Metastasis of breast cancer cells to distant organs is responsible for ∼50% of breast cancer-related deaths in women worldwide. SHIP2 (also known as INPPL1) is a phosphoinositide 5-phosphatase for phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P3] and phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P2]. Here we show, through depletion of SHIP2 in triple negative MDA-MB-231 cells and the use of SHIP2 inhibitors, that cell migration appears to be positively controlled by SHIP2. The effect of SHIP2 on migration, as observed in MDA-MB-231 cells, appears to be mediated by PI(3,4)P2. Adhesion on fibronectin is always increased in SHIP2-depleted cells. Apoptosis measured in MDA-MB-231 cells is also increased in SHIP2-depleted cells as compared to control cells. In xenograft mice, SHIP2-depleted MDA-MB-231 cells form significantly smaller tumors than those formed by control cells and less metastasis is detected in lung sections. Our data reveal a general role for SHIP2 in the control of cell migration in breast cancer cells and a second messenger role for PI(3,4)P2 in the migration mechanism. In MDA-MB-231 cells, SHIP2 has a function in apoptosis in cells incubated in vitro and in mouse tumor-derived cells, which could account for its role on tumor growth determined in vivo.
A large number of phosphatases participate in the regulation of phosphoinositide 3-kinase (PI3K) signaling in cancer cells (Balla, 2013; Hawkins and Stephens, 2016). In particular, loss of PTEN or INPP4B promotes tumor growth in breast and prostate cancer (Rodgers et al., 2017). Both enzymes have been reported to be mutated or deleted in breast cancer (Fedele et al., 2010; Kofuji et al., 2015). Multiple isoforms of inositol polyphosphate 5-phosphatase INPP5J, synaptojanin 2, INPP5K and the SH2 containing inositol 5-phosphatases SHIP1 and SHIP2 (also known as INPP5D and INPPL1, respectively) have been shown to behave either as tumor promotors or tumor suppressors (Erneux et al., 2016). Moreover, both in cell lines and in vivo in mice, synaptojanin 2 and SHIP1 appear to be druggable targets (Ben-Chetrit et al., 2015; Gumbleton et al., 2017). Both phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P3] and phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P2] are often reported as substrates of the substrates of the phosphoinositide (PI) 5-phosphatases OCRL, SHIP2 and INPP5K (Balla, 2013). This is in contrast to PTEN, which can use both PI(3,4,5)P3 and PI(3,4)P2 as substrate acting thus as PI 3-phosphatases (Malek et al., 2017).
Elevated SHIP2 expression is observed in very aggressive cancers, such as triple negative breast cancer and colorectal cancer (Fu et al., 2014; Hoekstra et al., 2016; Prasad et al., 2008b). SHIP2 is actually both a scaffold protein for multiple cytoskeletal proteins and a PI phosphatase that preferentially uses PI(3,4,5)P3 as substrate (Blero et al., 2005; Erneux et al., 2016). It is therefore producing the important lipid PI(3,4)P2, which can interact with specific proteins, such as lamellipodin (Krause et al., 2004), Tks5 at the invadopodium (Sharma et al., 2013) and the BAR domain protein SNX9 at late-stage endocytic intermediates (Ketel et al., 2016). Recent data obtained in the breast cancer MDA-MB-231 cell model identified SHIP2 as a modulator of carcinoma invasiveness and a potential target for metastatic disease (Rajadurai et al., 2016). The question remains as to whether SHIP2 is acting by producing a key lipid, such as PI(3,4)P2, by reducing PI(4,5)P2 or through its scaffold properties. SHIP2 is also important in the control of cell migration and adhesion, as has been observed in both normal and different cancer cell types, such as keratinocytes (Yu et al., 2010), glioblastoma cells (Elong Edimo et al., 2016a) and colorectal cancer cells (Hoekstra et al., 2016). In glioblastoma 1321 N1 cells, SHIP2 was reported to control the content of PI(4,5)P2 and to affect focal adhesion dynamics (Elong Edimo et al., 2016a,b). Whether this could also occur in breast cancer cells is unknown. It is also possible that both PI(4,5)P2 and PI(3,4)P2 are affected in cancer cells and control cell migration or invasion by different mechanisms.
The INPPL1 gene, which encodes SHIP2, has been found to be mutated in opsismodysplasia (OPS) a rare autosomal recessive disease characterized by growth plate defects and delayed bone maturation (Below et al., 2013; Huber et al., 2013). Interestingly, the SHIP2 inhibitor AS1949490 (Suwa et al., 2009) has been shown to inhibit migration in fibroblasts that express SHIP2 but had no effect in fibroblasts from OPS patients (Ghosh et al., 2017). The data therefore suggest the inhibitor could be an important tool to probe SHIP2 function in cells or in vivo, at least for migration.
SHIP2 was initially presented as a tumor suppressor due to its ability to dephosphorylate PI(3,4,5)P3 although this may be depending on the cell context (Ye et al., 2016). Here, we aimed to address whether we could inhibit cell migration in breast cancer cells by the use of SHIP2 inhibitors, and if this result could be generalized to different breast cancer cells. We also questioned whether this could be seen in other cell lines, such as rat chondrosarcoma cells [RCS; chosen as cell model of OPS as those cells maintain a chondrocyte differentiation status (Mukhopadhyay et al., 1995)]. In addition, because the level of PI(3,4)P2 appears to be lower at focal adhesions in SHIP2-depleted MDA-MB-231 cells than in control cells (Fukumoto et al., 2017), we assessed whether we could observe a direct and specific effect of this lipid on focal adhesions and cell migration. Finally, as SHIP2 has been reported to promote tumor formation and metastasis in MDA-MB-231 cells (Prasad et al., 2008a), we determined whether we could confirm and interpret this result by the use of SHIP2-positive and SHIP2-depleted cells prepared in our hands.
The influence of SHIP2 on cell migration in breast cancer cells: MDA-MB-231 and MDA-MB-468 cells
Previous studies have reported that the SHIP2 inhibitor AS1949490 can be used as a tool to target SHIP2 catalytic activity in human fibroblasts (Ghosh et al., 2017). It has been shown that treatment with this molecule at 10 µM inhibits cell migration in SHIP2-containing fibroblasts and, as expected, it has no effect on fibroblasts derived from OPS patients that do not express SHIP2 (Ghosh et al., 2017). Here, we asked whether this result could be generalized to different breast cancer cells and, hence, cell migration of basal and luminal breast cancer cells was tested in the presence of AS1949490 either at 5 or 10 µM. Western blotting demonstrated the presence of SHIP2 in all cells at its expected molecular mass of 155 kDa (Fig. 1A). In the presence of AS1949490, cell migration velocity was inhibited in MDA-MB-231, MCF7, T47D, BT549 and HCC38 cells as compared to control (DMSO). It was increased in MDA-MB-468 and BT20 cells (Fig. 1B).
SHIP2-deficient MDA-MB-231 cells were made through shRNA lentiviral transduction. The SHIP2-depleted cells are referred to as shSHIP2(1) or shSHIP2(2) cells and compared to cells transduced with scrambled shRNA (shScramble cells). As we could not prepare shSHIP2 MDA-MB-468 cells, we depleted SHIP2 by siRNA transfection in this model. SHIP2 expression was decreased in the two breast cancer models (as determined by western blotting, Fig. 2A). SHIP2-depleted cells showed a decrease in cell migration in MDA-MB-231 cells [shSHIP2(2) cells] as compared to control shScramble cells (Fig. 2B). This result was confirmed in cells with SHIP2 siRNA transfection (Fig. 2B). Most of our experiments were undertaken with shSHIP2(2) cells, which gave a significant and reproducible reduction in SHIP2 expression (data not shown). In contrast to MDA-MB-231 cells, siRNA-transfected MDA-MB-468 cells showed an increase in cell migration as compared to control cells (Fig. 2B). This is in agreement with the data obtained when adding the SHIP2 inhibitor AS1949490 to the same cells (Fig. 1B).
MDA-MB-468 cells have been used as an EGF-inducible epithelial-mesenchymal transition (EMT) model (Turtoi et al., 2013). When cells were allowed to grow in the presence of EGF for 72 h, migration was increased to the same extent as with the SHIP2 inhibitor AS1949490 (Fig. 2D). Cells were more elongated and polarized in the presence of the SHIP2 inhibitor added for 16 h (Fig. S1A), and SHIP2 staining at the cell periphery was increased in cells incubated with AS1949490 (Fig. S1B). Moreover, EGF-treated cells were also more elongated than control MDA-MB-468 cells (Fig. S1C). We asked whether AS1949490 had an impact on cell migration after EGF treatment of MDA-MB-468 cells. AS1949490 does not activate, but rather inhibits, migration of cells treated with EGF for 3 days. This response to the inhibitor in MDA-MB-468 is comparable to the effect of AS1949490 in MDA-MB-231 cells (Fig. 2C,D). We conclude that the inhibitory effect of AS1949490 on MDA-MB-468 cell motility actually depends on whether or not cells have been treated with EGF.
Specificity of AS1949490 as a SHIP2 inhibitor in MDA-MB-231 and RCS cells
We aimed to compare MDA-MB-231 shSHIP2(2) and shScramble cells to assess whether specific inhibition of SHIP2 by AS1949490 is responsible for the effects on cell migration. As expected, the inhibitor at 10 µM or 5 µM did not affect cell migration of SHIP2-depleted cells in contrast to what was seen in SHIP2-containing cells (Fig. S2A). This result was confirmed when using another SHIP2 inhibitor, K149, which has a different chemical structure, as previously reported in a study of SHIP2 in colorectal cancer (Hoekstra et al., 2016). When tested in our cells, K149 had the same effect as AS1949490 on cell migration (Fig. S2B), confirming our results with AS1949490. Finally, as SHIP2 plays a role in chondrocyte maturation (Fradet and Fitzgerald, 2017), we used RCS cells in which SHIP2 was deleted by CRISPR/Cas9 (SHIP2Crispr cells) as a model. Cells plated on fibronectin and in 10% serum showed a decrease in cell migration as compared to control cells (Fig. S3A). AS1949490 did not affect cell migration in two independent RCS SHIP2Crispr cell lines (SHIP2c−/− and SHIP2g−/−). The absence of SHIP2 in SHIP2Crispr cells was verified by western blotting (Fig. S3B).
Taken together, we confirmed the specificity of AS1949490 as an inhibitor of SHIP2, and that treatment with this compound affects the control of cell migration in MDA-MB-231 and other cells. Moreover, the migration data obtained upon depleting SHIP2 expression in breast cancer cells are in agreement with the inhibitor data, suggesting that SHIP2 activity and/or expression control cell migration in MDA-MB-231 cells as well as in other breast cancer cells.
Influence of PI3K inhibitors and Akti on cell migration
SHIP2 preferred substrate is PI(3,4,5)P3, which leads to the generation of PI(3,4)P2 (Giuriato et al., 2002). PI(3,4)P2 can be dephosphorylated by PI 4- and 3-phosphatase activities activities (Hawkins and Stephens, 2016). This pathway is controlled by class I or class II PI3K activities. Migration of MDA-MB-231 cells has been shown to be dependent on Akt proteins (Ooms et al., 2015). We therefore tested the effect of the PI3K inhibitor LY294002 and Akt inhibitor (Akti) on cell migration. LY294002 and Akti inhibited migration in MDA-MB-231 cells. LY294002 had no effect in MDA-MB-468 cells (untreated with EGF) whereas Akti stimulated cell migration velocity (Fig. 2C,D). In EGF-treated cells, cell migration velocity was decreased in the presence of Akti and LY294002, suggesting that migration is both PI3K and PKB dependent in those conditions (Fig. 2D). The data with LY294002 in MDA-MB-231 and MDA-MB-468 cells were confirmed with the class I specific PI3K inhibitor PI-103 (as shown in Fig. S3C).
PI(3,4)P2 content in shSHIP2 MDA-MB-231 cells as compared to control cells
Given the specific binding of PI(3,4)P2 to lamellipodin (Krause and Gautreau, 2014), which controls lamellipodium formation, we aimed to quantify the level of PI(3,4)P2 at the cell periphery in SHIP2-depleted MDA-MB-231 cells as compared to control cells. In SHIP2-depleted MDA-MB-231 cells stimulated with 10% serum for 5 min, PI(3,4)P2 was decreased as compared to control cells (Fig. 3). This was shown with two detection methods: immunostaining with an anti-PI(3,4)P2 antibody and by the use of GFP–PH-TAPP1 as a biosensor for PI(3,4)P2 (Fig. 3A–C). PI(3,4)P2 immunoreactivity at the cell periphery was inhibited in the presence of the PI3K inhibitor LY294002 (data not shown). The decrease in PI(3,4)P2 between the two cell types was also observed in cells maintained in 10% serum (Fig. S4A,B). The levels of PI(4,5)P2 were not modified, as determined by comparing the two types of cells (Fig. S4C,D) and PI(3,4,5)P3 was slightly upregulated [i.e. 26% in shSHIP2 MDA-MB-231 as compared to control cells (Fig. S4E,F)]. AS1949490 also inhibited PI(3,4)P2 in MDA-MB-231 control cells, as determined by immunostaining or by the use of GFP–PH-TAPP1 as a biosensor (Fig. 3D–F). We conclude that PI(3,4)P2 is decreased both in SHIP2-depleted MDA-MB-231 cells and in cells treated with AS1949490.
Lamellipodin binds to Ena/VASP proteins (Carmona et al., 2016) and controls actin and lamellipodial dynamics (Li and Marshall, 2015): phalloidin staining was always increased in SHIP2-depleted cells as compared to control cells (Fig. S5A). MDA-MB-231 shSHIP2(2) cells were also less polarized as compared to shScramble cells, as shown by phalloidin staining (Fig. S5B).
The influence of bio-specific probes for phosphoinositides on cell migration in breast cancer cells
The use of protein domains specific for binding to different phosphoinositides in competition experiments to prevent cellular lipid activity has been reported (Davies et al., 2014; Ooms et al., 2015) and was undertaken in our study in 1321 N1 cells (Elong Edimo et al., 2016a). We asked whether migration in MDA-MB-231 cells would be affected when blocking PI(3,4)P2 through overexpression of GFP–PH-TAPP1, a specific probe for this lipid. The same approach was also used with other probes: GFP–PH-Btk, specific for PI(3,4,5)P3, and GFP–PH-PLCδ1, specific for PI(4,5)P2. Cell migration velocities were compared upon transfecting GFP–PH-TAPP1, GFP–PH-Btk and GFP–PH-PLCδ1 to prevent PI function (Fig. 4A). In MDA-MB-231 cells, the most efficient probe that inhibited cell migration was GFP–PH-TAPP1, further suggesting that PI(3,4)P2 is involved in the control of cell migration. We confirmed this result by adding a GFP-PH mutant of TAPP1 that is no longer able to interact with PI(3,4)P2, namely GFP-PH-TAPP1-R211/L (Marshall et al., 2002); only wild-type GFP–PH-TAPP1 inhibited cell migration whereas the TAPP1 mutant did not have an effect in this assay (Fig. 4B). The same comparison was made by measuring cell migration in MDA-MB-468 cells that were not treated with EGF. In this model, the most potent probe that inhibited cell migration was GFP–PH-PLCδ1 (Fig. 4C). The PLCδ1 mutant GFP–PH-PLCδ1-R40L (Várnai and Balla, 1998) had no effect on cell migration.
Cell migration in SHIP2-depleted MDA-MB-231 cells can be increased by direct addition of diC8 PI(3,4)P2
Data reported by others indicate that a short chain PI, that is diC8 PI5P, can be directly added to cells to probe a signaling function (Viaud et al., 2014). We followed the same approach and directly added diC8 PI(3,4)P2 to MDA-MB-231 cells deficient for SHIP2. This was carried out after a pre-incubation of diC8 PI(3,4)P2 of 6 h to facilitate penetrance of the lipid. Migration was increased by adding the lipid at 5 or 10 µM. diC8 PI(3,5)P2 had no effect at the same concentrations in similar experiments (Fig. 5A). We noticed that lamellipodia formation was always decreased in SHIP2-depleted cells as compared to control or shScramble cells (Fig. 5B). This was also shown in shScramble cells in the presence of SHIP2 inhibitor AS1949490. In our experiments adding lipids, we also found that lamellipodia formation was increased by the addition of diC8 PI(3,4)P2 to the same cells. This was not seen with the isomer PI(3,5)P2 (Fig. 5B). We conclude that, in MDA-MB-231 cells, PI(3,4)P2 can directly control cell migration.
Akt activity in shSHIP2 MDA-MB-231 cells as compared to control cells
PI(3,4)P2 can bind to Akt with an affinity which is similar to that of PI(3,4,5)P3 (Franke et al., 1997). PI(3,4)P2 can be dephosphorylated by PI 4- and 3-phosphatase (INPP4A/B and PTEN) (Malek et al., 2017). Therefore the influence of SHIP2 on Akt activity depends on cell context. We previously reported that Akt1 phosphorylated on S473 (pAkt S473) was upregulated in SHIP2-deficient mouse fibroblasts as compared to control cells (Blero et al., 2005). This was shown after serum stimulation for 5–10 min. We repeated the same type of experiments in MDA-MB-231 cells upon a short time EGF stimulation from 5–60 min (Fig. 6A). Levels of pAkt S473 normalized to total Akt, were upregulated in EGF-stimulated cells but only after 5 min EGF stimulation (Fig. 6B). In the same western blot, we confirmed the decrease in SHIP2 expression by 80% in shSHIP2(2) cells (Fig. 6A). The levels of phosphorylated ERK protein (ERK1/2) were not significantly modified between the two different cell types at the same time points (data not shown). In contrast, when cells were maintained in 10% serum, pAkt S473 was decreased in SHIP2-depleted cells as compared to control cells (Fig. 6C).
Tumor-promoting role of SHIP2 in MDA-MB-231 xenograft SCID mice
Data reported by Prasad et al. revealed an oncogenic role of SHIP2 through xenografting SHIP2-depleted MDA-MB-231 cells (Prasad et al., 2008a). The authors explained their data by showing that there was an increase in EGF receptor internalization in SHIP2-depleted cells compared to control cells. In contrast to our data (Fig. 6A), SHIP2 depletion decreased EGF-induced Akt phosphorylation in this model (Prasad et al., 2008a).
We confirmed the tumor-promoting role of SHIP2 in xenografts of 106 shScramble cells and 106 shSHIP2(2) cells by measuring both tumor volume and tumor weight (Fig. 7A–C). EGF receptor expression was, however, unchanged between shSHIP2 and shScramble cells (Fig. 6A). The number of focal adhesions per cell, estimated by determining the immunoreactivity of phosphorylated FAK (pFAK; FAK is also known as PTK2), was also increased in shSHIP2 cells as compared to control cells both before and after xenografting (Fig. 7D–F). SHIP2 inhibitors AS1949904 and K149 had no effect on the number of focal adhesions per cell (Fig. 7G). diC8 PI(3,4)P2 also had no impact on the number of focal adhesions for both MDA-MB-231 shSHIP2(2) and shScramble cells (Fig. 7H).
Metastasis in lungs of xenograft mice was assessed by staining for human vimentin. The staining of vimentin was decreased 6 weeks after grafting animals with shSHIP2(2) cells as compared to control cells (Fig. 8A,B). This was confirmed in SCID mice tail vain-injected with GFP-expressing shSHIP2(2) cells: lowering SHIP2 expression decreased the appearance of metastasis. After 4 weeks, lung sections were analyzed. A decrease in GFP staining of mice injected with GFP-expressing shSHIP2(2) cells as compared to control-injected mice was observed (Fig. S6A,B). We conclude that lowering SHIP2 expression decreases the appearance of metastasis. In an invadopodium degradation assay, we observed that the lowered SHIP2 expression in shSHIP2(2) cells provoked an important decrease in matrix degradation as compared to shScramble cells (Fig. S7). Cell invasion was also measured in these cells through the use of invasion chambers coated with Matrigel matrix. Invasion was very much decreased in shSHIP2(2) MDA-MB-231 cells as compared to shScramble cells. AS1949490 at 5 or 10 µM inhibited invasion of shScramble cells but had no effect on shSHIP2(2) cells (Fig. 8C).
SHIP2 depletion in MDA-MB-231 increased cell adhesion and spreading on fibronectin
Cell adhesion on fibronectin is increased in OPS-derived fibroblasts as compared to SHIP2-containing control fibroblasts (Ghosh et al., 2017). Increased adhesion was also shown both in MDA-MB-231 shSHIP2 cells as compared to shScramble cells and in SHIP2Crispr cells as compared to wild-type RCS cells (Fig. 8D,E). These data were obtained on 96-well plates coated with fibronectin as described in the Materials and Methods. Cell spreading on fibronectin was also enhanced in shSHIP2(2) MDA-MB-231 cells as compared to shScramble cells. Interestingly, there was a higher level of cell death for shSHIP2 MDA-MB-231 cells than for shScramble MDA-MB-231 cells (Fig. S8A). FACS analysis of activated caspase 3 confirmed that shSHIP2(2) cells showed a higher level of apoptosis than shScramble cells both before and after xenografting (i.e. in tumor-derived cells) (Fig. S8B,C).
In this study, evidence is provided that, in MDA-MB-231 breast cancer cells, SHIP2 depletion decreases cell migration and increases the adherence to fibronectin. A similar effect on migration in MDA-MB-231 cells was obtained by inhibiting SHIP2 activity with two unrelated SHIP2 phosphatase inhibitors which, as expected, had no effect in SHIP2-depleted cells. In MDA-MB-231 cells, the amount of PI(3,4)P2, the major SHIP2 reaction product, is decreased both in SHIP2-depleted cells and in cells incubated in the presence of AS1949490. We show for the first time that diC8 PI(3,4)P2, added to SHIP2-depleted MDA-MB-231 cells, increased cell migration, an effect which is specific for this isomer. We thus establish that in some breast cancer cells, particularly in MDA-MB-231, PI(3,4)P2 plays a positive second messenger role in the control of cell migration through promoting lamellipodia formation. Consistent with our data, it has been shown that the actin-binding protein profilin 1 regulates PI(3,4)P2 and lamellipodin accumulation at the leading edge of MDA-MB-231 cells and influences their motility (Bae et al., 2010). Although our migration data are in agreement with the data of Prasad, we could not confirm an alteration of EGF receptor internalization upon SHIP2 silencing (Prasad, 2009). The reason for this is not understood, although it could be related to clonal differences in SHIP2 depletion/silencing between the two studies.
In 1321 N1 glioblastoma cells, SHIP2-depleted cells migrate faster, concomitant with PI(4,5)P2 upregulation and increasing pFAK activities (Elong Edimo et al., 2016a). We measured PI(4,5)P2 at the plasma membrane in MDA-MB-231 cells and no significant differences between control and shSHIP2 MDA-MB-231 cells was detected. This was in contrast to what was seen for PI(3,4)P2; in MDA-MB-231 cells, PI(3,4)P2 at the cell periphery was reduced in shSHIP2 cells as compared to control cells, in agreement with a recent report where SHIP2 siRNA had been used (Fukumoto et al., 2017).
To further support a role of PI(3,4)P2, we overexpressed a specific probe for either PI(4,5)P2 or PI(3,4)P2 to see whether migration velocity was affected in MDA-MB-231 cells. We observed a minor decrease in cells expressing GFP–PH-PLCδ1 but a very significant effect in cells expressing GFP–PH-TAPP1. Therefore, PI(3,4)P2 plays a major role in the control of cell migration in this breast cancer cell line. It is likely that the absence of INPP4B in this model (Fedele et al., 2010) favors the accumulation of PI(3,4)P2 at the plasma membrane. INPP4B expression is lost in a cohort of basal-like breast cancers and its reduced expression is associated with worse survival. Therefore, SHIP2-produced PI(3,4)P2 could play an important function in cancer cells where INPP4B is either deleted or mutated. Moreover, in MDA-MB-231 cells, PI(3,4)P2 accumulates at nascent invadopodia, which leads to the recruitment of effector proteins, such as Tks4/Tks5, which are important for invadopodia maturation (Sharma et al., 2013). SHIP2 inhibition with AS1949490 leads to a 50% decrease in Matrigel invasion of cells (Sharma et al., 2013). In the same cells, we showed a 50% decrease in PI(3,4)P2 after incubation of shScramble cells with the inhibitor and confirmed the decrease in invasion both in shSHIP2 cells as compared to control cells and by adding SHIP2 inhibitors to MDA-MB-231 cells.
Several protein interactors of SHIP2, such as vinexin, filamin and Cbl (Dyson et al., 2001; Paternotte et al., 2005; Vandenbroere et al., 2003), and SHIP2 itself are members of the integrin adhesome, and could generate dynamic complexes of proteins, such as the integrin adhesion complex (Horton et al., 2016). This will have an impact on the actin cytoskeleton. Phalloidin staining has been reported to be increased in glioblastoma 1321 N1 cells upon SHIP2 depletion or in SHIP2−/− mouse kidney sections as compared to wild type (Elong Edimo et al., 2016a,b; Sayyed et al., 2017). An increase in phalloidin staining was also shown in our study in SHIP2-depleted MDA-MB-231 cells (Fig. S5A). The number of focal adhesions per cell was also increased in SHIP2-depleted cells and in tumor-derived cells in xenograft shSHIP2 cells (Fig. 7D–F). Importantly, we could not detect any change in pFAK staining in cells pre-incubated with a SHIP2 inhibitor or upon the addition of PI(3,4)P2 to cells. Therefore, in addition to changes in PI(3,4)P2 content, which influences cell migration, it is likely that SHIP2 depletion affects protein complexes at the adhesome, which promotes focal adhesions through a phosphatase-independent mechanism.
MDA-MB-468 cells had a far lower level of cell migration as compared to MDA-MB-231 cells as also observed by others (Juvin et al., 2013). Indeed, MDA-MB-468 cells migrated, but very slowly as compared to MDA-MB-231 cells. SHIP2 inhibition with AS1949490 increased the cell migration velocity in MDA-MB-468 cells not treated with EGF but inhibited cell migration of cells treated with EGF for 3 days. Therefore the inhibitory effect of AS1949490 on MDA-MB-468 cell migration actually depends on the EMT status of the cells, which influences the morphology of the cells and their ability to form lamellipodia. This generalizes the observation we made in MDA-MB-231 cells about the role of SHIP2 in cell migration in breast cancer cells.
The tumor volume was decreased in xenograft MDA-MB-231 SHIP2-depleted cells as compared to control cells supporting a tumor-promoting role for SHIP2. The data are in agreement with previous observations made by others (Fu et al., 2014; Prasad et al., 2008a).
In serum-containing medium, we observed a decrease in pAkt S473 in MDA-MB-231 shSHIP2 cells as compared to control cells. This result is comparable to pAkt S473 data reported in colorectal cancer cells upon SHIP2 depletion (Hoekstra et al., 2016). These data suggested that SHIP2 had an oncogenic role in colorectal cancer through enhancement of cell migration and cell invasion. This is similar to our data in MDA-MB-231 cells.
Depletion of SHIP2 in cells or the addition of a SHIP2 inhibitor resulted in a decrease in living cells. SHIP2-depleted cells showed a higher level of apoptosis than control cells and this was observed both in cells in culture (shSHIP2 cells as compared to shScramble cells) and, importantly, in tumor-derived cells obtained after 6 weeks in vivo. A possible link between cell survival and SHIP2 has been reported in a different cell context in CDAAP-deficient podocytes where inhibition of SHIP2 by AS1949490 reduced oxidative stress, but resulted in a deleterious increase in apoptosis (Saurus et al., 2017). It will be interesting to compare the effect of SHIP2 inhibitor(s) on apoptosis of different cancer cells as this could explain the fact that SHIP2 is oncogenic in breast cancer cells but a tumor suppressor in some glioblastoma cells (Elong Edimo et al., 2011; Taylor et al., 2000) and in squamous cell carcinoma (Yu et al., 2008).
We previously reported that cell migration was very much decreased in fibroblasts derived from OPS patients as compared to control individuals, supporting a role of SHIP2 in cell migration (Ghosh et al., 2017). In contrast, cell adhesion on fibronectin was increased in OPS fibroblasts, as also shown by the data presented for MDA-MB-231 cells. In RCS chondrocytes, SHIP2 was identified to be part of the fibroblast growth factor receptor 3 (FGFR3) interactome in a complex of proteins also containing proteins of the adhesome (Balek et al., 2018). SHIP2-deleted RCS cell lines have been generated by gene editing using CRISPR/Cas9 (P.K., unpublished data). We confirmed in this model the decrease in migration in two SHIP2-deleted RCS clones and the increase in adhesion on fibronectin. Importantly, as expected, in SHIP2-deleted RCS cells, SHIP2 inhibitor had no effect on cell migration. The inhibition of cell migration by AS1949490 was observed in wild-type RCS, MDA-MB-231, MDA-MB-468 (treated with EGF for 3 days), MCF7, T47D, HCC38 and BT549 cells (Fig. 1B) and also in patient-derived xenografts (PDX) of breast cancer (S.G., unpublished data). It will be important to establish whether or not PI(3,4)P2 is decreased in all these cells.
In conclusion, we have shown that cell migration in glioblastoma, human fibroblasts and breast cancer cells is very much controlled by SHIP2 activity and expression. We have validated that SHIP2 inhibitors can be used as tools to rapidly test for an impact on cell migration in any cell type or PDX. It will be interesting to use those inhibitors in migration assays in mice models in vivo. As SHIP2 plays an important and very specific role in OPS, it will be important to ask whether migration or adhesion on fibronectin is affected in the presence of FGF or IGF1 in the RCS chondrocyte model.
MATERIALS AND METHODS
Anti-SHIP2 monoclonal antibody used for total SHIP2 immunostaining was from Novus (Cambridge, UK; catalog number H00003636-M01) and used at a dilution of 1:3500 in MDA-MB-231 cells. LY294002 hydrochloride (catalog number L9908), unconjugated gelatin from porcine skin (catalog number G-2500), extra cellular matrix (catalog number E1270), fibronectin from bovine plasma (catalog number F1141), collagenase I (catalog number C0130) and hyaluronidase (catalog number H3884) were from Sigma-Aldrich (Diegem, Belgium). Polyclonal pFAK Y397 antibody (catalog number 3283) was from Cell Signaling Technology (Bioke, The Netherlands; used at 1:100 for immunofluorescence). Antibodies against pAkt S473 (catalog number, 9271L), pAkt T308 (catalog number, 9275L), total Akt (catalog number, 9272S), pErk1/2 Thr202/Tyr204 (catalog number 9106S) were from Cell Signaling Technology and were used at 1:500. Anti-Erk2 antibody (sc-154) was from Santa Cruz Biotechnology (Dallas, TX) and was used at 1:500. Antibodies to PI(4,5)P2 (2C11) (catalog number sc-53412, used at 1:200) and to PI(3,4)P2 (catalog number 117Z-P034b, used at 1:150) were from Santa Cruz Biotechnology and Echelon (Le-Perray-en-Yvelines, France), respectively. Lipofectamine™ (catalog number 11668-019), Lipofectamine™ 3000 (catalog number L3000008), PI-103 hydrochloride (catalog number inh-pi10), Alexa-Fluor-488-conjugated donkey (catalog number A21206) and Alexa-Fluor-594-conjugated goat (catalog number A11012) anti-rabbit-IgG, Alexa-Fluor-488-conjugated donkey (catalog number A21202) and Alexa-Fluor-594-conjugated goat (catalog number A11032) anti-mouse-IgG, Alexa Fluor 488–gelatin (catalog number G13187) and normal goat serum (catalog number 16210-064) were from Invitrogen (Breda, The Netherlands). The Glycergel mounting medium (catalog number C0563) was from Dako (Heverlee, Belgium). Acti-stain 555 (phalloidin) (catalog number PHDH1) was from Cytoskeleton (Denver, CO). GFP–PH-PLCδ1, GFP–PH-Btk and GFP–PH-TAPP1 were generously provided by Dr Tamas Balla (NIH, USA). The pSico vector expressing EGFP was kindly provided by Dr Sabrina Ena (Laboratory of Neurophysiology, ULB) and was used to prepare GFP-positive cells required for the study of metastasis. The GFP–C1-TAPP1-R211/L construct was from MRC-PPU Reagents (University of Dundee, UK). GFP–PH-PLCδ1-R40L was from Dr Tamas Balla. SHIP2 inhibitor AS1949490 (catalog number 3718) was from Tocris and K149 (catalog number B-0345) was kindly provided by Dr Colin Ferguson (Echelon Inc.). Four-well and eight-well µ-slides, ibiTreat: #1.5 polymer coverslip, tissue culture treated, sterilized (catalog number 80426) were purchased from Ibidi.
Human breast cancer cell lines (MDA-MB-231, MCF7 and MDA-MB-468 cells) were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco, CA), supplemented with 10% fetal calf serum (FCS), 1% sodium pyruvate, 1% non-essential amino acids and 2% penicillin-streptomycin. The cell lines T47D, BT20, BT549, HCC1937 and HCC38 were kindly provided by Dr Pierre Roger (IRIBHM, ULB) and are described in Raspé et al. (2017). They were authenticated by Eurofins Medigenomics and regularly checked to be mycoplasma free. The cells were cultured in RPMI medium supplemented with 10% FCS, 1% sodium pyruvate and 2% penicillin-streptomycin. RCS WT, RCS SHIP2c−/− and SHIP2g−/− cells (P.K., unpublished data) were cultured in DMEM supplemented with 10% FCS and 2% penicillin-streptomycin. All the cells were maintained at 37°C under 5% CO2. Cells were regularly checked to be free of mycoplasma. MDA-MB-231 cells were transduced with Sigma MISSION lentiviral transduction particles containing a pLK0.1-Puro plasmid encoding either a non-mammalian shRNA (Scramble SHC016) or one of two SHIP2-specific targeting shRNAs [reference number TRCN 0000052810 and TRCN 0000052808 for shSHIP2(1) and shSHIP2(2), respectively] at a multiplicity of infection of 2.8. This was performed overnight at 37°C in growth medium supplemented with 8 μg/ml of polybrene. Transduced cells were selected in growth medium supplemented with 8 μg/ml puromycin for 3 weeks then subsequently maintained in growth medium without puromycin. Cells were rapidly frozen after three passages and used between passage 3 and 12.
Transfection of cells with SHIP2-specific siRNA
Cells were transfected with siRNA targeting INPPL1 (catalog number HS S105471 and HS S179945 from Invitrogen) and negative control low GC duplex using the Lipofectamine RNAi Max or Lipofectamine 3000 for MDA-MB-468 cells according to the manufacturer's instructions. siRNA reagents were chosen based on a previous report (Awad et al., 2013).
Cell viability assay
The proportion of viable cells with or without any treatment was determined using the WST1 reagent (catalog number 11644807001, Roche, Vilvorde, Belgium) according to the manufacturer's instructions. Briefly 25,000 cells were plated in 96-well plates, in triplicates for each condition, and then the required treatment was undertaken for the indicated times in 100 µl of medium. At the end of the treatment, 10 µl of the WST1 reagent was added to 100 µl of medium directly on each well and cells were incubated at 37°C for 2 h. The optical density was determined at 460 nm and compared to a blank made without cells.
Breast cancer cell xenografting and preparation of primary cells
All procedures involving mice for xenograft tumor growth assays were approved by the Animal Care and Ethics Committee of the Université Libre de Bruxelles. 8-week old female, athymic immunodeficient NOD/SCID mice were purchased from Charles River Laboratories (France). 106 cells were suspended in 50 µl DMEM+50% growth factor-reduced extracellular matrix (catalog number E1270, Sigma) and injected subcutaneously in 12 mice per cell line. At 6 weeks post injection mice were killed, and lungs were harvested and fixed in paraffin for further histological analysis. Sections of these lungs were stained for human vimentin (Dako; catalog number M0725, used at 1:100) to identify lung metastasis. The weight of the tumors was determined and volume was measured using the formula length×(width)2/2. For preparation of primary cells, tumors were washed with HBSS (catalog number 14025092, Thermo Fisher Scientific) and chopped into small pieces. The chopped tumors were incubated in HBSS containing 300 U/ml collagenase and 300 mg/ml of hyaluronidase for 2 h at 37°C with constant agitation. A P1000 pipette was used to physically dissociate the tumor every 15 min throughout the enzymatic digestion. The reaction was stopped using EDTA at a final concentration of 5 mM. The tissue suspension was filtered through 70 µm mesh, pelleted (200 g for 10 min) and washed with PBS, plated with DMEM and allowed to grow at 37°C.
To study metastasis, MDA-MB-231 shScramble and shSHIP2(2) cells were labeled with GFP through a lentiviral-mediated pSico vector. 1.5×105 GFP-positive cells were injected into the tail vein of 8-week-old NOD/SCID mice. At the end of fourth week, the mice were killed and the lungs fixed. Cryosections of the lungs were stained with an anti-GFP antibody (1:200, catalog number A11122, Invitrogen) to identify the metastasis in lungs. Preparations were observed on an Axiozoom V16 microscope (Zeiss, Oberkochen, Germany) equipped with PlanNeoFluar Z 1×/0.25 N.A. objective and an optical zoom to give an actual magnification of 200× as indicated in the figure legend. The percentage of GFP-positive (metastatic) area for the lungs was quantified for each mouse using ImageJ software.
Cells were lysed in buffer A (10 mM Tris-HCl pH 7.5, 150 mM KCl, 12 mM 2-mercaptoethanol, 100 mM NaF, 0.5% NP-40, 0.01 μM sodium vanadate and 2 mM EDTA) and protease inhibitors (Roche). The cells were scraped, and the lysates were cleared by centrifugation at 14,000 g. The lysates were normalized based on total protein content using the Bradford assay (Bio-Rad). Samples were loaded on 8% polyacrylamide gels and run for 1.5 h. Proteins were transferred onto nitrocellulose membranes, blocked in Odyssey blocking buffer (LI-COR Biosciences), and incubated with primary antibodies for 2 h at room temperature. After three washes in PBS with Odyssey blocking buffer, membranes were incubated with DyLight 680 anti-rabbit- and anti-mouse-IgG or DyLight 800 anti-rabbit- and anti-mouse-IgG (Thermo Fisher Scientific) for 1 h at room temperature, and then washed three times in PBS with 0.1% Tween 20 (Bio-Rad). Membranes were scanned using a high-sensitivity Odyssey Infrared Imaging System (LI-COR Biosciences) as reported previously (Elong Edimo et al., 2016a).
Migration, adhesion, spreading and invasion assays
Cells were plated on four- or eight-chambered μ-slide wells and allowed to migrate for 15 h as reported previously (Elong Edimo et al., 2016a). They were analyzed with a Leica DM6000B microscope, using a 10× magnification objective, for live-cell imaging. Each cell was tracked over the period of time using the manual tracking plugin from ImageJ software. To determine cell migration velocity, cells were tracked over time using chemotaxix and migration software from Ibidi and plotted using Graphpad Prism. At least 50 cells were analyzed for each condition and each experiment was repeated at least twice.
The formation of lamellipodia was assessed from the videos acquired for migration assay. Each field was observed for a period of 1 h to note the number of cells per field that formed lamellipodia. This was undertaken for at least five fields per condition and repeated over at least three independent experiments.
For cell adhesion determination, breast cancer cells were plated on fibronectin coated 96-well plates for 30 min. Cells were washed five times with PBS. They were fixed using 96% ethanol for 10 min and stained with 0.1% Crystal Violet (catalog number C3886, Sigma) for 30 min at room temperature. Each well was washed twice and permeabilized using 0.2% Triton X-100. Absorbance was read at 595 nm. Each condition was run in triplicate. We verified that a 15 or 30 min incubation was optimal to observe a difference in cell adhesion. When an effect was observed, it was always confirmed by using different amounts of cells (25,000 or 50,000 cells per well). For the spreading assay, cells were plated on fibronectin-coated coverslips for 30 min and then fixed using 4% paraformaldehyde followed by immunofluorescence using the required primary antibodies.
Invadopodium degradation assay
This assay was performed as reported in Sharma et al. (2013). Briefly, Alexa Fluor 488-conjugated gelatin was coated on coverslips. Cells were plated on labeled gelatin matrix for 24 h. Invadopodium precursors were identified as cortactin- and phalloidin-positive puncta without a degradation hole. Mature invadopodia were identified as cortactin- and phalloidin-positive puncta colocalizing with a degradation hole. Confocal imaging was realized on a Zeiss LSM-710 confocal microscope using a Zeiss ×63/1.4 Plan Apochromat objective and with specific excitation using a 488 nm argon ion laser, a 594 nm helium/neon laser and a 633 nm laser diode. A specific gallium arsenide phosphide (GaAsP) Airyscan detector was used to increase signal detection, the signal:noise ratio and resolution. Images were acquired using Zeiss ZenBlack software (Zeiss, Oberkochen, Germany).
Matrigel invasion assay
Cell invasion was determined using Corning Matrigel invasion chambers (catalog number 354480) according to manufacturer's protocol. In brief, MDA-MB-231 shScramble or shSHIP2 cells were washed with DMEM, and 25,000 cells were plated on the upper chambers of the coated transwell chambers. The lower chamber was supplemented with DMEM containing 10% FCS. After 22 h of incubation, cells were fixed and stained with hematoxylin/eosin. The number of invaded cells was counted in seven randomly selected microscopic fields and photographed.
Immunostaining of PI(3,4)P2 and PI(4,5)P2 at the plasma membrane
Cells plated on coverslips were fixed in formaldehyde 4% and 0.2% glutaraldehyde for 15 min at 4°C. After three washes with PBS containing 50 mM NH4Cl, the cells were incubated in buffer A (PIPES 20 mM pH 6.8, NaCl 137 mM, KCl 2.7 mM) containing 0.05% saponin, 5% NGS, 5% NHS and 50 mM NH4Cl for 45 min on ice to reduce background staining. The coverslips were left overnight at 4°C in a humid chamber with the primary antibodies diluted in buffer A containing 0.1% saponin, 5% NGS and 5% NHS, then they were rinsed in buffer A and then incubated for 3 h on ice with biotin-conjugated goat anti-mouse-IgM antibody diluted in buffer A containing 0.1% saponin, 5% NGS and 5% NHS for PI(3,4)P2 or PI(4,5)P2 staining. They were then rinsed in buffer A and incubated in the dark for 1 h on ice in buffer A containing 0.1% saponin, 5% NGS and 5% NHS containing secondary antibodies. Nuclei were stained with 5 μM DAPI in buffer A for 2 min in the dark at room temperature.
Immunostaining in the presence of Triton X-100 for pFAK, SHIP2, GFP–PH-TAPP1 and GFP–PH-Btk
Cells plated on coverslips were fixed in 4% paraformaldehyde for 30 min at room temperature and incubated in 10 mM Tris-HCl pH 7.4 with 0.15 M sodium chloride (TBS) containing 0.1% (v/v) Triton X-100 (TBS-TX), 10% NHS (Hormonologie Laboratoire, Marloie, Belgium) and 10% NGS (Invitrogen) for 1 h at room temperature. The coverslips were left overnight at 4°C in a humid chamber with the primary antibodies diluted in TBS-T containing 1% NHS and 1% NGS; they were then rinsed in TBS and then incubated in the dark for 1 h at room temperature in TBS containing the secondary antibodies. Coverslips were then rinsed in TBS and nuclei were stained with DAPI 5 μM in 0.05 M Tris-HCl (pH 7.4) for 2 min in the dark at room temperature. Primary antibodies raised in different species and secondary antibodies coupled to different fluorochromes were combined to specifically label one marker in green (Alexa Fluor 488) and the other in red (Alexa Fluor 594). After three rinses in TBS, the coverslips were mounted with Glycergel anti-fade mounting medium. The number of pFAK-stained focal adhesions was quantified following the protocol described previously (Horzum et al., 2014) using ImageJ software. ImageJ plugins ‘CLAHE’ and ‘Log3D’ were used for processing the images and the ‘analyze particles’ plugin was used to quantify the results.
Transfection of GFP bio-probes for PI(3,4,5)P3, PI(4,5)P2 and PI(3,4)P2
GFP or GFP–PH-PLCδ1, GFP–PH-Btk, GFP–PH-TAPP1, to bind PI(4,5)P2, PI(3,4,5)P3 and PI(3,4)P2, respectively, were transfected in MDA-MB-231 shScramble or shSHIP2 cells using Lipofectamine 2000. Briefly, 5×105 cells/well were plated on six-well dishes in complete medium (DMEM medium supplemented with 10% FCS, 1% penicillin/streptomycin and 250 µg/ml fungizone). Lipofectamine and DNA were allowed to form a complex for 30 min in DMEM at room temperature. Cells were washed twice in DMEM followed by the drop wise addition of the DNA–Lipofectamine complex. Cells were incubated for 2 h at 37°C followed by overnight incubation in DMEM supplemented with 20% FCS. The next morning the medium was changed and replaced with DMEM containing 10% FCS and 1% penicillin-streptomycin, and cells were allowed to grow for another day. The same GFP-tagged bio-probes were transfected in MDA-MB-468 cells using Lipofectamine 3000 according to the manufacturer's protocol. Expression of the GFP constructs was verified by fluorescence imaging.
FACS analysis was used to determine the proportion of apoptotic cells using the PE Active Caspase-3 Apoptosis kit (catalog number 550914) from BD Pharmingen. Cells were prepared according to the manufacturer's protocol: briefly, 1×106 cells were trypsinized and fixed in 500 µl of the BD cytofix/cytoperm solution provided with the kit, on ice for 20 min. After washing the fixed cells twice with 1× BD perm/wash buffer, cells were stained with the phycoerythrin (PE)-conjugated active anti-caspase3 antibody for 30 min at room temperature. Following the incubation, cells were washed using the wash buffer and analyzed using the BD Fortessa cell cytometer.
Data are means±s.e.m.; they were analyzed by GraphPad Prism using non-parametric Student's t-test (two-tailed) followed by Mann–Whitney test. Statistical significance was defined as P<0.05 (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001).
We would like to thank Dr Jean-Marie Vanderwinden (Limif and Laboratory of Neurophysiology, ULB) for his help in the imaging portion of our studies. We are grateful to Drs Pierre Roger, Eric Raspé, Viviane De Maertelaer (IRIBHM, ULB) and Andrei Turtoi (Université Montpellier, Montpellier, France) for helpful discussions, and Dr Tamas Balla for the bio-specific probes. We also thank Dr Ievgenia Pastushenko (Stem Cell and Cancer group, ULB) for providing the breast cancer PDX.
Conceptualization: S.G., C.E.; Methodology: S.G., S.S., A.R.R., S.D., C.C., P.K.; Validation: S.G.; Formal analysis: S.G., A.R., C.E.; Investigation: S.G., S.S., A.R.R.; Resources: A.R., S.D., P.K.; Data curation: S.S.; Writing - original draft: S.G.; Writing - review & editing: A.R.R., P.K., C.E.; Visualization: C.E.; Supervision: C.E.; Funding acquisition: C.E.
This work was supported by grants from the Fonds de la Recherche Scientifique Médicale (FRSM) (grant number J.0078.18 to C.E.) and from the Université Libre de Bruxelles. S.G. and A.R.R. are supported by Fondation Rose et Jean Hoguet and Belgium Télévie fellowships.
The authors declare no competing or financial interests.