Macrophage-dependent tumor cell transendothelial migration is mediated by Notch1/MenaINV-initiated invadopodium formation

The process of intravasation involving transendothelial migration is a key step in metastatic spread. How the triple cell complex composed of a macrophage, Mena over-expressing tumor cell and endothelial cell, called the tumor microenvironment of metastasis (TMEM), facilitates tumor cell transendothelial migration is not completely understood. Previous work has shown that the physical contact between a macrophage and tumor cell results in the formation of invadopodia, actin-rich matrix degrading protrusions, important for tumor cell invasion and transendothelial migration and tumor cell dissemination. Herein, we show that the macrophage-induced invadopodium is formed through a Notch1/MenaINV signaling pathway in the tumor cell upon macrophage contact. This heterotypic tumor cell – macrophage interaction results in the upregulation of MenaINV through the activation of MENA transcription. Notch1 and MenaINV expression are required for tumor cell transendothelial migration, a necessary step during intravasation. Inhibition of the Notch signaling pathway blocked macrophage-induced invadopodium formation in vitro and the dissemination of tumor cells from the primary tumor in vivo. Our findings indicate a novel role for Notch1 signaling in the regulation of MenaINV expression and transendothelial migration and provide mechanistic information essential to the use of therapeutic inhibitors of metastasis.


Notch1 signaling is required for macrophage-induced formation of invadopodia in tumor cells.
To evaluate if Notch signaling is required for macrophage -induced invadopodium formation (Fig. 1A), cells were treated with DAPT, a γ -secretase inhibitor, which inhibits intracellular Notch signaling by preventing its cleavage into the active NICD 26 . A mature invadopodium is defined herein as having cortactin and Tks5 positive staining as well as being co-localized with a discreet area of matrix degradation. Tks5 is required for anchoring the invadopodium core to the plasma membrane via its binding to PI (3, 4) P2 27 and its association with these other two markers is a definitive identifier of mature invadopodia. In the absence of macrophages and in serumstarved conditions, DAPT treatment has no significant effect on invadopodium assembly by MDA-MB-231 human breast tumor cells (Fig. 1B). When BAC1.2F5 macrophages are added to the MDA-MB-231 culture there is a significant increase in the number of mature invadopodia per tumor cell but the addition of DAPT to these co-cultures prevents the macrophage-mediated induction of invadopodia (Fig. 1B).
Increased Notch1 signaling is associated with a greater chance of metastasis and poor prognosis 23 . Therefore, to determine if Notch1 receptor is important in macrophage-induced invadopodium formation we used siRNA mediated knockdown of Notch1. Knockdown of Notch1 receptor in MDA-MB-231 cells resulted in a significant reduction in macrophage-induced invadopodia as well as a significant reduction in invadopodium-associated matrix degradation to baseline levels seen in the absence of macrophages (Fig. 1C-E). Notch1 inhibition had no significant effect on steady sate invadopodium formation in the absence of macrophages (Supplemental Figure 1A and B). Thus, we conclude that Notch1 is required for macrophage-induced invadopodium assembly and function. These data indicate a novel signaling pathway involving heterotypic cell-cell communication of Notch1 in macrophage-induced invadopodium formation in tumor cells.

Notch1 signaling is required for macrophage -tumor cell contact-induced Mena INV expression.
Notch1 has well defined roles in the regulation of gene and protein expression at the transcriptional level, although a role for Notch1 in the regulation of the macrophage-induced invadopodium pathway is unknown. We sought to determine if macrophage -induced Notch1 signaling in tumor cells results in expression changes in genes with known roles in invadopodium formation and function. We performed a qRT-PCR analysis to determine changes in mRNA levels of the known invadopodium pathway regulatory proteins ( Fig. 2A). The co-culture of MDA-MB-231 tumor cells with macrophages resulted in no significant mRNA changes in actin, cortactin, cofilin, N-WASp or the RhoGTPases Rac1, Cdc42, RhoA and RhoC, which all have well defined roles in invadopodia regulation 12,[28][29][30][31][32][33] . There was no significant change in the level of total (pan) Mena mRNA levels which are very abundant compared to Mena INV (Supplemental Figure 2B), but the Mena INV isoform was highly upregulated (~45x) while Mena11a isoform levels remain unchanged. This finding is of great interest because high expression Scientific RepoRts | 6:37874 | DOI: 10.1038/srep37874 of the Mena INV isoform has been implicated in the increased motility, invadopodium assembly and invasion by tumor cells 1,8,34,35 .
There is a well described paracrine signaling loop between human breast or mouse mammary tumor cells and macrophages 1,35,36 . Due to this well-defined soluble signaling pathway between tumor cells and macrophages, we wanted to confirm that the increase in Mena INV upon co-culture with macrophages was a result of cell-cell contact and not soluble signaling factors. We performed co-culture experiments utilizing 3 μ m pore transwells, which allow for the passage of soluble factors but not the tumor cells. Tumor cells were either plated in the transwell alone, with the two cell types mixed in the top of the well, or with macrophages added to the bottom of the well to prevent tumor cell -macrophage contact (Fig. 2B). When the cell types were mixed together in the top of the transwell, allowing cell-cell contact, we observed the increase in Mena INV mRNA whereas when the cells were plated on opposite sides of the transwell, allowing for the exchange of soluble factors but not direct contact, no significant change in Mena INV mRNA was observed.
To confirm that the increase in Mena INV mRNA is regulated by the macrophage-tumor cell contact initiated Notch1 signaling, we treated the tumor cells with Notch1 siRNA or DAPT. We found that both had the ability to abrogate the increase in Mena INV mRNA during co-culture of the two cell types ( Fig. 2C and D).
To determine if these events are macrophage-contact specific, we co-cultured tumor cells at high density to induce tumor cell-tumor cell contact, tumor cell and macrophage contact or tumor cells and endothelial cell contact. These are the three different cell types that comprise TMEM sites. The only contact that was able to induce an increase in Mena INV expression was the tumor cell and macrophage co-culture (Supplemental Figure 2A). In addition, the contact between tumor cells and HUVEC endothelial cells did not result in an increase in mature invadopodia or matrix degradation (Supplemental Figure 2B and C).
In addition to the MDA-MB-231 cells, we confirmed the macrophage-induced expression of Mena INV in primary tumor cells. Tumor cells were obtained by fine needle aspiration biopsy (FNA), which yields ~97% pure Primary cancer cells were also obtained by FNA from two human invasive ductal carcinomas immediately after surgical resection (Fig. 3B, patient #1 had ER+ disease and patient #2 had ER-disease). In cells from these 2 patient samples we also observed a dramatic increase in Mena INV mRNA when co-cultured with macrophages (Fig. 3B). These data indicate a novel pathway for the upregulation of the pro-migratory/invasive Mena INV isoform of Mena that is common to breast cancer cells with different ER expression status. To test the hypothesis that the increase in Mena INV mRNA seen in tumor cells in contact with macrophages is associated with increased Mena gene transcription, we designed fluorescent in-situ hybridization (FISH) probes to MENA. The different Mena isoforms are splice variants transcribed by the same gene 6,11 ; therefore the probes designed recognize the full length MENA gene transcript (Supplemental Figure 3). Macrophages were labeled with cell tracker red and co-cultured with tumor cells for 60 minutes (Fig. 4A). When MDA-MB-231 tumor cells are cultured alone, only ~10% have an active MENA transcription site, as seen by FISH (Fig. 4B). When tumor cells and macrophages are plated in co-culture for 1 hour, tumor cells that are not in contact with macrophages still only display ~10% of cells actively transcribing MENA, while tumor cells touching a macrophage have over 45% of cells with active transcription sites ( Fig. 4A and B). Figure 4A shows a tumor cell in direct contact with a macrophage displaying active MENA transcription, and a tumor cell not in contact with a macrophage displaying no active MENA transcription site. The addition of DAPT to the tumor cell-macrophage co-culture prevented the macrophage induction of MENA transcription, even when the tumor cells were in direct contact with macrophages (Fig. 4C).
In addition, we measured the probability of active MENA transcription as a function of the distance of a tumor cell to the nearest macrophage in control conditions or in the presence of DAPT ( Fig. 4D and E). When tumor cells are touching macrophages (0 μ m distance), the probability of active MENA transcription is ~0.65, a high probability of transcription. As the distance becomes greater than 35 μ m away, there is zero probability that the tumor cell has an active MENA transcription site. In the presence of DAPT treatment, tumor cells show a scattered pattern of low level transcription versus distance from a macrophage indicating that by blocking Notch signaling we have inhibited the macrophage contact induced MENA transcription.
These FISH data and the qPCR data above indicate that when tumor cells come into direct contact with a macrophage there is a rapid Notch-dependent induction of MENA transcription.
To Mena INV and Notch1 are required for invadopodium-driven transendothelial migration. The above results demonstrate that macrophage -induced tumor cell invadopodium formation requires Notch1 signaling. We also know from previous work that macrophages induce transendothelial migration of tumor cells both in vitro 13 and in vivo 2 , and that Mena INV is required for transendothelial migration 1 . Therefore, we hypothesized   (Fig. 6A and B). Therefore, Mena INV is the key isoform that regulates macrophage-induced invadopodia.
To determine the role of Notch1 in tumor cell intravasation-directed transendothelial migration (iTEM) activity, we either depleted Notch1 in tumor cells with siRNA or inhibited Notch signaling with DAPT and quantified the tumor cell iTEM activity in the presence and absence of macrophages. Notch1 depletion had no effect on the iTEM activity of tumor cells alone, but either depletion of Notch1 with siRNA or its inhibition with DAPT inhibited macrophage-induced tumor cell iTEM activity (Fig. 6C-E).

Notch1 inhibition decreases macrophage-dependent invadopodium formation in vitro and the dissemination of photo-converted tumor cells in vivo. Formation of invadopodia in vitro as
well as invasion in vivo has been well characterized using the rat mammary adencocarcoma MTLn3 cells. Like MDA-MB-231 cells, MTLn3 cells can form invadopodia and undergo migration in vivo when they upregulate Mena INV expression 6,35 . Here, we tested their ability to form macrophage-induced invadopodia. When MTLn3 cells are co-cultured with macrophages, we observed a highly significant increase in formation of invadopodia ( Fig. 7A and B). As an additional method of inhibiting Notch1 signaling applicable to the MTLn3 cells, we treated MTLn3 cells with a Notch1 specific blocking antibody or control isotype IgG. Cells co-cultured with macrophages and treated with the Notch1 blocking antibody demonstrate significantly fewer invadopodia ( Fig. 7A and B). These data demonstrate that the macrophage-induced invadopodia can be stimulated in a different well-characterized invasive and metastatic tumor cell line and this is Notch1 dependent.
We utilized MTLn3 cells expressing photo-convertible Dendra2 to measure the effects of Notch1 inhibition on loss of tumor cells from the primary tumor in vivo as described previously 12,38 . Peri-vascular regions within the MTLn3 -Dendra2 tumors were photo-converted and were tracked every 24 hours over a 72 hour period to measure intravasation as described previously 10,14,38 (Fig. 7 C and D). Mice were treated daily with 1 mg/kg control IgG or Notch1 blocking IgG. In control treated mice, there was a significant decrease every 24 hour period in the percent of photo-converted-Dendra2 cells remaining in the region 10,14,38 . In mice treated with the Notch1 blocking antibody there was a significant reduction in the percent of cells that were leaving the photo-converted regions of the tumor, indicating these tumor cells were less efficient at leaving the primary tumor.
To find out if cells leaving the tumor site were correlated with circulating tumor cells, mice with MTLn3 xenographs were treated with control or Notch1 blocking IgG for 6 hours and the circulating tumor cells were measured after treatment. There was a significant decrease in the number of circulating tumor cells in mice that were treated with the Notch1 blocking antibody compared to control treated mice (Fig. 7E), consistent with the idea that transendothelial migration is involved in CTC (Circulating Tumor Cells) number. These data indicate a role for Notch1 signaling in tumor cells leaving the primary tumor.  endothelial cells 2 . In this study we define the molecular mechanism by which the direct contact between macrophages and tumor cells leads to invadopodium formation and transendothelial migration. We show that these events require the Notch1 receptor on the tumor cell and that Notch1 signaling induces Mena INV expression via activation of transcription. We found that Notch1 is required for transendothelial migration of tumor cells and  1 and, conversely, that Mena INV over-expression drives invadopodium assembly and function, and transendothelial migration 10 .
The relative expression of Mena INV to that of Mena11a is associated with TMEM assembly, metastatic recurrence and death of breast cancer patients 1,39,40 . This is significant because TMEM is the doorway for intravasation of tumor cells into the blood vessels in breast tumors and the number of TMEM sites in a breast tumor is highly predictive of risk of distant recurrence in patients 3,4 . Previous work demonstrated the signals that lead to decreased expression of Mena11a 11 . However, until now the signals responsible for the induction of Mena INV were not known. Here we show that the expression of the Mena INV isoform is induced by activation of Notch1 signaling in tumor cells. In this regard our findings indicate a novel role for Notch1 signaling in the regulation of Mena INV expression and transendothelial migration at TMEM sites.
The induction of invadopodia in tumor cells by macrophages identifies an important step in tumor cell dissemination. Invadopodia are F-actin-rich protrusions on tumor cells capable of degrading extracellular matrix and assisting in tumor cell chemotaxis and migration 41 . Invadopodium initiation has been described during tumor hypoxia emphasizing the importance of the tumor microenvironment in the regulation of invasive protrusion initiation and function 25 . While tumor cells form invadopodia in response to both growth factor and integrin signaling 29,42,43 our studies described herein show, for the first time, that macrophages can initiate invadopodium assembly and this requires Notch signaling uniquely. These results emphasize selectivity differences in how invadopodia are initiated, and the importance of the tumor microenvironment in determining the different invadopodium functions that result from differences in how invadopodia are initiated; chemotaxis during invasive migration involving invadopodium initiation in response to growth factors 44 , fibronectin directed invasion involving invadopodium initiation by integrin beta-1 43 , and transendothelial migration as described here involving invadopodium initiation via Notch signaling in response to heterotypic cancer cell-macrophage contact.
The analysis of the gene expression pattern of migratory and disseminating tumor cells in breast tumors revealed an "invasion signature" that is associated with distant recurrence in breast cancer patients 5,35,37 . A prominent pathway in the invasion signature is the Mena-Cofilin pathway that regulates actin polymerization during chemotaxis and invasion of tumor cells 9,34,45 and which is associated with poor outcome in breast cancer patients 39,40,46 . Investigation of the gene expression pathways of the invasion signature revealed that differentially spliced Mena INV is upregulated, while the invasion-suppressing Mena11a isoform is downregulated in migratory/disseminating tumor cells 6,[8][9][10]35,37,47 . This isoform splicing pattern of Mena (Mena INV-high /Mena11a low ) is associated with directional cell migration towards chemotactic factors such as EGF and HGF, matrix degradation, TMEM assembly and transendothelial migration 8,10,34,35,37,48 , as well as poor outcome in breast cancer patients 5,39,40 . In addition, Mena INV dramatically increases the sensitivity of receptor tyrosine kinases to their ligands EGF, IGF1 and HGF to increase cell protrusion and locomotion of tumor cells toward blood vessels 9,10,49 .
Previous studies have shown that heterotypic interactions among cells surrounding intratumoral vasculature can promote cancer cell dissemination. For example, fibroblast-derived lysly oxidase, a matrix cross-linking enzyme that stiffens collage fibers, driven by myeloid cell-derived TGFβ , promotes cancer metastasis 50 . Another study demonstrated involvement of endosialin-expressing pericytes in cancer cell transendothelial migration and dissemination 51 . Endothelial cells are also actively involved in regulation of cancer cell dissemination. The interaction of CXCR12 (SDF-1), secreted by endothelial cells, with tumor cell expressed CXCR4 is sufficient to stimulate transendothelial migration of the tumor cells 52 . Interestingly, the induction of SDF-1 in endothelial cells seems to be mediated by hypoxia. CXCL12 (SDF-1) can also be expressed by tumor cells and it results in increased macrophage and microvessel density and in vivo invasiveness 53 . Increased macrophage density has been shown to contribute to cancer cell invasiveness and metastasis as shown by several laboratories including ours 1,2,13,54 . The presence of macrophages greatly enhances the ability of both tumor cell lines and primary breast tumor cells to undergo intravasation-directed transendothelial migration (iTEM) and that iTEM requires Mena INV expression and invadopodium formation 1,10,13 . Our study here has added the mechanistic insight into how Mena INV upregulation is achieved in tumor cells and opens the future exploration of how different macrophage ligands might activate Notch signaling on tumor cells to lead to phenotypes associated with tumor metastasis.
Our results are directly relevant to how Mena isoform expression and TMEM number can predict distant recurrence in breast cancer patients [3][4][5]39,40 . The molecular characterization of the Notch1 -dependent Mena INV expression shown here opens the possibility of developing additional markers that might be used in combination with TMEM sites to better predict the risk of distant metastatic recurrence of breast cancer patients and their response to treatment. We think that combining measures of Mena INV expression with the presence of TMEM sites could be useful in determining the relative activity of TMEM sites in transendothelial migration and the response to inhibitors designed to suppress either Mena INV expression and/or TMEM activity. Further work will be required to explore this possibility in patient cohorts of known outcome.
The spatial heterogeneity of expression of Mena INV in primary mammary tumor cells and its consequences has been well described in previous studies 8,10,34,55  with these particular tumor cells 57 . Furthermore, the CD11b positive tumor cell associated myeloid cells have been expression profiled to determine their identity as invasive macrophages 57 . High resolution intravital imaging has shown that transendothelial migration of tumor cells resulting in intravasation in mammary tumors occurs only in association with these invasive macrophages at TMEM. Furthermore, conditional depletion of the macrophages and/or knock out of the macrophage specific VEGF gene completely blocks intravasation in vivo 2 further implicating macrophages in intravasation.
There is significant interest in targeting Notch1 signaling for the treatment of a number of cancers but previous studies have shown chronic Notch inhibition can lead to detrimental secondary effects and in some cases increased vascular tumor development 21,58 . Therefore, further study of the mechanisms of Notch1 signaling in breast cancer progression, as we show here, might lead to the identification of novel therapeutic targets within the Notch1 signaling cascade that might be better tolerated in patients. Assay for Detection of Invadopodia. The 405 gelatin-labeled Mattek dishes were prepared as previously described 43,59 . Tumor cells were plated in complete media for 6 h on the Alexa 405-labeled gelatin dishes. Dishes were fixed and immunostained for cortactin and Tks5 as previously described. Cells were imaged on a wide-field microscope (Inverted Olympus IX70) and images were acquired with a cooled CCD camera (Sensicam QE cooled CCD camera) with a 60 × NA = 1.4 oil immersion objective using IP Laboratory 4.0 software. Invadopodia were detected as punctate structures that were positive for both cortactin and tks5 and capable of degrading Alexa 405-gelatin.

Methods
To detect macrophage-induced invadopodia, MDA-MB-231 cells were serum-starved for 16 h. BAC1.2F5 cells were cell tracker-labeled (CMPTX, Invitrogen). A total of 25 K MDA-MB-231 cells were incubated with 125 K BAC1.2F5 cells in serum-starvation media on 405-labeled gelatin-coated dishes for 6 h, fixed and immunostained for invadopodium markers as described above. For MTLn3 experiments, tumor cells were serum starved for 4 hrs before being plated with BAC1.2F5 cells for 6 hrs as described above.
qPCR. qRT-PCR for Mena splice variants was performed as described previously 35 . Briefly, the data analysis was conducted using the Δ Δ Ct method in which all MENA Ct values in the carcinoma samples were first normalized to GAPDH. qRT-PCR analyses were performed using a SyBR Green kit (Qiagen) and analyzed with a Qiagene Rotor Gene-Q detector and associated software.
Fine Needle Aspiration Biopsy. For primary cancer cells from two human invasive ductal carcinomas fine needle aspiration (FNA) was completed as previously described 1 . Briefly, lumpectomy and mastectomy specimens received at the Albert Einstein College of Medicine/Montefiore Medical Center, Moses and Weiler Divisions for pathological examination were used for FNA-based tissue collection under institutional review board approval. Four to five FNA aspiration biopsies per tumor were performed on grossly visible lesions using 25-gauge needles.
The TN human tissue transplant HT17 was previously described 1,37 . Briefly, the tumor originated from human patient samples and have since only been propagated in SCID mice. Tumors were harvested once they reached 1-to 1.2-cm diameter. Cells were obtained by FNA from human tumors grown in mice. All procedures were conducted in accordance with the NIH regulations and approved by the Albert Einstein College of Medicine animal use committee.

FISH.
In Situ Probes. 48 oligodeoxynucleotide probes for MENA were designed with online Stellaris RNA FISH probe designer (Biosearch Tehchnologies). Each probe was 20 nt long and contained a 5′ amino-modified nucleotide that was chemically coupled to CAL Fluor 610 fluorescent dye. The sequences for probes used to detect MENA mRNA are provided in Supplementary Information.

In Situ Hybridization.
For FISH experiments, MDA-MB-231 cells we starved overnight as described above. Tumor cells were plated in the presence or absence of BAC1.2F5 macrophages that were labeled with cell-tracker red on glass coverslips for 1 h then fixed with 4% paraformaldehyde for 20 minutes at room temperature. After washing away the fixative, the cells were stored in 70% (v/v) ethanol at 4 °C. Prior to hybridiazation, stored coverslips were washed with 1 × PBS and pretreated with 10% formamide/2 × SSC at room temperature for 5 minutes.
Image Acquisition and Analysis. Images were acquired on an Olympus BX61 epi-fluorescence microscope with an UPlanApo 60x, 1.35 numerical aperture oil immersion objective (Olympus). X-Cite 120 PC (EXFO) light source was used for illumination with filter sets 31000 (DAPI), 41001 (Cell-tracker Green), and SP-103v1 (CAL Fluor 610) (Chroma Technology). Vertical stacks of 30 images with a Z step size of 0.2 μ m were acquired using a CoolSNAP HQ camera (Photometrics) with 6.4 μ m pixel size CCD. IPLab (BD Biosciences) software platform was used for instrument control as well as image acquisition. Automated detection and counting of mRNAs was performed by fitting Gaussians to fluorescent spots with FISH-quant as described previously 60,61 . Transendothelial Migration Assay (iTEM). The transendothelial migration assay was performed as described previously 1,13 and briefly described here. The transwell was prepared so that tumor cell transendothelial migration was in the intravasation direction (from subluminal side to luminal side of the endothelium). We measure transendothelial migration as the intravasation-directed transendothelial migration (iTEM) from the tissue to the blood side of the endothelium. To prepare the endothelial monolayer, the underside of each transwell was coated with 50 μ l of Matrigel (2.5 μ g/ml; Invitrogen). Approximately 100,000 HUVEC cells were plated on the Matrigel coated underside of the transwells. Transwells were then flipped into a 24-well plate containing 200 μ l of EGM-2 and monolayers were formed over a 48 hours period. The integrity of the endothelium used in this assay has been validated using electrical resistance and blockade of diffusion of small molecules 13 . Macrophages and tumor cells were labeled with cell tracker dyes (CMFDA, CMPTX from Invitrogen) before the experiment. Then, 15,000 macrophages and 37,500 tumor cells were added to the upper chamber in 200 μ l of DMEM supplemented with 0.5% FBS while the bottom chamber contained EGM-2 supplemented with 3000 u/ml of CSF-1. After 18 hours of transmigration, the tranwells were fixed and stained for ZO-1 as previously described. Transwells were imaged using a Leica SP5 confocal microscope using a 60 × 1.4NA objective and processed using Image J [National Institutes of Health (NIH)]. Quantitation was performed by counting the number of tumor cells that had crossed the endothelium within the same field of view (60x) and represented as normalized values from at least 3 independent experiments. The quantitation of this assay is across at least 3 independent experiments, with 12 fields counted per transwell and transwells done in duplicate for each experiment.

Production of Mena
In vivo circulating tumor cells and photo-converted tumor cell dissemination assay. All procedures involving animals were conducted in accordance with NIH regulations, and approved by the Albert Einstein College of Medicine Animal Use Committee.
MTLn3 cells (parental and Dendra2 expressing) were injected into the mammary glands of 5 to 7 well old SCID mice and allowed to grow tumors for 3 weeks, until the tumors reached approximately 0.8-1.0 cm. Circulating tumor cell count was determined as previously described 36 . Briefly, 1 ml of blood was drawn from the right ventricle of anesthetized mice and plated in α -MEM media supplemented with 20% FBS. Tumor cell were counted as plated cells.
For the in vivo dissemination assay, a mammary imaging window was implanted and dissemination was measured as previously described 35,36,38 . Briefly, two days after the implantation of the imaging window regions of the tumor located adjacent to blood vessels were photo-converted. Approximately 5 regions were converted per tumor. After the photo-conversion and imaging of the regions (time 0), mice were treated with 1 mg/kg control IgG or Notch1 function blocking IgG (R&D Systems). The photo-converted regions were imaged at 24 hr, 48 hr, and 72 hr post treatment to determine the number of tumor cells that remained in the photo-converted primary tumor. The number of photo-converted cells remaining were counted at each time point and represented as the percent of photo-converted cells remaining compared to time 0. It should be noted that we have previously shown in the mouse mammary tumor model used in this study that the disappearance of photo-converted cells from the primary tumor gives rise to a disseminating population of tumor cells that is observed to arrive in distant organ sites seeding new metastases 14 . Furthermore, the tumor cells that disseminate from the photo-converted site are always associated with blood vessels 38 and require functional invadopodia to disseminate 14 . In addition, the imaging method used to document the disappearance of photo-converted cells detects and counts tumor cells that simply "disperse" away from the conversion site ensuring that the disappearance of converted cells requires their dissemination from the primary tumor.