverse 5’-CCAGGGGTGCTAAGCAGTT-3’, GDE3 forward 1 5’-TGTTTGAGACTGATGTGATGGTC-3’, GDE3 reverse 1 5’-TTCGGGTTGGGAATACAGAG-3’

Western blotting

For Western blotting, cells were washed with cold PBS, lysed in RIPA buffer supplemented with protease inhibitors and spun down. Protein concentration was measured using a BCA protein assay kit (Pierce) and LDS sample buffer (NuPAGE, Invitrogen) was added to the lysate or directly to the medium. Equal amounts were loaded on SDS-PAGE pre-cast gradient gels (4–12% Nu-Page Bis-Tris, Invitrogen) followed by transfer to nitrocellulose membrane. Non-specific protein binding was blocked by 5% skimmed milk in TBST; primary antibodies were incubated overnight at 4°C in TBST with 2% skimmed milk. Secondary antibodies conjugated to horseradish peroxidase (DAKO, Glostrup, Denmark) were incubated for 1 hr at room temperature; proteins were detected using ECL Western blot reagent.

Triton X-114 phase separation

HEK293 cells transiently transfected with GDE2, GDE3 or empty plasmid, were plated on PEI-coated 6-well plates. After 24 hr complete medium was replaced with 1 ml serum free medium, 24 hr thereafter the conditioned medium was collected. 2% pre-condensed Triton X-114 was added to ice-cold conditioned medium and phases were separated as previously described (Doering et al., 2001). The top aqueous phase and the bottom detergent phase were analyzed by SDS-PAGE and Western blotting.

Microscopy

Cells cultured on 24 mm, #1,5 coverslips were washed and fixed with 4% PFA, permeabilized with 0.1% Triton X-100 and blocked with 2% BSA for 1 hr. Incubation with primary antibodies was done for 1 hr followed by incubation with Alexa-conjugated antibodies or Phalloidin for 45 min at room temperature. For confocal microscopy, cells were washed with PBS, mounted with Immnuno-MountTM (Thermo Scientific) and visualized on a LEICA TCS-SP5 confocal microscopy (63 x objective). Super-resolution imaging was done using a SR-GSD Leica microscope equipped with an oxygen scavenging system, as previously described (Matas-Rico et al., 2016). In short, 15000 frames were taken in TIRF mode, at 10 ms exposure time. After post image analysis, movies were analyzed and corrected using the ImageJ plugin Thunderstorm (http://imagej.nih.gov/ij/) followed by correction with an ImageJ macro using the plugin Image Stabilizer. For Total Internal Reflection (TIRF) microscopy, HEK293 cells stably expressing UPAR-GFP and transiently transfected with GDE3-mCherry were imaged using a Leica AM TIRF MC microscope with a HCX PL APO 63x, 1.47 NA oil immersion lens. Excitation was at 488 and 561 nm and detection of fluorescence emission was by a GR filter cube (Leica). Before each experiment, automatic laser alignment was carried out and TIRF penetration depth was set to 200 nm. Data were acquired at 500 ms frame rate. Basolateral uPAR patches were visualized using confocal microscopy and the area was quantified using an ImageJ macro that measured the uPAR patches based on fluorescence intensity.

Flow cytometry

HEK293 cells stably expressing uPAR-GFP were left untreated, treated with PI-PLC or transiently transfected with GDE3-mCherry. MDA-MB-231 cells were stably transfected with GDE3-mCherry. Cells were trypsinized, blocked in 2%BSA and stained with rabbit anti-GFP primary antibody or mouse anti-uPAR (13F6) antibody followed by AlexaFluor-647 coupled anti-rabbit secondary antibody. Cells were analyzed using a BD LSR Fortessa flow cytometer.

Mouse xenograft assay

Equal groups (n = 16) of eight-week-old female NMRI nude mice were injected subcutaneously into the mammary fat pads. Prior to injection, GDE3-expressing or control MDA-MB-231 cells (5 × 105) were suspended in an equal volume of cold phosphate buffered saline and diluted 1:1 with Matrigel. The tumor growth was monitored 3 times a week for 9 weeks by caliper measurements. The tumor volumes were calculated with the formula: 0,5 x length x width¬2 and statistical analysis of the tumor growth was done using a multiple t-test corrected for multiple comparison. A P value < 0.05 was considered statistically different. The mouse experiments were approved by the Animal Ethics Committee of the Netherlands Cancer Institute (protocol number: 30100 2015 407 appendix 1 WP 6061).

Statistical analysis

For all single comparisons, a two-tailed unpaired Student’s t-test was used; for multiple comparisons, an ordinary ANOVA with Tukey’s test was used. A P value < 0.05 was considered statistically significant. Error bars shown in the bar diagrams were calculated as the standard error of the mean (SEM); whiskers in the box plots depict 95% confidence intervals.

Acknowledgements

We thank the people from the Preclinical Intervention Unit of the Mouse Clinic for Cancer and Ageing (MCCA) at the NKI for performing the animal experiments. This work was supported by the Netherlands Organisation for Scientific Research (NWO) and the Dutch Cancer Society (KWF).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Wouter H Moolenaar, Email: w.moolenaar@nki.nl.

Jonathan A Cooper, Fred Hutchinson Cancer Research Center, United States.

Funding Information

This paper was supported by the following grants:

  • Nederlandse Organisatie voor Wetenschappelijk Onderzoek Graduate Student Fellowship to Daniela Leyton-Puig.

  • KWF Kankerbestrijding Graduate Student Fellowship to Katarzyna M Kedziora.

Additional information

Competing interests

No competing interests declared.

Author contributions

Conceptualization, Data curation, Formal analysis, Validation, Investigation, Methodology.

Conceptualization, Data curation, Formal analysis, Validation, Investigation, Methodology.

Formal analysis, Investigation, Methodology.

Formal analysis, Investigation, Methodology.

Formal analysis, Investigation, Methodology.

Investigation, Methodology.

Investigation, Methodology.

Methodology.

Supervision, Investigation, Methodology.

Resources, Methodology, Writing—review and editing.

Formal analysis, Supervision, Validation, Investigation, Methodology.

Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing—original draft, Writing—review and editing.

Ethics

Animal experimentation: The mouse experiments were approved by the Animal Ethics Committee of the Netherlands Cancer Institute (protocol number: 30100 2015 407 appendix 1).

Additional files

Transparent reporting form
elife-23649-transrepform.docx (244.4KB, docx)
DOI: 10.7554/eLife.23649.014

References

  1. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, Wilson CJ, Lehár J, Kryukov GV, Sonkin D, Reddy A, Liu M, Murray L, Berger MF, Monahan JE, Morais P, Meltzer J, Korejwa A, Jané-Valbuena J, Mapa FA, Thibault J, Bric-Furlong E, Raman P, Shipway A, Engels IH, Cheng J, Yu GK, Yu J, Aspesi P, de Silva M, Jagtap K, Jones MD, Wang L, Hatton C, Palescandolo E, Gupta S, Mahan S, Sougnez C, Onofrio RC, Liefeld T, MacConaill L, Winckler W, Reich M, Li N, Mesirov JP, Gabriel SB, Getz G, Ardlie K, Chan V, Myer VE, Weber BL, Porter J, Warmuth M, Finan P, Harris JL, Meyerson M, Golub TR, Morrissey MP, Sellers WR, Schlegel R, Garraway LA. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature. 2012;483:603–307. doi: 10.1038/nature11003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Blomen VA, Májek P, Jae LT, Bigenzahn JW, Nieuwenhuis J, Staring J, Sacco R, van Diemen FR, Olk N, Stukalov A, Marceau C, Janssen H, Carette JE, Bennett KL, Colinge J, Superti-Furga G, Brummelkamp TR. Gene essentiality and synthetic lethality in haploid human cells. Science. 2015;350:1092–1096. doi: 10.1126/science.aac7557. [DOI] [PubMed] [Google Scholar]
  3. Boonstra MC, Verspaget HW, Ganesh S, Kubben FJ, Vahrmeijer AL, van de Velde CJ, Kuppen PJ, Quax PH, Sier CF. Clinical applications of the urokinase receptor (uPAR) for cancer patients. Current Pharmaceutical Design. 2011;17:1890–1910. doi: 10.2174/138161211796718233. [DOI] [PubMed] [Google Scholar]
  4. Brinkman EK, Chen T, Amendola M, van Steensel B. Easy quantitative assessment of genome editing by sequence trace decomposition. Nucleic Acids Research. 2014;42:e168. doi: 10.1093/nar/gku936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Caiolfa VR, Zamai M, Malengo G, Andolfo A, Madsen CD, Sutin J, Digman MA, Gratton E, Blasi F, Sidenius N. Monomer dimer dynamics and distribution of GPI-anchored uPAR are determined by cell surface protein assemblies. The Journal of Cell Biology. 2007;179:1067–1082. doi: 10.1083/jcb.200702151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, Hsu PD, Wu X, Jiang W, Marraffini LA, Zhang F. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;339:819–823. doi: 10.1126/science.1231143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Corda D, Kudo T, Zizza P, Iurisci C, Kawai E, Kato N, Yanaka N, Mariggiò S. The developmentally regulated osteoblast phosphodiesterase GDE3 is glycerophosphoinositol-specific and modulates cell growth. Journal of Biological Chemistry. 2009;284:24848–24856. doi: 10.1074/jbc.M109.035444. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Corda D, Mosca MG, Ohshima N, Grauso L, Yanaka N, Mariggiò S. The emerging physiological roles of the glycerophosphodiesterase family. FEBS Journal. 2014;281:998–1016. doi: 10.1111/febs.12699. [DOI] [PubMed] [Google Scholar]
  9. Cunningham O, Andolfo A, Santovito ML, Iuzzolino L, Blasi F, Sidenius N. Dimerization controls the lipid raft partitioning of uPAR/CD87 and regulates its biological functions. The EMBO Journal. 2003;22:5994–6003. doi: 10.1093/emboj/cdg588. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Doering TL, Englund PT, Hart GW. Detection of glycophospholipid anchors on proteins. Current Protocols in Protein Science. 2001;Chapter 12:Unit 12.5. doi: 10.1002/0471140864.ps1205s02. [DOI] [PubMed] [Google Scholar]
  11. Ferraris GM, Schulte C, Buttiglione V, De Lorenzi V, Piontini A, Galluzzi M, Podestà A, Madsen CD, Sidenius N. The interaction between uPAR and vitronectin triggers ligand-independent adhesion signalling by integrins. The EMBO Journal. 2014;33:2458–2472. doi: 10.15252/embj.201387611. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Ferraris GM, Sidenius N. Urokinase plasminogen activator receptor: a functional integrator of extracellular proteolysis, cell adhesion, and signal transduction. Seminars in Thrombosis and Hemostasis. 2013;39:347–355. doi: 10.1055/s-0033-1334485. [DOI] [PubMed] [Google Scholar]
  13. Furlan F, Galbiati C, Jorgensen NR, Jensen JE, Mrak E, Rubinacci A, Talotta F, Verde P, Blasi F. Urokinase plasminogen activator receptor affects bone homeostasis by regulating osteoblast and osteoclast function. Journal of Bone and Mineral Research. 2007;22:1387–1396. doi: 10.1359/jbmr.070516. [DOI] [PubMed] [Google Scholar]
  14. Haupt TH, Petersen J, Ellekilde G, Klausen HH, Thorball CW, Eugen-Olsen J, Andersen O. Plasma suPAR levels are associated with mortality, admission time, and Charlson Comorbidity Index in the acutely admitted medical patient: a prospective observational study. Critical Care. 2012;16:R130. doi: 10.1186/cc11434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hayek SS, Quyyumi AA, Reiser J. Soluble Urokinase Receptor and Chronic Kidney Disease. The New England Journal of Medicine. 2016;374:891. doi: 10.1056/NEJMc1515787. [DOI] [PubMed] [Google Scholar]
  16. Høyer-Hansen G, Rønne E, Solberg H, Behrendt N, Ploug M, Lund LR, Ellis V, Danø K. Urokinase plasminogen activator cleaves its cell surface receptor releasing the ligand-binding domain. The Journal of Biological Chemistry. 1992;267:18224–18229. [PubMed] [Google Scholar]
  17. Kinoshita T, Fujita M. Biosynthesis of GPI-anchored proteins: special emphasis on GPI lipid remodeling. Journal of Lipid Research. 2016;57:6–24. doi: 10.1194/jlr.R063313. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kjøller L, Hall A. Rac mediates cytoskeletal rearrangements and increased cell motility induced by urokinase-type plasminogen activator receptor binding to vitronectin. The Journal of Cell Biology. 2001;152:1145–1158. doi: 10.1083/jcb.152.6.1145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. LeBeau AM, Duriseti S, Murphy ST, Pepin F, Hann B, Gray JW, VanBrocklin HF, Craik CS. Targeting uPAR with antagonistic recombinant human antibodies in aggressive breast cancer. Cancer Research. 2013;73:2070–2081. doi: 10.1158/0008-5472.CAN-12-3526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Madsen CD, Ferraris GM, Andolfo A, Cunningham O, Sidenius N. uPAR-induced cell adhesion and migration: vitronectin provides the key. The Journal of Cell Biology. 2007;177:927–939. doi: 10.1083/jcb.200612058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Maeda Y, Kinoshita T. Structural remodeling, trafficking and functions of glycosylphosphatidylinositol-anchored proteins. Progress in Lipid Research. 2011;50:411–424. doi: 10.1016/j.plipres.2011.05.002. [DOI] [PubMed] [Google Scholar]
  22. Mann KJ, Hepworth MR, Raikwar NS, Deeg MA, Sevlever D. Effect of glycosylphosphatidylinositol (GPI)-phospholipase D overexpression on GPI metabolism. Biochemical Journal. 2004;378:641–648. doi: 10.1042/bj20031326. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Matas-Rico E, van Veen M, Leyton-Puig D, van den Berg J, Koster J, Kedziora KM, Molenaar B, Weerts MJ, de Rink I, Medema RH, Giepmans BN, Perrakis A, Jalink K, Versteeg R, Moolenaar WH. Glycerophosphodiesterase GDE2 Promotes Neuroblastoma Differentiation through Glypican Release and Is a Marker of Clinical Outcome. Cancer Cell. 2016;30:548–562. doi: 10.1016/j.ccell.2016.08.016. [DOI] [PubMed] [Google Scholar]
  24. Matas-Rico E, van Veen M, Moolenaar WH. Neuronal differentiation through GPI-anchor cleavage. Cell Cycle. 2017;16:388–389. doi: 10.1080/15384101.2016.1259894. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Mehlert A, Ferguson MAJ. Proteomic scale high-sensitivity analyses of GPI membrane anchors. Glycoconjugate Journal. 2009;26:915–921. doi: 10.1007/s10719-008-9116-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Montuori N, Ragno P. Multiple activities of a multifaceted receptor: roles of cleaved and soluble uPAR. Frontiers in Bioscience. 2009;14:2494–2503. doi: 10.2741/3392. [DOI] [PubMed] [Google Scholar]
  27. Multhaupt HA, Mazar A, Cines DB, Warhol MJ, McCrae KR. Expression of urokinase receptors by human trophoblast. A histochemical and ultrastructural analysis. Laboratory Investigation; a Journal of Technical Methods and Pathology. 1994;71:392–400. [PubMed] [Google Scholar]
  28. Munch EM, Sparks AE, Gonzalez Bosquet J, Christenson LK, Devor EJ, Van Voorhis BJ. Differentially expressed genes in preimplantation human embryos: potential candidate genes for blastocyst formation and implantation. Journal of Assisted Reproduction and Genetics. 2016;33:1017–1025. doi: 10.1007/s10815-016-0745-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Mustjoki S, Sidenius N, Sier CF, Blasi F, Elonen E, Alitalo R, Vaheri A. Soluble urokinase receptor levels correlate with number of circulating tumor cells in acute myeloid leukemia and decrease rapidly during chemotherapy. Cancer Research. 2000;60:7126–7132. [PubMed] [Google Scholar]
  30. Park S, Lee C, Sabharwal P, Zhang M, Meyers CL, Sockanathan S. GDE2 promotes neurogenesis by glycosylphosphatidylinositol-anchor cleavage of RECK. Science. 2013;339:324–328. doi: 10.1126/science.1231921. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Paulick MG, Bertozzi CR. The glycosylphosphatidylinositol anchor: a complex membrane-anchoring structure for proteins. Biochemistry. 2008;47:6991–7000. doi: 10.1021/bi8006324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Pedersen N, Schmitt M, Rønne E, Nicoletti MI, Høyer-Hansen G, Conese M, Giavazzi R, Dano K, Kuhn W, Jänicke F. A ligand-free, soluble urokinase receptor is present in the ascitic fluid from patients with ovarian cancer. Journal of Clinical Investigation. 1993;92:2160–2167. doi: 10.1172/JCI116817. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Persson M, Skovgaard D, Brandt-Larsen M, Christensen C, Madsen J, Nielsen CH, Thurison T, Klausen TL, Holm S, Loft A, Berthelsen AK, Ploug M, Pappot H, Brasso K, Kroman N, Højgaard L, Kjaer A. First-in-human uPAR PET: Imaging of Cancer Aggressiveness. Theranostics. 2015;5:1303–1316. doi: 10.7150/thno.12956. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Pierleoni C, Samuelsen GB, Graem N, Rønne E, Nielsen BS, Kaufmann P, Castellucci M. Immunohistochemical identification of the receptor for urokinase plasminogen activator associated with fibrin deposition in normal and ectopic human placenta. Placenta. 1998;19:501–508. doi: 10.1016/S0143-4004(98)91043-8. [DOI] [PubMed] [Google Scholar]
  35. Ploug M, Eriksen J, Plesner T, Hansen NE, Danø K. A soluble form of the glycolipid-anchored receptor for urokinase-type plasminogen activator is secreted from peripheral blood leukocytes from patients with paroxysmal nocturnal hemoglobinuria. European Journal of Biochemistry. 1992;208:397–404. doi: 10.1111/j.1432-1033.1992.tb17200.x. [DOI] [PubMed] [Google Scholar]
  36. Scallon BJ, Fung WJ, Tsang TC, Li S, Kado-Fong H, Huang KS, Kochan JP. Primary structure and functional activity of a phosphatidylinositol-glycan-specific phospholipase D. Science. 1991;252:446–448. doi: 10.1126/science.2017684. [DOI] [PubMed] [Google Scholar]
  37. Shariat SF, Roehrborn CG, McConnell JD, Park S, Alam N, Wheeler TM, Slawin KM. Association of the circulating levels of the urokinase system of plasminogen activation with the presence of prostate cancer and invasion, progression, and metastasis. Journal of Clinical Oncology. 2007;25:349–355. doi: 10.1200/JCO.2006.05.6853. [DOI] [PubMed] [Google Scholar]
  38. Sidenius N, Sier CF, Ullum H, Pedersen BK, Lepri AC, Blasi F, Eugen-Olsen J. Serum level of soluble urokinase-type plasminogen activator receptor is a strong and independent predictor of survival in human immunodeficiency virus infection. Blood. 2000;96:4091–4095. [PubMed] [Google Scholar]
  39. Smith HW, Marshall CJ. Regulation of cell signalling by uPAR. Nature Reviews Molecular Cell Biology. 2010;11:23–36. doi: 10.1038/nrm2821. [DOI] [PubMed] [Google Scholar]
  40. Stephens RW, Nielsen HJ, Christensen IJ, Thorlacius-Ussing O, Sørensen S, Danø K, Brünner N. Plasma urokinase receptor levels in patients with colorectal cancer: relationship to prognosis. Journal of the National Cancer Institute. 1999;91:869–874. doi: 10.1093/jnci/91.10.869. [DOI] [PubMed] [Google Scholar]
  41. Yanaka N, Imai Y, Kawai E, Akatsuka H, Wakimoto K, Nogusa Y, Kato N, Chiba H, Kotani E, Omori K, Sakurai N. Novel membrane protein containing glycerophosphodiester phosphodiesterase motif is transiently expressed during osteoblast differentiation. Journal of Biological Chemistry. 2003;278:43595–43602. doi: 10.1074/jbc.M302867200. [DOI] [PubMed] [Google Scholar]
  42. Yang J, Yan R, Roy A, Xu D, Poisson J, Zhang Y. The I-TASSER Suite: protein structure and function prediction. Nature Methods. 2015;12:7–8. doi: 10.1038/nmeth.3213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Zhao B, Gandhi S, Yuan C, Luo Z, Li R, Gårdsvoll H, de Lorenzi V, Sidenius N, Huang M, Ploug M. Mapping the topographic epitope landscape on the urokinase plasminogen activator receptor (uPAR) by surface plasmon resonance and X-ray crystallography. Data in Brief. 2015;5:107–113. doi: 10.1016/j.dib.2015.08.027. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision letter

Editor: Jonathan A Cooper1

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Negative regulation of uPAR activity by a GPI-specific phospholipase C" for consideration by eLife. Your article has been favorably evaluated by Jonathan Cooper (Senior Editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors. The following individuals involved in review of your submission have agreed to reveal their identity: Michael Ploug (Reviewer #2); Reinhard Fässler (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

GPI-membrane-bound uPAR can be cleaved from the cell surface to produce a soluble form (soluble uPAR – suPAR), which is readily detected in blood and urine and is markedly increased in pathological conditions, such as inflammation and in various types of cancer. The mechanism of uPAR release and its physiological implications are still largely unknown. The manuscript by van Veen et al., reports the novel finding that the transmembrane glycerophosphodiesterase GDE3 can promote shedding of uPAR in cultured cells by cleavage of uPAR's GPI-anchor via a phospholipase C-like activity. This finding is interesting because of suPAR's role as a prognostic biomarker for poor patient prognosis. Moreover, increased uPAR expression in solid tumors (and the corresponding activated stroma) is being evaluated by PET-imaging for patient stratification (Persson et al. 2015, Theranostics 5, 1303-1306). However, despite some clear strengths of the submitted manuscript, the research appears to have some important shortcomings and limitations, which the authors should address in a revised submission:

Major points:

1) The bulk of the biochemical data on the shedding of uPAR driven by GDE3 expression/activity performed in transfected HEK293 cells (co-transfection of both uPAR & GDE3) or MDA-MB-231 breast carcinoma cell line (transfection of GDE3). Under these conditions it appears that only a fraction of the total uPAR is actually shed from the cell surface. Consequently, it is recommended the GDE3-mediated release of uPAR should be tested in cell lines that endogenously co-express both uPAR and GDE3 in which GDE3 expression is inhibited or silenced by RNA interference or CRISPR/CAS9 strategies.

2) In their Introduction the authors posit that the mechanism(s) of shedding of soluble uPAR remained enigmatic until their discovery of the potential role of GDE3 in cleaving uPAR's GPI linkage. This is an over-simplification since it is that uPAR can be shed by proteolysis. Indeed, the proteolytically released domain I fragment is actually the more powerful biomarker for poor prognosis and correlates better to cancer burden in a GEM model of breast cancer (Thurison et al., 2015 Mol Carcinog 55: 717-731). Consequently, the authors should clarify the relative role of the different mechanisms of uPAR shedding in the text.

3) In Figure 1B and 1C the GDE3-mediated release of uPAR to the medium and the reduction in uPAR at the cell surface (Figure 1D) is convincingly shown. However, it also seems that this is a relatively ineffective shedding (despite GDE3 being expressed by transfection). Indeed, the flow cytometry analysis appears to indicate that GDE3 promotes only 10% of the uPAR shedding compared to PI-PLC. This should be quantified and commented on. Moreover, how does the subcellular localization of GDE3 (filopodia, microclusters) compare to that of uPAR in cells that do not so heavily overexpress GDE3 or in the invasive front of solid tumors where uPAR is believed to be concentrated?

4) Figure 1 demonstrates that GDE3 promotes uPAR shedding by a phospholipase C-like mechanism. However, it has been noted by others (Ploug et al., 1991 JBC 266: 1926-1933) that GPI-anchors carrying an additional acylation at the 2-hydroxyl position of the inositol ring would be refractory to the action of GDE3 (and experimentally to GPI-PLC). Since a substantial proportion (~50%) of uPAR may carry this acylation event in some cells, the authors should discuss how this might alter uPAR shedding by the GDE3 mechanism in some cells.

5) In Figure 4 the authors investigate the effect of GDE3 transfection on MDA-MB-231 cells ability to cleave gelatin (denatured collagen). This is an unusual approach since it is well established that uPAR functions by focalizing uPA-mediated plasminogen activation to the cell surface and that the physiologically important substrate for this is fibrin – not collagen (Connolly et al., 2010 Blood 116: 1593-1603). Moreover, the control for inhibition of this activity should not be a broad spectrum MMP inhibitor but alpha2-antiplasmin or prevention of uPA binding by adding a neutralizing antibody or an inactivated variant of uPA (DFP-treated uPA, active site mutated (S356A) or simply the uPAR binding fragment ATF.

6) In the second paragraph of the Results and Discussion the authors speculate, based on in silico modeling, that the selectivity of GDE3 for uPAR arises due to an exosite interaction with the protein moiety of uPAR (Figure 1F). This appears highly speculative since the authors have not experimentally demonstrated selectivity of GDE3 for uPAR over other other GPI-anchored uPAR homologs in parallel (e.g. CD177, C4.4A, TEX101 or Haldisin). Moreover, the cleavage of the uPAR-GFP-hybrid (Figure 1E) and uPAR-FLAG (Figure 1—figure supplement 2) would tend to argue against such specificity. In their Materials and methods they state that these gene fusions were generated as described by Cunningham et al. 2003 (although no GFP version was described in that paper). In the experiments described by Cunningham et al., the FLAG epitope tag or TM-region is fused at position 275 or 277 in uPAR, which eliminates the region of uPAR proximal to the GPI-attachment site (position 283) thereby eliminating the protein sequence that van Veen et al. propose to serve in exosite binding. It is possible that the citation of Cunningham et al. 2003 may be in error and what the authors actually used was a more recent strategy in which GFP is inserted between uPAR and the GPI-anchoring signal (Hellriegel et al. 2011, FASEB J 25, 2882-2897) leaving residues 276-283 of uPAR in close enough proximity to occupy a putative exosite in GDE3. Although potentially interesting, this suggestion is nevertheless also complicated by the fact that this region is completely different between mouse and human uPAR (and should be mirrored by a similar difference in GDE3). If not elaborated on experimentally this speculation should thus be omitted.

7) In the seventh paragraph of the Results and Discussion it states that invasion of pre-implantation embryos is a known uPA/uPAR-dependent process is incorrect. Although uPAR is expressed by giant trophoblasts, the fact that genetic ablation of uPAR has no effect on mouse fertility (or development, or hemostasis) suggests that this system is not essential for the invasion of pre-implantation mouse embryos (Bugge et al. 1995, JBC 270, 16886).

8) It would be interesting, and potentially important, to test the effects of GDE3 over-expression on the in vivo tumorigenicity and/or metastasis of MDA-MB-231 cells in appropriately xenografted mice. More specifically, to what extent does GDE3 over-expression diminish the in vivo tumorigenicity of MDA-MB-231 cells and to what extent can any GDE3-mediated inhibitory effects on tumorigenesis be reversed by the expression of the GDE3 resistant uPAR-TM construct?

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Negative regulation of urokinase receptor activity by a GPI-specific phospholipase C in breast cancer cells" for further consideration at eLife. Your revised article has been favorably evaluated by Jonathan Cooper (Senior editor), a Reviewing editor, and two reviewers.

The revised paper is much improved but the new data have raised additional questions. The paper now provides strong evidence that over-expressing GDE3 (but not GDE2) causes release of a soluble form of uPAR (suPAR). Combined with the in vitro experiments showing GPI-PLC activity of GDE3 but not GDE2 on uPAR, the paper makes a strong case for GDE3 down-regulating uPAR by cleavage of the inositol phosphodiester of the GPI anchor. Consistent with this hypothesis, over-expressed GDE3 opposes uPAR in various functional assays done with HEK-uPAR and MDA-MB-231 cells. The reviewers were concerned, however, that the functional assays could have been done with better controls, such as catalytically-inactive GDE3 and an uncleavable form of uPAR. Experiments to show whether GDE3 is necessary for uPAR cleavage would have been better done in cells with high levels of GDE3 and uPAR gene expression. In addition, the% decrease in uPAR when GDE3 is over-expressed seems rather small, raising questions whether or not uPAR is the primary substrate. However, the original round of review did not bring out these issues prominently, so we are only recommending minor revisions to be clear about the limitations of the results.

1. Results: "Flow cytometry analysis confirmed loss of uPAR from the plasma membrane by GDE3 (Figure 1D), while TIRF microscopy revealed that uPAR loss occurred largely at the basolateral plasma membrane, where integrins normally mediate cell-matrix adhesion (Figure 1E)."

"Loss" is a bit strong, and basolateral refers to a membrane domain that is larger than that seen by TIRF microscopy. Perhaps revise to:

"Flow cytometry analysis of GDE3 over-expressing cells confirmed decreased levels of uPAR in the plasma membrane (Figure 1D), while TIRF microscopy revealed substantial loss of uPAR loss from the ventral surface (Figure 1E)."

2. Figure 1F: Given the speculative nature of the model, please relabel the dashed yellow line as "proposed substrate interaction border".

3. Figure 4D: upper images show confocal images (focal plane unclear, please clarify) and the lower images show super-resolution, which uses TIRF and thus only the basal membrane. The results show that over-expressed WT GDE3, but not catalytically dead GDE3, segregates away from the uPAR signal (lack of co-localization). Please provide quantification of co-localization over multiple cells. In addition, if the conclusions are based on the TIRF and not confocal data, then the conclusion should be modified to make it clear that GDE3 may induce uPAR relocalization rather than (or as well as) cleavage, because only the basal membrane was visualized.

4. Figure 4D-F: Please clarify how the experiments were carried out. In particular, because uPAR may affect ECM composition and vice versa, what ECM were the cells plated on? Please clarify if n=3 is the number of independent experiments. Why wasn't suPAR release measured in these experiments? If suPAR release from MDA cells is as low as for HEK cells in Figures 1B,C, then it would not be feasible to see a decrease in MDA cells deleted for GDE3, but if MDA cells normally release detectable suPAR then the media of the GCE3 knockout cells should be assayed for suPAR.

5. Figure 7A: This experiment shows a modest but statistically significant reduction in tumor volume due to GDE3 over-expression. It is unfortunate that catalytically-dead GDE3 or uPAR-TM were not used as controls but we appreciate the time constraints. In light of the limited controls, we agree with the cautious interpretation in the discussion. However, even more caution may be justified because IHC analyses of solid human cancers show that the majority of uPAR is actually not found on the cancer cells themselves (only a few cancers deviate from this) but on the activated tumor microenvironment - mostly comprising immune cells. Therefore, the reviewers recommend adding a further caveat that the full implications of GDE3 expression on uPAR-dependent tumor growth need to be explored in more detail.

6. Please clarify in the methods the exact structure of the uPAR-TM construct. The Sidenius lab made several such constructs with the tag in different positions.

7. Figure 5D: The anti-uPAR antibody detects two bands. Are both bands depicting uPAR proteins?

8. Figure 1 – figure supplement 1C: It is shown that GDE3 competes with exogenous PI-PLC to deplete uPAR. It is unclear if PI-PLC is not working at all if GDE3 is highly expressed or less efficient. An additional lane showing uPAR and GDE3-expressing cells without PI-PLC would clarify this point.

eLife. 2017 Aug 29;6:e23649. doi: 10.7554/eLife.23649.016

Author response

Major points:

1) The bulk of the biochemical data on the shedding of uPAR driven by GDE3 expression/activity performed in transfected HEK293 cells (co-transfection of both uPAR & GDE3) or MDA-MB-231 breast carcinoma cell line (transfection of GDE3). Under these conditions it appears that only a fraction of the total uPAR is actually shed from the cell surface. Consequently, it is recommended the GDE3-mediated release of uPAR should be tested in cell lines that endogenously co-express both uPAR and GDE3 in which GDE3 expression is inhibited or silenced by RNA interference or CRISPR/CAS9 strategies.

We have addressed these points as follows:

- We have used CRISPR-based knockout strategies to knockout both GDE3 and uPAR in breast cancer cells, showing that GDE3-deficient MDA-MB-231 cells show the predicted uPAR-dependent phenotype.

- We now show that an independent triple-negative breast cancer cell line (Hs578T, which is uPAR-positive; NCI-60 panel analysis) has the same phenotype as MDA-MB-231 cells upon GDE3 overexpression.

- FACS analysis shows reduced uPAR expression at the MDA-MB-231 cell surface upon GDE3 expression. Loss of uPAR by GDE3 is indeed partial, as shown by dual-color microscopy.

- Confocal and dual-color super-resolution microscopy in TIRF mode show where GDE3 and uPAR localize at the basolateral plasma membrane, and which uPAR fraction is depleted by active GDE3 but not by inactive GDE3(H229A).

2) In their Introduction the authors posit that the mechanism(s) of shedding of soluble uPAR remained enigmatic until their discovery of the potential role of GDE3 in cleaving uPAR's GPI linkage. This is an over-simplification since it is that uPAR can be shed by proteolysis. Indeed, the proteolytically released domain I fragment is actually the more powerful biomarker for poor prognosis and correlates better to cancer burden in a GEM model of breast cancer (Thurison et al., 2015 Mol Carcinog 55: 717-731). Consequently, the authors should clarify the relative role of the different mechanisms of uPAR shedding in the text.

Proteolytic fragmentation of uPAR resulting in the release of the domain 1 fragment from cells is in our opinion most correctly referred to as uPAR-cleavage as the majority of the receptor remains anchored to the plasma membrane after the event. For clarity, we now mention proteolytic fragmentation of uPAR in the Introduction, but we believe a further expansion on this point would only add confusion, as we do not investigate proteolytic cleavage of uPAR in this study. Along the same line, we do not present any biomarker data that would allow us to conclude anything on the possible importance of the GDE3-uPAR axis for the power of (s)uPAR and (s)uPAR-fragments as biomarkers. Although very interesting, we believe discussing these issues would be to speculative.

3) In Figure 1B and 1C the GDE3-mediated release of uPAR to the medium and the reduction in uPAR at the cell surface (Figure 1D) is convincingly shown. However, it also seems that this is a relatively ineffective shedding (despite GDE3 being expressed by transfection). Indeed, the flow cytometry analysis appears to indicate that GDE3 promotes only 10% of the uPAR shedding compared to PI-PLC. This should be quantified and commented on. Moreover, how does the subcellular localization of GDE3 (filopodia, microclusters) compare to that of uPAR in cells that do not so heavily overexpress GDE3 or in the invasive front of solid tumors where uPAR is believed to be concentrated?

See point #1 above. New flow cytometry data show that part of endogenous uPAR in MDA-MB-231 breast cancer cells is shed by active GDE3 (Figure 4). In addition, we have used confocal and TRIF dual-color super-resolution microscopy showing that uPAR (both transfected and endogenous) colocalizes with catalytically dead GDE3(H229A) but not with active wild-type uPAR at distinct microdomains at the ventral plasma membrane (Figure 4). Thus, GDE3 activity regulates endogenous uPAR only in select membrane domains, which is consistent with the observation that active GDE3 does not deplete the entire uPAR membrane pool (Figure 4).

4) Figure 1 demonstrates that GDE3 promotes uPAR shedding by a phospholipase C-like mechanism. However, it has been noted by others (Ploug et al., 1991 JBC 266: 1926-1933) that GPI-anchors carrying an additional acylation at the 2-hydroxyl position of the inositol ring would be refractory to the action of GDE3 (and experimentally to GPI-PLC). Since a substantial proportion (~50%) of uPAR may carry this acylation event in some cells, the authors should discuss how this might alter uPAR shedding by the GDE3 mechanism in some cells.

This is a somewhat confusing issue. GPI-anchors are modified by inositol acylation during biosynthesis, but it is far from clear if mature GPI-anchored proteins at the plasma membrane still carry this modification. In fact, normal inositol de-acylation could be impaired under overexpression conditions, such in HEK293 cells. GPI-anchor acylation of erythrocyte acetylcholinesterase renders it resistant to bacterial PI-PLC, but not to GPI-PLD (Roberts et al. 1988, JBC 263: 18766-75), but this seems to be an exception. Consistent with previous results (for example, Hoyer-Hansen et al. 2001, Biochem J 358:673), we find that mature uPAR is efficiently shed by PI-PLC, as well as by GDE3, arguing against a GDE3-resistant fraction. Actually, Ploug et al., 1991 (JBC 266:1926) also describe efficient uPAR release by exogenous PI-PLC. That being said, we cannot find evidence for the notion that “a substantial proportion (~50%) of uPAR may carry this acylation”. We therefore prefer not to discuss this hypothesis.

5) In Figure 4 the authors investigate the effect of GDE3 transfection on MDA-MB-231 cells ability to cleave gelatin (denatured collagen). This is an unusual approach since it is well established that uPAR functions by focalizing uPA-mediated plasminogen activation to the cell surface and that the physiologically important substrate for this is fibrin – not collagen (Connolly et al., 2010 Blood 116: 1593-1603). Moreover, the control for inhibition of this activity should not be a broad spectrum MMP inhibitor but alpha2-antiplasmin or prevention of uPA binding by adding a neutralizing antibody or an inactivated variant of uPA (DFP-treated uPA, active site mutated (S356A) or simply the uPAR binding fragment ATF.

We agree that collagen is not a direct substrate for plasmin. However, it has long been known that plasmin can degrade collagen indirectly through activation of latent collagenase. See for example, Werb et al., N Engl J Med 296:1017–1023, 1977). In a tumor context, uPA/uPAR knockdown has strong effects on matrigel invasion and tumor cell angiogenesis, See for example: Subramanian et al., Int. J. Oncol. 28:831-6; 2006; Montgomery et al., Breast Cancer Res. 14:R84; 2012. Furthermore, the involvement of the uPAR-uPA signaling axis in Matrigel and gelatin degradation has been documented in various cell systems (Fleetwood, et al., J Immunol, 2014. 192:3540; Kunigal et al., Int J Cancer, 2007. 121:2307; Kargiotis et al., Int J Oncol, 2008. 33: 937; Duriseti et al. J Biol Chem, 2010. 285:26878). Since we focused on regulation of uPAR by GDE3 in breast cancer cells, we reasoned that degradation of gelatin (mixed with vitronectin and serum factors) would be a relevant model system.

To show that GDE3 expression decreases gelatin degradation through uPAR release, we originally used two hairpins against uPAR. We have now confirmed this finding by using CRISPR-based uPAR knockout cells (Figure 6). We used a widely used MMP inhibitor (GM6001) as a control, simply to show that loss of gelatin occurs through degradation.

6) In the second paragraph of the Results and Discussion the authors speculate, based on in silico modeling, that the selectivity of GDE3 for uPAR arises due to an exosite interaction with the protein moiety of uPAR (Figure 1F). This appears highly speculative since the authors have not experimentally demonstrated selectivity of GDE3 for uPAR over other GPI-anchored uPAR homologs in parallel (e.g. CD177, C4.4A, TEX101 or Haldisin). Moreover, the cleavage of the uPAR-GFP-hybrid (Figure 1E) and uPAR-FLAG (Figure 1—figure supplement 2) would tend to argue against such specificity. In their Materials and methods they state that these gene fusions were generated as described by Cunningham et al. 2003 (although no GFP version was described in that paper). In the experiments described by Cunningham et al., the FLAG epitope tag or TM-region is fused at position 275 or 277 in uPAR, which eliminates the region of uPAR proximal to the GPI-attachment site (position 283) thereby eliminating the protein sequence that van Veen et al. propose to serve in exosite binding. It is possible that the citation of Cunningham et al. 2003 may be in error and what the authors actually used was a more recent strategy in which GFP is inserted between uPAR and the GPI-anchoring signal (Hellriegel et al. 2011, FASEB J 25, 2882-2897) leaving residues 276-283 of uPAR in close enough proximity to occupy a putative exosite in GDE3. Although potentially interesting, this suggestion is nevertheless also complicated by the fact that this region is completely different between mouse and human uPAR (and should be mirrored by a similar difference in GDE3). If not elaborated on experimentally this speculation should thus be omitted.

These points are well taken. Yet we feel that some speculation is allowed in this case, also because the reviewer considers our suggestion ‘potentially interesting’. The uPAR-GFP and uPAR-FLAG constructs used in this study are those described in Caiolfa et al. JCB 2007 and Cunningham et al. 2003, respectively. In both of these constructs, the tag has been inserted within the disordered region between the membrane proximal domain 3 and the GPI anchoring signal. Although the exact insertion site of the tag within the disordered region is not the same in the two constructs, both insertions leave the mature uPAR protein intact and accessible for putative exosite interactions. Mapping the location of the exact exosite interaction and determining the selectivity of GDE3 towards uPAR-homologs goes beyond the scope the present paper, in our frank opinion.

It is true that GPI-anchor variants and the sequence of their pro-peptide could determine protein localization or selective cleavage by GPI-PLC’s. We now have obtained new data supporting the view that GDEs not only recognize PI-lipid moieties at the water-membrane interface (or the adjacent amino acids) but also the attached protein. We previously showed that GDE2 cleaves GPI-anchored glypican 6 (GPC6) (E. Matas-Rico et al. 2016). To show that GDE substrate specificity is not solely dependent on the GPI-anchor attachment or the unstructured region upstream of this signal, we constructed uPAR-GPC6 chimeras. We swapped the C-terminal region of uPAR with that of GPC6, and analyzed GDE2 enzymatic activity. GDE2 did not shed the uPAR-GPC6 chimaera, whereas GDE3 did (Van Veen et al., manuscript in preparation). Therefore, surface charge distributions could well determine protein-protein interactions and, consequently, selective substrate specificity of GDE3 versus GDE3.

7) In the seventh paragraph of the Results and Discussion it states that invasion of pre-implantation embryos is a known uPA/uPAR-dependent process is incorrect. Although uPAR is expressed by giant trophoblasts, the fact that genetic ablation of uPAR has no effect on mouse fertility (or development, or hemostasis) suggests that this system is not essential for the invasion of pre-implantation mouse embryos (Bugge et al. 1995, JBC 270, 16886).

Our original phrasing was a bit inaccurate. Although uPAR is not essential for embryonic development (neither is GDE3, for that matter; S. Sockanathan, personal communication), its absence may be overcome by redundant or compensatory mechanisms (Teesalu et al., 1998). Thus, uPAR and GDE3 appear to be ‘involved’ in the migration and invasion of pre-implantation embryos. We have revised the text accordingly, with new references.

8) It would be interesting, and potentially important, to test the effects of GDE3 over-expression on the in vivo tumorigenicity and/or metastasis of MDA-MB-231 cells in appropriately xenografted mice. More specifically, to what extent does GDE3 over-expression diminish the in vivo tumorigenicity of MDA-MB-231 cells and to what extent can any GDE3-mediated inhibitory effects on tumorigenesis be reversed by the expression of the GDE3 resistant uPAR-TM construct?

We agree that it would be interesting to test the effects of GDE3 over-expression in xenografted mice, although this would make revision of our manuscript within two months impossible. Nevertheless, we have performed the mice experiments. The new results show that GDE3 overexpression slows tumor growth (Figure 6). We did not the test the uPAR-TM construct in mice, because of time and costs considerations

[Editors' note: further revisions were requested prior to acceptance, as described below.]

[…] 1. Results: "Flow cytometry analysis confirmed loss of uPAR from the plasma membrane by GDE3 (Figure 1D), while TIRF microscopy revealed that uPAR loss occurred largely at the basolateral plasma membrane, where integrins normally mediate cell-matrix adhesion (Figure 1E)."

"Loss" is a bit strong, and basolateral refers to a membrane domain that is larger than that seen by TIRF microscopy. Perhaps revise to:

"Flow cytometry analysis of GDE3 over-expressing cells confirmed decreased levels of uPAR in the plasma membrane (Figure 1D), while TIRF microscopy revealed substantial loss of uPAR loss from the ventral surface (Figure 1E)."

We agree with the reviewers and have revised the text accordingly. Furthermore, we have included a picture showing the subcellular localization of GDE6 (Figure 1-figure supplement 1A; previously “results not shown”), and describe its cDNA cloning in the Material and methods.

2. Figure 1F: Given the speculative nature of the model, please relabel the dashed yellow line as "proposed substrate interaction border".

We have relabeled the yellow line as requested.

3. Figure 4D: upper images show confocal images (focal plane unclear, please clarify) and the lower images show super-resolution, which uses TIRF and thus only the basal membrane. The results show that over-expressed WT GDE3, but not catalytically dead GDE3, segregates away from the uPAR signal (lack of co-localization). Please provide quantification of co-localization over multiple cells. In addition, if the conclusions are based on the TIRF and not confocal data, then the conclusion should be modified to make it clear that GDE3 may induce uPAR relocalization rather than (or as well as) cleavage, because only the basal membrane was visualized.

In HEK-uPAR2 cells, GDE3 expression results in a moderate decrease of total membrane-anchored uPAR (shown by FACS). When plated on vitronectin, however, both HEK-uPAR and MDA-MB-231 cells show a substantial localized loss of uPAR from the basal membrane. We have quantified uPAR patches on the plasma membrane of wild-type, GDE3-expressing and GDE3 knockout cells using confocal microscopy (ventral membrane plane). In the revised manuscript, we have done additional experiments to allow for adequate quantification of GDE3 and GDE3(H229A) co-localization with those uPAR patches (new panel in Figure 4D, showing the Mander’s coefficient). We think it unlikely that GDE3 activity could induce uPAR ‘relocalization’, but we have weakened our conclusion.

4. Figure 4D-F: Please clarify how the experiments were carried out. In particular, because uPAR may affect ECM composition and vice versa, what ECM were the cells plated on? Please clarify if n=3 is the number of independent experiments. Why wasn't suPAR release measured in these experiments? If suPAR release from MDA cells is as low as for HEK cells in Figures 1B,C, then it would not be feasible to see a decrease in MDA cells deleted for GDE3, but if MDA cells normally release detectable suPAR then the media of the GCE3 knockout cells should be assayed for suPAR.

We now explain that the cells were plated on vitronectin, and that N=3 is the number of independent experiments. Concerning suPAR measurements in GDE3-depleted cells, this is a valid question. The problem is that GDE3 expression in breast cancer cell lines is relatively low (Barretina et al., 2012). We therefore preferred to rely on sensitive morphological assays in GDE3 knockout cells, very similar to the protocol used in our studies on GDE2-mediated glypican release. In that case, relatively little GDE2-mediated glypican release accounts for a prominent phenotype (Matas-Rico et al., Cancer Cell. 2016;30:548). Of note, also in GDE3-expressing HEK-uPAR cells the most striking response was an altered phenotype when the cells plated were on VN. In conclusion, highly localized release of GPI-anchored proteins (by either GDE3 or GDE2) can result in dramatic phenotypic outcomes. We have emphasized this point in the text.

5. Figure 7A: This experiment shows a modest but statistically significant reduction in tumor volume due to GDE3 over-expression. It is unfortunate that catalytically-dead GDE3 or uPAR-TM were not used as controls but we appreciate the time constraints. In light of the limited controls, we agree with the cautious interpretation in the discussion. However, even more caution may be justified because IHC analyses of solid human cancers show that the majority of uPAR is actually not found on the cancer cells themselves (only a few cancers deviate from this) but on the activated tumor microenvironment - mostly comprising immune cells. Therefore, the reviewers recommend adding a further caveat that the full implications of GDE3 expression on uPAR-dependent tumor growth need to be explored in more detail.

We thank the reviewers for these comments. We have rephrased the text as suggested.

6. Please clarify in the methods the exact structure of the uPAR-TM construct. The Sidenius lab made several such constructs with the tag in different positions.

The structure of the uPAR-TM construct used has been described by Cunningham et al (2003), as mentioned in the text. We now elaborate on this construct in the Material and methods.

7. Figure 5D: The anti-uPAR antibody detects two bands. Are both bands depicting uPAR proteins?

We have used two different uPAR antibodies in our study. Consistent with literature data on uPAR, we often see two bands (sometimes a smear, depending on assay conditions). According to previous studies (Hoyer-Hansen et al. JBC 267:18224;1992), the upper band represents full-length uPAR, the lower band proteolytically cleaved uPAR(D2+D3) that lacks the D1 domain. We mention this in the legend, including the above reference.

8. Figure 1 – figure supplement 1C: It is shown that GDE3 competes with exogenous PI-PLC to deplete uPAR. It is unclear if PI-PLC is not working at all if GDE3 is highly expressed or less efficient. An additional lane showing uPAR and GDE3-expressing cells without PI-PLC would clarify this point.

We have added a panel showing blots of GDE3-expressing HEK-uPAR cells without PI-PLC treatment. Together, the results show that PI-PLC is working less efficiently upon GDE3 overexpression.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Transparent reporting form
    elife-23649-transrepform.docx (244.4KB, docx)
    DOI: 10.7554/eLife.23649.014

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES