Phosphoinositide 3-kinase signaling is critical for ErbB3-driven breast cancer cell motility and metastasis

Abstract

Many malignancies show increased expression of the epidermal growth factor (EGF) receptor family member ErbB3 (HER3). ErbB3 binds heregulin β-1 (HRGβ1) and forms a heterodimer with other ErbB family members, such as ErbB2 (HER2) or EGF receptor (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 phosphoinositide 3 (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, whereas 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.

Figure 1. ErbB3 mutations block tyrosine phosphorylation, PI3K binding and show a defect in in vitro chemotaxis and invasion to HRGβ1

A. Flow cytometry analysis of surface ErbB3 levels in pLXSN (solid line), ErbB3WT (dotted line), and ErbB3-Mutant (dashed line) cells. A mouse anti-ErbB3 primary followed by an APC-labeled secondary was used. The curve with solid shading represents ErbB3WT cells incubated with just the APC-conjugated secondary antibody. B. HRGβ1-induced ErbB3 tyrosine phosphorylation and association to the p85 subunit of PI3K. Serum starved MTLn3 ErbB3WT or ErbB3-Mutant cells were stimulated with buffer or HRGβ1 (0 or 0.4 HRG) for 5 minutes. ErbB3 immunoprecipitates (IP) and whole cell lysates (WCL) were resolved by SDS/PAGE, and probed for (PTyr) or p85α. Membranes were stripped and reprobed to control for ErbB3 levels. C Chemotaxis to HRGβ1 of pLXSN (light gray bars), ErbB3WT (dark gray bars) and ErbB3-Mutant (patterned bars). Data are mean and SEM of 6-27 wells in 3-6 independent experiments. D. In vitro invasion responses of ErbB3WT (black bars) and ErbB3-Mutant cells (patterned bar) into Matrigel-coated transwells stimulated by buffer or 12.5 nM HRGβ1. Data are presented as % Area invaded, and are mean and SEM of 3 independent experiments. *:p< 0.02

Figure 2. PI3-kinase binding sites contribute to intravasation and lung metastasis

A. Tumor growth after orthotopic injection into the mammary fat pad. There are no statistically significant differences between the average primary tumor volumes formed by MTLn3 pLXSN, ErbB3WT or ErbB3-Mutant lines. Data are means and SEM of 9-27 tumor measurements determined on average 36 days after injection for all cell lines. B. Intravasation measured 4-5 weeks post injection into the mammary fat pad. Blood drawn from the right atrium of the heart was plated into cell culture medium and colonies were counted 7 days later. Data are means and SEM of total numbers of colonies normalized per ml of blood drawn for 15 – 17 mice per cell line. *: p<0.004, **: p<0.05. C. Spontaneous metastatic potential measured 4-5 weeks post injection into the mammary fat pad was determined by quantifying metastases in H&E stained sections through the middle of the lungs. Data are means and SEM of counts from 11 – 13 animals carrying tumors from each cell line. *: p< 0.02 D. Lung seeding efficiency determined by evaluating metastases in H&E stained sections through the middle of the lungs 2 weeks post injection via tail vein. Data are means and SEM of 9 – 12 animals per cell line. *: p<0.02, **: p<0.03

Figure 3. PI3-kinase binding sites are required for ErbB3 enhancement of in vivo motility and invasion

A. Intravital imaging using multiphoton microscopy of primary mammary tumors of GFP-labeled tumor cells. Time-lapse z-series were acquired and the average total cell motility was determined per 40 μm z-stack (5 sections imaged at 10 μm intervals). Data are means and SEM of 24 – 33 separate z stacks from 7-8 mice per cell line. *: p<0.025, **: p<0.0000042. i/ii panels illustrate examples of (i) ErbB3WT GFP and (ii) ErbB3-Mutant GFP cell motility in vivo. Cells are green with collagen fibers detected by second harmonic scattering in purple. Frames are 6 minutes apart, arrows show migrating tumor cell; Scale bar = 25 μm. Asterisks mark a single cell in i and ii and arrows mark a moving cell in i. B. Length/width ratio comparisons. Top: representative images. Bottom: Mean and SEM of ratios of 50-100 cells from 3 tumors, 5 slices analyzed per z stack. Scale bar = 25 μm. *: p<0.002, **: p<6×10^-5 C. In vivo invasion responses. Microneedles containing buffer or 50 nM HRGβ1 were inserted into primary tumors and invasive cells were quantified. *: p<4×10^-5, **: p< 5×10^-6. Data are means and SEM of 9-13 measurements from 3-4 tumors per cell line (pLXSN: gray, ErbB3WT: black, ErbB3-Mutant: patterned). D. In vitro transendothelial migration assay. Cells were plated on the basal side of an endothelial monolayer either alone or mixed with macrophages (+M) and allowed to migrate for 18 hours. for comparison of ErbB3WT+M with ErbB3-Mutant+M: p<4×10^-6. Data are represented as the fold migration over pLXSN alone, with mean and SEM of 4 independent experiments with each cell line.

Figure 4. The PI3-kinase inhibitor PIK-75 blocks HRGβ1 induced chemotaxis, in vivo invasion and total tumor cell motility

A. In vitro chemotaxis is inhibited by PIK-75. Migration of ErbB3WT cells in response to a gradient of either 1.25 nM HRGβ1 (black bar) or 1.25 nM HRGβ1 + 0.2 μM PIK-75 (patterned bar). *: p< 0.0001. Data are mean and SEM of measurements from 3 independent experiments. B. In vivo invasion is inhibited by PIK-75. Microneedles containing 50 nM HRGβ1or 50 nM HRGβ1 + 1 μM PIK-75 were inserted into the primary tumor and invasive cells were collected and counted. Data are mean and SEM from 6 needles for HRGβ1 and 10 needles for HRGβ1 + PIK-75 from 3 tumors for each condition. *: p< 1 × 10^-5. C. In vivo invasion stimulated by 50 nM HRGβ1 is inhibited by systemic application of PIK-75. Mice bearing ErbB3WT tumors were injected i.p. either with vehicle control (black bar) or with 1.5 mg PIK-75 (pattern) 3 hours prior to assay. Microneedles containing 50 nM HRGβ1 were inserted into primary tumors and invasive cells were quantified as described in the methods.*: p< 0.00035. Data are means and SEM of 4-8 measurements from 4 tumors per condition. D. Cell motility in vivo using intravital imaging. Mice bearing tumors were injected i.p. either with vehicle control (black bar) or with 1.5 mg PIK-75 (pattern) 3 hours prior to imaging. Tumors were exposed by skin-flap surgery and time-lapse z-series were acquired. The average total cell motility was determined per 40 μm z-stack (5 sections imaged at 10 μm intervals). Top: Data are means and SEM of 8 (vehicle control) and 15 (PIK-75) separate z stacks from 3 mice per condition. *: p< 0.016 Bottom: Representative panels from tumors of animals injected with (i) MTLn3-ErbB3--GFP vehicle control or (ii) PIK-75 3 hours prior to imaging. Cells are green with collagen fibers detected by second harmonic scattering in purple. Frames are 6 minutes apart, arrow shows migrating tumor cell; * indicates a single cell. Scale bar = 25 μm. E. Length/width ratio comparison. Top: representative images Bottom: Mean and SEM of 40-50 cells from 3 tumors per condition, 5 slices analyzed per field. Scale bar = 25 μm.*: p<0.002

Figure 5. Analysis of HRGβ1 induced signaling in the ErbB3 WT and ErbB3-Mutant cells

A. Erk phosphorylation was determined in serum-starved cells after 5 minute treatment with 0 (-) or 0.4 nM HRGβ1 (+). Similar results were observed in three independent experiments. B. MTLn3 pLXSN, ErbB3WT, and ErbB3-Mutant cells were stimulated with 0.4 nM HRGβ1for 0, 5, 10, and 15 minutes (Min HRG). A delay in phosphorylation of Akt on Thr 308 was observed in ErbB3-Mutant cells response to HRGβ1. C. Akt (T308) and Erk phosphorylation activation was determined in serum-starved cells after 5 min treatment with 0 (-) or 5 nM (+) EGF. Similar results were observed in three independent experiments. D. Akt (T308) and Erk phosphorylation activation was determined in serum-starved cells after 10 min treatment with PDGF, IGF-I, and EGF. Similar results were observed in three independent experiments. E. Matrigel-coated transwells were used to test 12.5 nM HRGβ1induced in vitro invasion responses of MTLn3 ErbB3WT in the presence of vehicle (gray bars) or inhibitor (patterned bars). PIK-75 is a p110α selective inhibitor, Triciribine is an Akt inhibitor, and NSC23766 is a Rac1 inhibitor. Data are mean and SEM from 3 independent experiments. *: p< 0.05, **: p< 1.83E-05