|Home | About | Journals | Submit | Contact Us | Français|
Many malignancies show increased expression of the EGF receptor family member ErbB3 (HER3). ErbB3 binds beta-1 (HRGβ1), and forms a heterodimer with other ErbB family members, such as ErbB2 (HER2) or EGFR (HER1), enhancing phosphorylation of specific C terminal tyrosine residues and activation of downstream signaling pathways. ErbB3 contains six YXXM motifs that bind the p85 subunit of PI3-kinase. Previous studies demonstrated that overexpression of ErbB3 in mammary tumor cells can significantly enhance chemotaxis to HRGβ1 and overall metastatic potential. We tested the hypothesis that ErbB3-mediated PI3-kinase signaling is critical for heregulin-induced motility, and therefore crucial for ErbB3-mediated invasion, intravasation and metastasis. The tyrosines in the six YXXM motifs on the ErbB3 C-terminus were replaced with phenylalanine. In contrast to overexpression of the wild-type ErbB3, overexpression of the mutant ErbB3 did not enhance chemotaxis towards HRGβ1 in vitro or in vivo. We also observed reduced tumor cell motility in the primary tumor by multiphoton microscopy, as well as a dramatically reduced ability of these cells to cross the endothelium and intravasate into the circulation. Moreover, while mutation of the ErbB3 C-terminus had no effect on tumor growth, it had a dramatic effect on spontaneous metastatic potential. Treatment with the PI3-kinase inhibitor PIK-75 similarly inhibited motility and invasion in vitro and in vivo. Our results indicate that stimulation of the early metastatic steps of motility and invasion by ErbB3 requires activation of the PI3-kinase pathway by the ErbB3 receptor.
The epidermal growth factor receptor (EGFR) family has been a major target of anticancer therapy development (Di Cosimo and Baselga, 2010). Its members can contribute to a wide range of cell phenomena including proliferation, apoptosis, survival, invasion, and differentiation in both normal and neoplastic cells. Members of this family include the epidermal growth factor receptor (EGFR or ErbB1), ErbB2 (Her2/neu), ErbB3 and ErbB4 (Burgess, 2008). ErbB1 and ErbB2 have been most thoroughly studied, with a number of different inhibitors developed in hopes of identifying a treatment that will improve patient survival. However, the functions of ErbB1 and ErbB2 can be dependent upon ErbB3 expression through heterodimerization, and this dependency has repercussions for how tumors may respond to inhibitor treatment (Baselga and Swain, 2009). In NSCLCs that are driven by activating EGFR mutations, high ErbB3 expression is an indicator for gefitinib sensitivity (Engelman et al., 2005; Fujimoto et al., 2005), suggesting that ErbB1/ErbB3 heterodimers may be critical oncogenic units in these tumors. Indeed the development of resistance to EGFR inhibitors in NSCLCs can occur through restoration of ErbB3 activation by upregulation of c-Met (Engelman et al., 2007). In breast cancer, the ErbB2/ErbB3 heterodimer can also form a potent oncogenic unit (Amin et al.; Holbro et al., 2003). In mouse models where ErbB2 overexpression in the mammary gland drives tumor formation, ErbB3 expression and phosphorylation are upregulated (Schade et al., 2007; Siegel et al., 1999). Increased ErbB3 expression correlates with higher hazard ratios for reduced survival of breast cancer patients (Chiu et al.; Sassen et al., 2008).
ErbB3 binds heregulin beta-1 (HRGβ1), but is unable to stimulate downstream signaling on its own as it has a defective kinase domain; however, heterodimerization with another ErbB family member, such as ErbB2 or EGFR, permits tyrosine phosphorylation of the ErbB3 C-terminal domain (Campbell et al., 2010). Downstream signaling from the ErbB receptors includes the activation of a number of pathways, including the PI3-kinase pathway. ErbB3 contains six YXXM motifs that bind the p85 subunit of PI3-kinase (Fiddes et al., 1998; Hellyer et al., 2001; Prigent and Gullick, 1994; Vijapurkar et al., 2003), emphasizing the potential importance of ErbB3 in PI3-kinase activation. In NIH 3T3 cells, mutation of specific tyrosines in the ErbB3 C-terminus uncouples ErbB3 from PI3-kinase, with a strong effect on HRGβ1-stimulated cell transformation and mitogenic responses (Hellyer et al., 2001; Vijapurkar et al., 2003). Previous studies from our laboratory demonstrated that in MTLn3 mammary tumor cells, ErbB3 expression significantly enhances the chemotactic response and in vivo invasion towards HRGβ1, as well as greatly increases metastatic potential without affecting primary tumor growth rate (Hernandez et al., 2009a; Zhang et al., 2006). Thus this model provides a valuable tool for examining how ErbB3 signaling affects metastatic properties beyond the enhancement of cell survival. PI3-kinase signaling via ErbB3 has the potential to modulate actin cytoskeleton rearrangement, thus influencing motility and chemotaxis (Adam et al., 1998; Cain and Ridley, 2009; Chausovsky et al., 2000).
In this paper we tested the hypothesis that the PI3-kinase signaling pathway coupled to ErbB3 is critical for motility, and therefore crucial for invasion, intravasation and metastasis. We created a version of the human ErbB3 receptor in which all six tyrosine residues responsible for binding the p85 subunit of PI3-kinase were replaced with phenylalanine, and evaluated breast cancer cell lines stably expressing either the wild-type ErbB3 or the mutant ErbB3 receptor. Our data revealed that mutation of the PI3K binding sites blocked a number of responses that are enhanced by overexpression of wild-type ErbB3. These include enhanced chemotaxis towards HRGβ1 in vitro and in vivo, increased tumor cell motility in the primary tumor, and increased transendothelial migration, intravasation, and metastasis. Taken together, our results demonstrate that coupling of the PI3-kinase pathway to ErbB3 is essential during the initial steps of breast cancer metastasis.
We created a version of the human ErbB3 receptor in which the tyrosines for all six YXXM motifs responsible for binding the p85 subunit of PI3-kinase were replaced with phenylalanine by site-directed mutagenesis. We then generated stable transductants of MTLn3 cells with the empty vector pLXSN (pLXSN), the wild-type human ErbB3 receptor (ErbB3WT), or the human ErbB3 receptor mutated in all six PI3-kinase binding sites (ErbB3-Mutant). Surface overexpression levels of the ErbB3WT and ErbB3-Mutant constructs were similar (Figure 1A). The endogenous rat ErbB3 was reduced using a rat-specific shRNA. All of the cell lines also stably expressed GFP. To measure ErbB3-Mutant and WT signaling activity, starved cultures were stimulated with HRGβ1, and ErbB3 was immunoprecipitated from MTLn3 ErbB3WT and ErbB3-Mutant cell lysates. ErbB3 signaling activity and coupling to the PI3K pathway was assessed by measuring ErbB3 immunoprecipitates (IP) for tyrosine phosphorylation (PTyr) and p85α association. ErbB3 IPs from HRGβ1-stimulated ErbB3WT cells had high levels of PTyr and complexed p85α, while ErbB3 IPs from ErbB3-Mutant cells had very low levels of PTyr and little or no associated p85α (Figure 1B). We then tested whether decreased signaling through PI3K had an effect on HRGβ1-induced chemotaxis (Figure 1C). ErbB3WT cells showed dramatically enhanced chemotaxis to HRGβ1 compared to the control pLXSN cell line. In contrast, ErbB3-Mutant cells were similar to control cells, indicating that removal of the PI3-kinase binding sites completely blocks chemotactic responses driven by HRGβ1. Chemotaxis in response to insulin did not show differences between ErbB3WT and ErbB3-Mutant cells (Supplementary Figure 1). We also measured tumor cell invasiveness using Matrigel-coated transwells. We found that HRGβ1 significantly enhanced invasiveness of ErbB3WT cells into Matrigel, while invasiveness of ErbB3-Mutant cells was significantly lower (Figure 1D). In summary, these data suggest that PI3K association with ErbB3 is required for HRGβ1 to induce chemotaxis and invasion.
We then investigated the metastatic potential of the three tumor cell lines. Cells were injected into the mammary fat pads of SCID mice. We observed no significant differences in tumor growth and final tumor volumes between the three tumor types (Figure 2A). TUNEL (for apoptotic cells), IBA-1 (for macrophages), pAKT, and pERK staining did not show any differences between the tumors (Supplementary Figures 2 A - D). Strikingly, intravasation as measured by the number of circulating tumor cells was significantly higher for ErbB3WT compared to both pLXSN and ErbB3-Mutant (Figure 2B). However, staining for CD34 revealed that this result was not due to obvious differences in the number of blood vessels, which were the same in all tumor types (Supplementary Figures 2 E and F). Our results indicate that the enhancement in intravasation produced by ErbB3 is dependent upon coupling to PI3K.
Intravasation can be a rate limiting step for metastasis (Wyckoff et al., 2000), and to test whether metastatic capability correlated with intravasation capability, H&E sections of the lungs of the mice were evaluated. The lungs of mice carrying ErbB3WT tumors showed greater than 10 times more metastases than the lungs of mice with pLXSN or ErbB3-Mutant tumors (Figure 2C). These results are consistent with intravasation rates regulating metastasis rates in this model. To evaluate lung seeding ability separately from intravasation efficiency, the three cell lines were injected into the lateral tail vein of SCID mice (experimental metastasis assay). Two weeks post- injection the mice were sacrificed and H&E sections of the lungs were examined for metastases. The lungs of mice injected with ErbB3WT cells had a small, but significant increase (approximately 2-fold) in lung seeding efficiency compared to the lungs of mice injected with the pLXSN and ErbB3-Mutant lines (Figure 2D). We conclude that the PI3-kinase binding sites are required in vivo for ErbB3 enhancement of intravasation and lung seeding, with a major contribution to metastasis occurring at the intravasation step.
To examine the importance of ErbB3/PI3K for invasion in the primary tumor microenvironment, we used multiphoton microscopy to perform intravital imaging. Intravital imaging enables visualization and quantification of the movement and orientation of fluorescent cells directly at the primary tumor without exogenously added stimulants or chemoattractants. Local motility and invasion in the primary tumor may contribute to the ability of tumor cells to locate blood vessels and to intravasate. GFP expressing lines were used in this assay, and 33-38 days post injection into the mammary fat pad, animals were anesthetized and skin flap surgery was performed to expose the primary tumors for imaging (Wyckoff et al., 2007). Total tumor cell motility in pLXSN, ErbB3WT and ErbB3-Mutant tumors was evaluated by analyzing multiple 30-minute z-series from 7-8 tumors generated from each cell line. We found that ErbB3WT expression significantly enhanced tumor cell motility compared to the pLXSN control (Figure 3A and Supplementary Movies 1 and 2). Importantly, expression of the ErbB3-Mutant did not enhance cell motility above the level of the control cell line.
The ErbB3WT cells were more polarized near vascular and stromal areas in the tumor compared to ErbB3-Mutant cells. We used the length/width ratio of cells in the proximity of vessels to quantify polarization (Wyckoff et al., 2000). There was a significant increase in the length/width ratio in ErbB3WT tumors (Figure 3B). This is consistent with increased sensitivity to endogenous gradients of ligands which could enhance both invasion and intravasation. We evaluated whether there was enhanced invasion in vivo in response to HRGβ1 using microneedles containing Matrigel and HRGβ1 inserted into the primary tumors to collect invasive cells (Figure 3C). ErbB3 expression had no effect on basal invasion into needles containing Matrigel alone. However, ErbB3WT tumor cells showed significantly enhanced in vivo invasion in response to 50 nM HRGβ1 over the empty vector control, while the mutant cells behaved like control cells. These data reveal that the ability of ErbB3 to directly couple to the PI3K pathway is essential for HRGβ1’s ability to stimulate in vivo invasion. To model the in vivo intravasation step, we used an in vitro transendothelial migration (TEM) assay (Figure 3D). Tumor cells alone, or mixed with macrophages, were plated on top of an endothelial monolayer and allowed to migrate through to the other side for 18 hours. For all lines, very few carcinoma cells traversed the endothelium in the absence of macrophages. The addition of macrophages dramatically enhanced the ability of ErbB3WT cells to cross the endothelium, but did not increase the ability of pLXSN to cross. The ErbB3-Mutant line showed enhancement by macrophages which was, however, significantly less than ErbB3WT cells. This is similar to the pattern of intravasation efficiency that we observed in vivo (Figure 2B). Importantly, IBA-1 staining of primary tumors revealed the presence of macrophages in all three tumor types (Supplementary Figure 2B), indicating that the defect in intravasation is not due to a difference in macrophage density in the tumor.
As an independent confirmation of the role of PI3-kinase in chemotaxis and invasion of ErbB3WT cells, we used a p110α-selective PI3-K inhibitor, PIK-75. At a concentration of 200 nM, PIK-75 dramatically reduced the in vitro chemotactic response to HRGβ1 (Figure 4A) and lowered pAkt levels by ~40% (Supplementary Figure 3). PI3K inhibition increased Erk activity by ~50%, consistently with a previously observed relief of feed-back inhibition following blockade of the PI3K/Akt pathway (Chandarlapaty et al., 2011; Serra et al., 2011). We then evaluated the effect of PIK-75 on invasion in vivo. Comparing microneedles containing HRGβ1 alone or HRGβ1 with PIK-75 inserted into ErbB3WT tumors, significantly fewer ErbB3WT cells migrated into microneedles containing HRGβ1 with PIK-75 compared to HRGβ1 alone (Figure 4B). Of note, a 4-hour treatment of cells in vitro with PIK-75 had no effect of cell viability (data not shown). We also examined the ability of systemic PIK-75 to inhibit HRGβ1-induced responses at the primary tumor. Mice bearing ErbB3WT tumors were IP-injected with PIK-75 three hours before performing the in vivo invasion assay using microneedles containing HRGβ1. Tumors in PIK-75-treated animals showed significantly reduced in vivo invasion (Figure 4C), compared to tumors from vehicle control treated animals confirming the importance of the PI3K pathway in HRGβ1-induced responses in vivo. Interestingly, PIK-75 did not inhibit EGF-induced in vivo invasion (Supplementary Figure 4), supporting our hypothesis that PI3K signaling is specifically critical in HRGβ1/ErbB3-driven invasion.
Next we used the same treatment regimen and evaluated tumor cell motility by intravital imaging. PIK-75 treatment significantly reduced total tumor cell motility in ErbB3WT primary tumors (Figure 4D, also see Supplementary Movies 3 and 4). We observed that the ErbB3WT cells treated with vehicle control were more polarized near vessels in the tumor, as compared to tumor cells in mice pretreated with PIK-75. To determine whether this observation was significant, we evaluated the length/width ratio of cells in close proximity to vessels employing the same method used previously to compare the polarization of ErbB3WT and ErbB3-Mutant tumor cells. There was a significant decrease in the length/width ratio in ErbB3 tumors pretreated with the inhibitor (Figure 4E). These effects were not due to differences in tumor cell survival, as TUNEL staining was similar in PIK-75 and vehicle-treated ErbB3WT tumors (data not shown).
To provide more mechanistic insight into the pathways that are important for motility and invasion, activity of the PI3K/Akt and the Erk pathways were measured in response to HRGβ1 treatment using phospho-specific antibodies. Cultures were stimulated with HRGβ1 for various times and pAkt (T308) and pERK levels were measured in lysates from the three cell lines. There were no significant differences in the basal levels of pERK, or in the HRGβ1-induced pERK levels in the three cell lines (Figure 5A), In contrast, basal pAKT levels were lower and the kinetics of HRGβ1-induced Akt activation were delayed in the ErbB3-Mutant cells in comparison to the other cell lines. ErbB3WT cells showed maximum pAKT levels by 5 min (Figure 5B and Supplementary Figure 5), while the ErbB3 mutant cells required 10-15 min to reach maximum.
We next examined whether the differential effect on signaling in ErbB3WT and ErbB3-Mutant cells was specific to HRGβ1 stimulation, by evaluating pAKT and pERK after stimulation with EGF (Figure 5 C and D), PDGF or with IGF-1 (Figure 5D). The results show that there was no difference in pAKT or pERK levels in EGF treated cells. IGF-1 caused an increase in pAKT, not pERK, and the pAKT levels were the same in both cell lines. Finally, PDGF had no effect on these pathways, likely due to absence of the receptor. Thus loss of the PI3K binding sites on ErbB3 resulted in AKT phosphorylation in response to HRGβ1.
These results suggest that HRGβ1’s ability to cause a rapid stimulation of PI3K signaling is important for its biological activity in motility and invasion. Next we employed an in vitro invasion assay using Matrigel-coated transwells in combination with specific inhibitors to investigate pathways downstream of ErbB3 in the ErbB3 WT cells (Figure 5E). We found that HRGβ1-induced invasion was totally blocked by inhibition of p110α and Akt. Next we looked at Rac1, which is downstream of PI3K, in the three cell lines. HRGβ1 rapidly stimulated Rac1 activity by 2-fold in the ErbB3 WT cells, while there was no induction in control and low levels in ErbB3-Mutant cells (Supplementary Figure 6). In accordance with these results treatment with a Rac 1 inhibitor (NSC23766) reduced invasion by approximately 50% (Figure 5E). In contrast, treatment with a Pak inhibitor had no effect on HRGβ1-induced invasion (data not shown). Taken together, the data suggest that invasion of the ErbB3 WT cells is mainly dependent upon the PI3K-Akt-Rac pathway.
In this study we utilized the MTLn3 cell line to evaluate the role of the YXXM sites in ErbB3 in breast cancer metastasis. We find that these sites are required for ErbB3-mediated chemotactic responses to HRGβ1 in vitro and invasion responses both in vitro and in vivo. Increased expression of wild-type ErbB3 had no effect on primary tumor growth, but significantly enhanced intravasation and spontaneous metastasis (Fig 2 and our previous publication (Xue et al., 2006)). Mutation of the YXXM sites dramatically reduced the enhancement of intravasation and metastasis, while having no effect on primary tumor growth. In an experimental metastasis assay, wild-type ErbB3 overexpression had a significant, albeit slight, effect on the lung seeding process, which also required ErbB3 coupling to PI3K. Similarly, an increase in spontaneous motility and cell polarization in the primary tumor induced by wild-type ErbB3 required the YXXM sites. Treatment of tumor-bearing mice with the p110alpha inhibitor PIK-75 caused a significant decrease in intravital motility and in vivo invasion, attesting to the important role of PI3K in invasion. Overexpression of ErbB3 in the aggressive MDA-MB-231 human breast cancer model also significantly enhanced the invasiveness of the cells in response to HRGβ1 treatment (Supplementary Figure 7). Taken together the data presented here indicate that the ability of ErbB3 to couple to the PI3K/Akt pathway confers tumors overexpressing this receptor with an enhanced ability to invade and to metastasize.
Based on the studies reported here, we have identified intravasation as a key stage in metastasis that is enhanced by ErbB3 signaling both in vivo and in the in vitro transendothelial migration assay. Intriguingly, EGFR overexpression is not equivalent to ErbB3 overexpression in the in vitro TEM assay, since the former do not show increased intravasation (data not shown), while as we show here ErbB3 overexpression has a positive impact. These results are consistent with our in vivo studies using EGFR inhibitors (Kedrin et al., 2009a), where we show that inhibition of EGFR had a slow inhibitory effect on intravasation and was only evident with a delay of several hours after EGFR inhibition. Conversely, inhibition of ErbB2 rapidly inhibited intravasation (Kedrin et al., 2009b). Since ErbB3 forms preferred heterodimers with ErbB2, this suggests that an ErbB3/ErbB2 heterodimer is important in enhancing the intravasation step, while EGFR is important for tumor cell invasion from the primary tumor to reach the blood vessels (Kedrin et al., 2009a). HRGβ1 is present in serum in subnanomolar concentrations (Ky et al., 2009), and by increasing ErbB3 levels in the tumor cells this might enable them to detect low levels of HRGβ1 and increase their intravasation. Endothelial cells have also been reported to express HRGβ1, which may also stimulate intravasation (Iivanainen et al., 2007; Kalinowski et al.).
HRGβ1 activated PI3K/Akt in both ErbB3 WT and ErbB3-Mutant cells but with different kinetics. In ErbB3WT cells the pathway was maximally activated after 5min of HRGβ1 stimulation, while it took between 10 and 15min to see the same signal intensity in the ErbB3-Mutant cells. This suggests that early pathway activation, which is dependent upon p85/PI3K binding to HRGβ1 activated ErbB3-containing heterodimers, is responsible for the major effects of the ligand on migration and invasion. The fact that blockade of PI3K or Akt activity using selective inhibitors totally blocked HRGβ1-induced invasion (Fig 5E), while ErbB3-Mutant cells were only blocked in their invasive ability by ~ 60% (Fig 1D), very likely reflects the fact that the PI3K/Akt pathway does get activated in these cells, but with slower kinetics. The importance of the kinetics of signaling pathway activation in response to ErbB ligands has been observed in other systems, (see, e.g. (Olayioye et al., 2001)).
While previous studies have shown the direct binding of p85 and subsequent signaling from all six YXXM motifs on the ErbB3 C-terminus (Hellyer et al., 2001; Prigent and Gullick, 1994), high-throughput methods such as protein microarray studies suggest that other molecules could bind these sites (Jones et al., 2006). To confirm that the phenotype in the ErbB3-Mutant MTLn3 cells is due to defective PI3K pathway signaling, we used the p110-α selective inhibitor PIK-75, revealing that the in vitro chemotaxis as well as in vivo invasion of MTLn3-ErbB3WT cells towards HRGβ1 was impaired following PI3K inhibition. Moreover pretreatment of mice carrying ErbB3WT cell-induced tumors with PIK-75 caused reduced tumor cell motility, as visualized by multiphoton microscopy. Furthermore, the cells in the treated tumors were also more rounded on average as quantified by length/width ratios, thus mimicking the phenotype of the mutant ErbB3 tumors. These experiments support the hypothesis that PI3K pathway activation by ErbB3 is a major contributor to enhancing the initial steps of the metastatic cascade of mammary tumor cells.
Rat mammary carcinoma MTLn3 cells (Neri et al., 1982) were grown in α-MEM supplemented with 5% fetal bovine serum (FBS) and penicillin/streptomycin solution (Life Technologies). HRGβ1 was from R & D Systems (396-HB). MDA-MB-231 cells (American Type Culture Collection) were cultured in DMEM (Invitrogen) with 10% fetal bovine serum.
All procedures involving mice were conducted in accordance with the National Institutes of Health regulations concerning the use and care of experimental animals. The study of mice was approved by the Albert Einstein College of Medicine animal use committee. MTLn3 derived cell lines were grown to 70-85% confluence before being harvested for in vivo assays. To measure spontaneous metastasis, 5×105 cells were detached by incubation in DPBS + 2 mmol/L EDTA, scraping with a rubber policeman, centrifuged, and resuspended in DPBS at 107 cells/mL. MTLn3 (5 × 105) cells were injected into the right fourth mammary fat pad from the head of 5- to 7-week-old female severe combined immunodeficient (SCID/NCr) mice (National Cancer Institute, Bethesda, MD) in 100 μL DPBS with calcium and magnesium (Mediatech 21-031-cv) through a 25-gauge needle. Tumor growth rate was monitored at weekly intervals by measuring in two dimensions, and tumor volumes were calculated using the formula: length × width2/2. At the end point for spontaneous metastasis, the blood burden assay was performed using blood taken from the right atrium via heart puncture with a 25-gauge needle and a 1 mL syringe coated with heparin and containing 0.1 mL of heparin. Blood was plated in tissue culture medium and after 7 days all tumor colonies in the dish were counted. Tumor blood burden was calculated as the total colonies in the dish divided by the volume of blood taken. To measure experimental lung metastasis, 5 × 105 cells were injected into the lateral tail veins of 5- to 7-week-old female SCID mice (National Cancer Institute). Two weeks after injection, the mice were sacrificed, and the lungs were removed, fixed in formalin, and stained H&E sections were counted for metastasis. For each lung sample, all micrometastases were counted using a 10x objective. The efficiency of lung metastasis was expressed in number of metastases per lung section for each animal. Mean and SE were then calculated for each cell line.
The measurement of cell invasion into needles placed in the primary tumor of anesthetized mice was carried out as described previously in detail (Hernandez et al., 2009b). Invasive cells were collected into 33-gauge Hamilton needles (Fisher 14-815-423) filled with Matrigel (Beckton Dickinson 356234) diluted 1:10 with L15-BSA +/- chemoattractant for 4 hours. At the end of collection, the contents of the needles were extruded using 0.5 μg/ml DAPI in PBS in a syringe onto a coverslip. The chemoattractants used in this assay include HRGβ1 at a concentration of 50 nM (R & D Systems 396-HB) and EGF at 25 nM (Life Technologies). To inhibit the activity of PI3-kinase, PIK-75 (Axon Medchem, Axon 1334), a p110-alpha-selective inhibitor (Chaussade et al., 2007), was used at 1 μM.
Orthotopic primary tumors were grown and imaged at 4-5 weeks post injection. The animal was anesthetized, and a skin flap surgery was performed to expose the tumor. The animal was placed on an inverted microscope and tumor cells were imaged using a Radiance 2000 MP multiphoton or on an Olympus Fluoview FV1000-MPE microscope using an excitation wavelength of 880nm with a 20X 0.95NA water objective (Wyckoff et al., 2007) or a 25X 1.05NA water objective at resolutions of 1.06 and 0.994 um per pixel, respectively. Cell motility was observed by time-lapse imaging over 30 min at 2-minute intervals, where a 100 μm z-series, with 10 μm steps, was collected at each time point. For quantification of average total cell motility, a motility event was defined as a protrusion of half a cell length or more. The length/width ratio (L/W) was calculated using ImageJ as follows: length was scored as the Feret’s diameter of the cell (the longest distance between any two points along the boundary); width was scored as the secondary axis of the best fit ellipse of the circled cells. Cells near stromal and vascular areas of the tumor in the top 40 um from the tumor surface were selected for these measurements. For inhibition of PI3-kinase in the tumor cells in vivo for imaging, PIK-75 was dissolved in DMSO and three hours prior to imaging, a dose calculated to be 50mg/kg of body mass was mixed with 200-300 ul PBS such that the final solution consisted of no more than 20% DMSO and 20% β-cyclodextrin, and injected intraperitoneally, following a previously described procedure (Hayakawa et al., 2007).
Additional reagents and methods are described in the Supplementary Materials and Methods.
Statistical significance was determined using unpaired, two-tailed Student’s t tests assuming unequal variances and an alpha level of 0.05. Differences were considered significant if the p value was <0.05. For blood burden assays the non-parametric Mann Whitney Wilcoxon rank sum test was used.
We thank John Koland and the members of the Segall, Condeelis, Cox, Hodgson, and Hynes laboratories for comments and suggestions. JES is the Betty and Sheldon Feinberg Senior Faculty Scholar in Cancer Research.
Financial support: This work was supported by grants from the NIH (JES: CA77522 and CA100324); CA1U01105490 and CA100324 (J.S. Condeelis and J. Wyckoff); NIH P01 CA100324 (JB and ARB).
CONFLICT OF INTEREST The authors declare no conflict of interest.