Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Apr 18.
Published in final edited form as: Oncogene. 2013 May 20;33(16):2065–2074. doi: 10.1038/onc.2013.166

Profilin-1 downregulation has contrasting effects on early vs late steps of breast cancer metastasis

Z Ding 1, M Joy 1, R Bhargava 2,3, M Gunsaulus 1, N Lakshman 4, M Miron-Mendoza 4, M Petroll 4, J Condeelis 5, A Wells 1,2,6, P Roy 1,2,6,7
PMCID: PMC3834125  NIHMSID: NIHMS501421  PMID: 23686314

Abstract

Profilin1 (Pfn1), a ubiquitously expressed actin-binding protein, has an indispensable role in migration and proliferation of normal cells. Seemingly contrary to its essential cellular functions, Pfn1’s expression is downregulated in breast cancer, the significance of which is unclear. In this study, expression profiling of Pfn1 in human breast cancer specimens correlates lower Pfn1 expression levels with propensity to metastasize. Xenograft experiments further establish a causal relationship between loss of Pfn1 expression and increased dissemination of breast cancer cells (BCCs) from the primary mammary tumor. BCCs exhibit a hyperinvasive phenotype (marked by matrix metalloproteinase-9 upregulation, faster invasion through collagen matrix) and acquire increased proficiency to transmigrate through endothelial barrier (an obligatory step for vascular dissemination) when Pfn1 expression is suppressed. In Pfn1-deficient cells, hyperinvasiveness involves a phosphatidylinositol 3-kinase-PI(3,4)P2 signaling axis while augmented transendothelial migration occurs in a vascular endothelial growth factor-dependent manner. Contrasting these dissemination promoting activities, loss of Pfn1, however, dramatically inhibits metastatic outgrowth of disseminated BCCs, suggesting that Pfn1 has a key role in the metastatic colonization process. In summary, this study shows that Pfn1 has a dichotomous role in early vs late steps of breast cancer metastasis.

Keywords: profilin-1, breast cancer, metastasis

INTRODUCTION

Dysregulation of the actin cytoskeleton is a hallmark of oncogenic transformation. During tumor progression, carcinoma cells acquire invasive properties that facilitate their dissemination to distant sites. This is a result of altered expression and/or activity of a wide range of actin-binding proteins, and deregulation of signaling pathways that control these actin-binding proteins. Profilin-1 (Pfn1), an evolutionarily conserved actin-binding protein and an essential control element for actin polymerization in cells, has been previously reported to be downregulated in various adenocarcinomas (breast, hepatic, pancreatic, bladder), and when overexpressed, it can also suppress the tumorigenic ability of breast cancer cells (BCCs).1–4 Although these findings have spawned interest in Pfn1’s possible involvement in cancer, whether Pfn1 downregulation causally relates to tumor progression has remained to be determined.

Pfn1 binds to three major classes of ligands: actin, poly-phosphoinositide-based lipids and a large number of proteins with poly-L-proline motifs, including those involved in the regulation of actin dynamics/organization, cell–cell adhesion, membrane trafficking, nuclear export of actin and gene splicing/transcription.5–7 Ablation of Pfn1, either alone or together with Pfn2 (a close homolog of Pfn1 that is primarily expressed in neuronal systems in vertebrates), leads to impaired proliferation and motility/invasion of single cells and developmental arrest and embryonic lethality in multi-cellular organisms.8–16 Pfn1’s interactions with actin and poly-L-proline ligands promote actin polymerization and consequently, serve important role in the execution of membrane protrusion and cytokinesis during cell migration and division, respectively.8,9

Contrasting the conventional pro-migratory function of Pfn1 in almost all physiological contexts, we previously demonstrated that acute depletion of Pfn1 can lead to faster migration of normal human mammary epithelial cells (HMECs) and BCCs, at least, in two-dimensional surface motility assays. Pfn1’s inhibitory effect on BCC motility is mediated mainly by its poly-phosphoinositide interaction.17,18 In light of these in vitro evidence of a contextual role of Pfn1 in cell migration, the present study investigated the in vivo effect of Pfn1 downregulation on breast cancer metastasis, a complex biological process that requires deregulated cell motility.

RESULTS

Lower Pfn1 expression is associated with increased metastatic potential in human breast cancer

The general phenomenon of Pfn1 downregulation in human breast cancer was first reported based on qualitative assessment of Pfn1 expression in several BCC lines and a small set (~40) of primary breast tumors relative to either non-tumorigenic mammary epithelial cell line or normal HMEC.2 However, whether Pfn1 expression has any correlation with the clinicopathological features of human breast tumors was not examined in that study. Therefore, we adopted a semi-quantitative immunohistochemistry approach to analyze Pfn1 expression in a larger panel (N =193) of primary human breast tumors of different histopathological stages (stages I–IV) relative to normal human breast biopsies prepared from reduction mammoplasty samples. Even though there was a considerable variability in Pfn1 immunoreactivity between the various primary tumor samples, overall, Pfn1 expression was found to be lower in primary tumors compared with epithelial cells in normal mammary glands (Figure 1a). According to H-score analyses, all major molecular subtypes (luminal (ER + and/or PR +, HER2 −), luminal/HER2 +, HER2-enriched (ER −, PR −, HER2 +) and triple-negative (ER −, PR −, HER2 −)) of breast cancer exhibit Pfn1 downregulation (Figure 1b; ER, estrogen receptor; PR, progesterone receptor; HER2, human EGF receptor-2), with most robust downregulation seen in those associated with distant metastasis (Figure 1c). In general, we observed an increasing trend in tumor population with negligible to very weak Pfn1 expression (H score: 0–100) with the progression of tumor stage (Figure 1d). Furthermore, tumors associated with marked lymph node infiltration (N score in TNM (Tumor, Node, Metastasis) classification: 2–3) exhibited a statistically significant ~20% lower average Pfn1 expression than tumors with no or limited lymph node involvement (N score: 0–1) (Figure 1e). Overall, these data indicate that lower Pfn1 expression correlates with increased metastatic potential in human breast cancer.

Figure 1.

Figure 1

Open in a new tab

Lower Pfn1 expression is correlated with increased metastatic propensity in human breast cancer. (a) Tissue microarray consisting of normal human breast biopsies (arrows show mammary gland) and primary tumors from breast cancer patients (below) were stained with anti-Pfn1 antibody (Bar =100 μm). (b, c) Bar graphs that compare the average H-scores of Pfn1 immunoreactivity in different molecular subtypes of breast cancer (luminal (ER + and/or PR +, HER2 −), n =127; luminal/HER2 + (ER + and/or PR +, HER2 +), n =19; HER2-enriched (ER −, PR −, HER2 +), n =9; triple-negative (TN: ER −, PR −, HER2 −), n =34) (panel b) and non-metastatic (n =180) vs metastatic (n =13) (panel c) breast tumors relative to normal breast tissue (n =34). (d) A histogram that shows the relative distribution of tumors with two ranges of H-score of Pfn1 immunoreactivity (0–100: negligible to weak; 101–200: moderate to strong Pfn1 expression) for different histopathological stages (H-score data for normal breast is shown alongside for comparison) (e) A bar graph that compares the average H scores of Pfn1 immunoreactivity in tumors associated with marked lymph node infiltration (N score: 2–3, n =30) vs those with no or mild lymph node involvement (N score: 0–1, n =150).

BCCs acquire hyperinvasive phenotype upon loss of Pfn1 expression and this requires phosphatidylinositol 3-kinase (PI3K) signaling

The foregoing clinical correlation data further prompted us to study whether there is any causal relationship between loss of Pfn1 expression and breast cancer metastasis. Specifically, we used MDA-MB-231 (MDA-231: a widely used triple-negative metastatic BCC line of basal subtype) cells to examine whether and how stable suppression of Pfn1 expression alters two fundamental motile activities that promote metastatic dissemination, including extracellular matrix (ECM) invasion and migration through endothelial cell (EC) barrier (an obligatory step for tumor cells’ entry into or exit from circulation). During metastasis, tumor cells encounter, at least, two major types of ECM that differ in their biophysical and biochemical properties, including collagen IV/laminin-rich basement membrane (BM—encapsulates the primary tumor and underlines blood vessels) and collagen-I-rich stroma.19 We previously showed that acute depletion of Pfn1 enhances matrigel (a reconstituted BM) invasion of MDA-231 cells.20 As biophysical and biochemical properties of ECM greatly influence cell motility, we asked whether Pfn1 depletion has similar effect on BCC invasion through fibrillar collagen-I. We found that stable knock-down (KD) of Pfn1 increases the random invasion of MDA-231 cells through collagen-I by >1.5-fold and this can be rescued by re-expression of a shRNA-resistant version of green fluorescent protein (GFP)-tagged Pfn1 (Figure 2a, b). The effect of Pfn1 KD on collagen-I invasiveness was even more pronounced in a directional invasion assay when cells were allowed to invade from a stiffer inner matrix to a softer outer matrix (better mimics the in vivo emigration of carcinoma cells from a tumor into the surrounding stroma) (Figure 2c). In this assay, Pfn1 KD increased the number of invading cells and the maximum invasion distance by 5.8- and 2.2-folds, respectively (Figure 2d, e). Similarly, collagen-I invasiveness of MDA-MB-468 (MDA-468), another human BCC line of basal subtype, increased by ~two-fold when Pfn1 expression was suppressed (Supplementary Figure S1). These data demonstrate that loss of Pfn1 induces hyperinvasiveness in BCCs.

Figure 2.

Figure 2

Open in a new tab

Pfn1 depletion induces hyperinvasiveness in MDA-231 cells. (a) Pfn1 immunoblot of total lysates derived from MDA-231 cells stably expressing either luc (control)- or Pfn1-shRNA (#1, 3—two independent Pfn1-shRNA-bearing clones). GAPDH blot serves as the loading control. (b) A bar graph summarizing the average invasion speed of Pfn1 KD cells, without or with rescue by GFP-Pfn1 re-expression, relative to that of control luc-shRNA expression in a random collagen invasion assay (n =48 (control), 31 (each Pfn1 KD clone), 15 (rescue group)). (c) Representative images of a nested collagen invasion assay showing the directional invasion of control and Pfn1 KD cells (clone #1) from the inner to the outer collagen-I matrix (the white line represents the interface of the two matrices). Maximum intensity projections from confocal stacks of TOTO3-labeled cells are shown here (Bar =100 μm). (d, e) Bar graphs comparing the number of invading cells (panel d) and distance (panel e) invaded in the control vs Pfn1 KD group in the nested collagen invasion assay.

Matrix metalloproteinases (MMPs) have a crucial role in ECM degradation during BM permeation and stromal invasion of malignant tumor cells. Therefore, we further examined the effect of Pfn1 depletion on secretion of three major types of MMPs that have been correlated with the invasive phenotype of BCCs, including MMP1 (degrades collagen-I), MMP2 and MMP9 (the latter two are gelatinases, which degrade BM matrix). Pfn1 KD in MDA-231 cells caused a marked elevation in MMP9 secretion and this effect could be blocked by re-expression of GFP-Pfn1 (Figure 3a); however, secretion of MMP1 or MMP2 remained unaffected (Supplementary Figure S2). MMP9 upregulation upon Pfn1 KD occurred at the transcriptional level (Figure 3b), at least, partly involving increased promoter activity (Supplementary Figure S3). Correlated with its effect on MMP9, Pfn1 KD also stimulated the actual gelatinous matrix degradation ability of MDA-231 cells by ~4.5-fold (Figure 3c, d). These data demonstrate that Pfn1 can regulate ECM degradation activity of BCCs.

Figure 3.

Figure 3

Open in a new tab

Pfn1 depletion promotes MMP9 secretion and matrix degradation ability of MDA-231 cells. (a) Gelatin zymography assay performed with MDA-231 conditioned media (CM) shows enhanced MMP9 secretion upon Pfn1 KD that can be reversed by re-expression of GFP-Pfn1. (b) RT–PCR showing the effects of the indicated transfections on MMP9 mRNA in MDA-231 cells (GAPDH RT–PCR serves as the loading control). (c, d) Panel c shows the representative images of fluorescein isothiocyanate-gelatin matrix degradation (marked by dark spots) by control vs Pfn1 KD cells in culture (Bar =20 μm). Quantification shows that average area of degraded matrix is robustly increased when Pfn1 expression is knocked down (bar graph in the left of panel (d) even though the average number of cells per field between the two groups is not statistically significant (bar graph on the right of panel (e). NS, not significant.

Tumor malignancy is caused to a significant extent by dysregulation of membrane phosphoinositide-derived signaling, particularly involving pathways downstream of activated PI3K. Signaling downstream of PI3K/AKT is also an important arm for transcriptional control of MMP9.21 Recently, we demonstrated that Pfn1 depletion increases two-dimensional motility of MDA-231 cells by promoting membrane targeting of lamellipodin (Lpd)-Ena (enabled)/VASP (vasodilator-stimulated phosphoprotein) protein complex (regulates F-actin architecture and lamellipodial dynamics) to the leading edge.17 Correlated with the requirement of PI(3,4)P2 (a PI3K-generated phosphoinositide) for membrane recruitment of Lpd,22,23 we further showed that Pfn1 negatively regulates membrane accumulation of PI(3,4)P2 at the leading edge.17 However, several gaps still need to be addressed. First, it is not known whether Pfn1 depletion alters the dynamics of PI3K signaling that may explain the changes in PI(3,4)P2 accumulation. Second, whether hypermotility of BCCs induced by Pfn1 depletion actually requires PI3K-PI(3,4)P2 signaling has not been directly tested. Third, two-dimensional surface motility assays, while informative, do not accurately mimic the biochemical and biophysical aspects of cell invasion through ECM as occurs in vivo. Therefore, the relevance of PI3K-PI(3,4)P2-Lpd pathway for hyperinvasiveness of Pfn1 KD cells through physiological matrix (for example, fibrillar collagen-I) is not clear. To address these gaps, we first examined the effect of Pfn1 KD on the kinetics of EGF-induced AKT phosphorylation (an event that occurs downstream of activated PI3K) in MDA-231 cells. Even though the peak AKT phosphorylation was found to be similar between control and Pfn1 KD cells, the attenuation of phosphorylation signal was delayed significantly in Pfn1 KD cells (Figure 4a). This suggests that Pfn1 depletion increases the sustenance of PI3K signaling and, clearly, this change is consistent with our previously observed effect of Pfn1 KD on PI(3,4)P2 accumulation.17 Inhibition of PI3K signaling by LY294002 markedly reduced MMP9 differential between control and Pfn1 KD cells (note that PI3K inhibition had no effect on MMP2 secretion; Figure 4b) and abrogated the hyperinvasive behavior of Pfn1 KD cells (Figure 4c). Similarly, blockade of PI(3,4)P2 signaling by transiently overexpressing either PH (pleckstrin-homology) domain of TAPP1 (a PI(3,4)P2-binding protein that should compete with other natural ligands of PI(3,4)P2, such as Lpd, for binding to PI(3,4)P2) or INPP4B (inositol polyphosphate 4-phophatase—a 4′-lipid phosphatase that selectively hydrolyzes and depletes PI(3,4)P224,25) markedly reduced the average invasion speed of Pfn1 KD cells (Figure 4d). Finally, when Lpd (a downstream target of PI(3,4)P2) was silenced, hyperinvasiveness of Pfn1 KD cells was also abrogated (Figure 4e; the immunoblot in Supplementary Figure S4 shows silencing of Lpd). Collectively, these data demonstrate that loss of Pfn1 promotes invasiveness of BCCs requiring a PI3K-PI(3,4)P2-Lpd signaling axis.

Figure 4.

Figure 4

Open in a new tab

PI3K signaling has an essential role in enhanced MMP9 secretion and hyperinvasiveness of MDA-231 cells induced by Pfn1 depletion. (a) Kinetics of T308 phosphorylation of AKT in control vs Pfn1 KD cells in response to EGF stimulation (phospho-AKT levels were normalized to the respective loading controls). (b) Gelatin zymogram of conditioned media from control or Pfn1 KD cells following treatment with either 25 μM LY294002 (PI3K inhibitor) or dimethyl sulfoxide (DMSO; vehicle control). (c) A bar graph summarizing the average invasion speed of control vs Pfn1 KD cells following either LY294002 or DMSO treatment (all data are normalized to the invasion speed vehicle of control cells in DMSO-treated conditions; 15–50 cells were analyzed in each group). (d) A bar graph summarizing the average invasion speed of Pfn1 KD cells following transient expression of either GFP (control, n =32) or GFP-PH-TAPP1 (n =30) or GFP-INPP4B (n =35). (e) A bar graph showing the average invasion speed of control and Pfn1 KD cells with or without simultaneous Lpd KD (24–64 cells were analyzed in each group). All invasion experiments were performed 2–3 times and data were pooled.

Loss of Pfn1 expression promotes transendothelial migration of BCCs in a vascular endothelial growth factor (VEGF)-dependent manner

Next, to determine whether loss of Pfn1 has any effect on the ability of BCCs to cross through vascular EC barrier, we performed a transendothelial cell migration (TECM) assay. In this assay, Pfn1 KD increased transmigration of MDA-231 through HUVEC (human umbilical vein EC) by nearly twofold, and this was rescued by re-expression of GFP-Pfn1 (Figure 5a). Tumor cells must disrupt intercellular junctions of EC in order to transmigrate. Immunostaining for β-catenin revealed that MDA-231 cells induce severe junctional disruption of HUVEC as expected, but the extent of junctional disruption of HUVEC seemed to be more pronounced when in co-culture with Pfn1 KD cells than with control cells (Figure 5b). This led us to postulate that Pfn1 KD promotes secretion of certain factors that compromise EC barrier function. Antibody array-based analyses of conditioned media (CM) from MDA-231 culture showed that Pfn1 KD was associated with increased secretion of several pro-metastatic factors, including uPA (urokinase plasminogen activator), interleukin-8, VEGF and endothelin-1 (Figure 5c). As VEGF is one of the most potent disruptors of EC–cell junctions, we next silenced VEGF expression, and this completely abrogated the TECM differential between control and Pfn1 KD cells (Figures 5d–e). This suggests that Pfn1 depletion confers BCCs increased proficiency to cross through EC barrier in a VEGF-dependent manner.

Figure 5.

Figure 5

Open in a new tab

Pfn1 KD promotes transendothelial migration ability of MDA-231 cells. (a) A bar graph showing that average number of transmigrated MDA-231 cells through HUVEC monolayer (per 10 × field of observation) for the indicated experimental perturbations (the inset shows the representative images; Bar =100 μm). (b) β-catenin immunostaining of HUVEC without or with co-culture of MDA-231 cells for the indicated time period (Bar =50 μm). (c) Antibody microarray analyses (two different exposures of blots are shown) of conditioned media from control vs Pfn1 KD MDA-231 cells show differential secretion of interleukin-8, VEGF, endothelin-1 and uPA between the two groups (the fold changes for these factors, averaged from five independent experiments, are tabulated on the right). (d) RT–PCR data show effective silencing of VEGF expression by RNAi (GAPDH RT–PCR serves as the loading control). (e) A bar graph summarizing the effect of VEGF silencing on the TECM of control and Pfn1 KD cells. All TECM experiments were repeated at least three times for statistical comparison.

Loss of Pfn1 expression enhances metastatic dissemination but suppresses colonization ability of BCCs

Based on the foregoing in vitro data, we further postulated that loss of Pfn1 promotes dissemination of BCCs from mammary tumor in vivo. As metastasis is dictated by both dissemination and growth/survival properties of cancer cells, commonly used end point metastasis measurement in xenograft models is not appropriate for selective assessment of tumor cell dissemination. Therefore, we deconstructed the metastasis cascade into three major steps, including vascular dissemination of cancer cells from the primary tumor (mediated by the combined act of invasion and intravasation), seeding of circulating tumor cells into the metastatic organ (mediated by extravasation) and metastatic colonization of seeded cells (influenced mainly by survival and growth of tumor cells at the ectopic site). First, to determine the effect of Pfn1 depletion on vascular dissemination of BCCs from the primary tumor, we performed orthotopic xenograft experiments in athymic nude mice. In these experiments, mammary tumors developed in five out of five and five out of six mice following mammary fat pad injection of control and Pfn1 KD MDA-231 cells, respectively. Even though there was no statistical difference (P =0.65) in the average tumor burden between the two groups of animals (Figure 6a), blood cancer burden (BCB—indirectly assessed by the ratio of human DNA content to mouse DNA content in cells isolated from whole blood26), a measure of circulating tumor cells in blood, was ~2.5-fold higher (P =0.02) in the Pfn1 KD group (Figure 6b). When we normalized BCB for each animal to its corresponding tumor volume (to account for the possible effect of animal-to-animal variation in primary tumor size on BCB), the average fold increase in the normalized BCB induced by Pfn1 KD was even higher (equal to 3.6-fold; P =0.04; Figure 6b). These data demonstrate that loss of Pfn1 promotes the escape of BCCs from the primary mammary tumor.

Figure 6.

Figure 6

Open in a new tab

Loss of Pfn1 promotes vascular dissemination of MDA-231 cells in xenograft model. (a) A schematic of orthotopic injection of BCCs in athymic nude mice is shown on the left. The plot on the right shows the kinetics of mammary tumor growth following injection of control (n =5 animals) vs Pfn1 KD (n =6 animals) cells. (b) A dot plot comparing the BCB of animals in the control vs Pfn1 KD groups (n =5 animals each); the average BCB calculated for control animals was set to 1 to display the fold change (See Materials and methods for BCB quantification). MFP, mammary fat pad.

Next, we performed a lung colonization assay where we injected control or Pfn1 KD MDA-231 cells directly into the lateral tail vein of athymic nude mice. In this in vivo assay, we did not find any statistically significant difference in the number of extravasated cells in lungs between the control and Pfn1 KD groups at 24 h after injection (Figure 7a, b), thus suggesting that initial metastatic seeding of circulating BCCs into lungs was not affected by Pfn1 depletion. However, at 1 week after injection, not only the number of metastatic foci was strikingly fewer in animals in the Pfn1 KD group, but the outgrowth of those foci was strongly impaired when compared with the control group of animals (Figure 7c). At 4 weeks after injection, control mice developed extensive pulmonary macro-metastatic nodules as expected, but those inoculated with Pfn1 KD cells either failed to exhibit such macro-metastatic lesions or, in some cases, developed nodules, which were strikingly fewer in number and much smaller in size than those seen in control animals (Figure 7d). Analyses of lung histology showed that animals injected with Pfn1 KD cells developed about 5.37-fold fewer metastatic foci and had 28.67-fold less area of lung affected by metastases than control animals (Figure 7e–g). These data demonstrate that Pfn1 has an obligatory role in distal colonization of disseminated BCCs.

Figure 7.

Figure 7

Open in a new tab

Loss of Pfn1 causes a major impairment in metastatic outgrowth of MDA-231 cells. (a) A schematic of in vivo lung colonization assay through tail-vein injection of BCCs. Lungs harvested 24 h after tail-vein injection of control (n =4 animals) vs Pfn1 KD cells (n =5 animals) were imaged by a wide-field fluorescence microscope to detect red fluorescent protein-positive extravasated tumor cells on lung surface (Bar =100 μm). (b) A bar graph summarizes the average percentage of area of lung surface represented by extravasated cells in the two groups (NS, not significant). (c) Representative fluorescence images of lungs harvested 1 week after tail-vein injection of tumor cells reveal a striking difference in the number of metastatic foci and growth between the control and Pfn1 KD groups (five animals were used in each group) (Bar =100 μm). (d–g) Gross morphology of lungs at 4 weeks after tail-vein injection of control vs Pfn1 KD cells (panel d—arrows indicate metastatic nodules). Panel e shows the representative hematoxylin and eosin staining of lung histosections in two groups of animals (areas outlined by dotted lines represent metastatic foci; Bar =100 μm). The bar graphs in panels f and g compare the average number of metastatic foci per 10 × field and the percentage of lung area affected by micro-metastasis, respectively, between the control and Pfn1 KD groups (these data are summarized from analyses of eight and nine animals in the control and Pfn1 KD groups, respectively).

DISCUSSION

The question as to whether Pfn1, found downregulated in breast cancer, confers functional advantages to tumor cells for dissemination and metastasis is critical to understanding this association. We herein demonstrate that while loss of Pfn1 expression promotes dissemination of BCCs from the primary tumor likely through facilitating invasion and intravasation steps of metastasis, successful metastatic colonization of disseminated BCCs absolutely requires Pfn1’s presence. In essence, a near complete Pfn1 depletion appears to confer functional advantage to metastatic BCCs only at the primary tumor site but not at the metastatic site. Interestingly, when we compared Pfn1 expression between the primary tumor and the matched lung metastases for a small subset (n =7) of patient samples, none of the lung metastases actually exhibited any further Pfn1 downregulation when compared with the corresponding primary tumor and, in a few cases, Pfn1 expression in metastases was even marginally higher (Supplementary Figure S5). Even though the small sample size used in this preliminary analysis prevents definitive interpretation, this observation is still consistent with the requirement for, at least, some levels of intracellular of Pfn1 for metastatic colonization of BCCs, as suggested by our in vivo studies. We previously showed that even a moderate two-fold overexpression of Pfn1 in MDA-231 cells (this, in principle, should restore the Pfn1 content in MDA-231 cells to a level comparable with that in normal HMEC) dramatically suppresses their tumorigenic and invasive potential.20 In light of those findings, the clinical correlation and in vivo data reported herein lead us to speculate, at least, two possible scenarios with regard to how Pfn1 downregulation might be coupled to BC progression in humans. One possibility is that Pfn1 expression is downregulated in mammary tumor cells as the disease progresses but not below a critical level that would impair their distal colonization ability. Alternatively, a subset of highly aggressive tumor cells experience a more drastic Pfn1 downregulation (possibly even to the extent of a near complete loss) compared with the bulk of the primary tumor, and this further facilitates their disseminative potential. However, similar to the recently described phenomenon of plasticity of E-cadherin expression in primary tumor vs distant metastases,27,28 Pfn1 expression in the disseminated cells could possibly increase to some extent at the metastatic sites (as seen in four out of seven patient samples in our preliminary analyses—see Supplementary Figure S5) to allow colonization.

A causal relationship between loss of Pfn1 and acquisition of increased disseminative ability of BCCs in vivo not only supports our previous in vitro demonstration of the contextual role of Pfn1 in cell migration but also establishes the biological significance of Pfn1 downregulation in breast cancer. We have identified two potential mechanisms by which loss of Pfn1 can augment motile behavior of BCCs. We linked a PI3K-PI(3,4)P2-Lpd signaling axis to the hyperinvasiveness of Pfn1-deficient cells, providing a direct evidence of Pfn1’s coupling of phosphoinositide signaling to cell motility as once speculated based on biochemical evidence of Pfn1’s ability to inhibit PI(4,5)P2 hydrolysis.29,30 Second, we showed that loss of Pfn1 promotes the ability of BCCs to cross through vascular endothelial barrier by modulating VEGF secretion. In addition to these molecular changes, we showed that secretion of several other pro-metastatic factors (uPA, interleukin-8, endothelin-1) are also elevated upon Pfn1 KD. Therefore, it is likely that Pfn1 regulates BC metastasis by affecting multiple pathways. Given that Pfn1 also has nuclear activities,31 perturbing Pfn1 expression could potentially cause a major change in the gene expression pattern of BCCs partly accounting for the pleotropic effects seen in our experiments.

One of the most unexpected findings of the present study is the severe colonization defect of BCCs induced by Pfn1 depletion. As in the orthotopic model, primary tumor growth was not significantly altered by Pfn1 KD, it further suggests that tumor microenvironment has a profound influence on how BCCs respond to loss of Pfn1 expression. Even though the exact mechanisms by which Pfn1 deficiency arrests the metastatic outgrowth of disseminated BCCs are not clear and need further investigation, we can speculate that this may be, at least, partly related to Pfn1-dependent cytoskeletal changes in BCCs. There is emerging evidence that metastatic outgrowth of BCCs is critically influenced by signaling triggered by the ECM component of tumor microenvironment and this involves changes in actin cytoskeletal architecture.32,33 It has been shown that preventing actin stress fiber assembly or blocking integrin signaling can dramatically inhibit metastatic outgrowth of BCCs. Consistent with Pfn1’s essential role in actin polymerization, we previously demonstrated that Pfn1 KD causes reduction in filamentous actin and focal adhesions, and conversely, Pfn1 overexpression stimulates assembly of actin stress fibers and focal adhesions with concomitant increase in FAK (focal adhesion kinase) phosphorylation (an indicator of adhesion-induced signaling) in BCCs (including MDA-231 cells) and/or HMEC.20,34 Therefore, it will be interesting to determine in the future whether hyperactivating integrin signaling or stimulating actin cytoskeletal assembly rescues metastatic colonization defect of Pfn1-deficient BCCs. It is also possible that Pfn1-deficient BCCs experience a survival disadvantage at the metastatic site. This could be a consequence of either attenuated cell-matrix adhesion signaling promoting anoikis or alteration in gene expression pattern (possibly linked to reduced nuclear activities of Pfn1) leading to induction of an unsupportive pre-metastatic niche. Finally, we cannot absolutely rule out an additional possibility that Pfn1 depletion puts circulating tumor cells to a survival disadvantage in the face of hemodynamic stress. If this is true, we may have not only underestimated the fold increase in vascular dissemination caused by Pfn1 depletion but a comparable readout in the initial lung seeding between control and Pfn1 KD cells must also mean that Pfn1 KD promotes BCC extravasation, as to be expected from our TECM data in vitro. These various possibilities need to be explored in detail in future studies.

In conclusion, this study uncovers a unique dichotomous role of Pfn1 in early vs late steps of breast cancer metastasis.

MATERIALS AND METHODS

Animal experiments

For inducing mammary tumor formation, 2 ×106 MDA-231 (source: ATTC) cells, suspended in 50 μl of 1:1 (v/v) mix of phosphate-buffered saline and growth factor-reduced matrigel (BD Biosciences, Bedford, MA, USA) were injected into the right inguinal mammary fat pad of 4–5-week-old female athymic nude mice. For lung colonization experiments, 1 ×106 cells suspended in 100 μl sterile phosphate-buffered saline were directly injected into the lateral tain vein of 4-week-old female nude mice. Because of low abundance of circulating tumor cells and limited availability of blood for analysis (<1 ml) in mouse models, we employed a commonly used PCR-based approach to obtain an indirect measure of BCB. Specifically, blood was drawn by performing cardiac puncture following which red blood cells were lysed and genomic DNA was extracted from the cell pellet using a commercial kit (Promega, Madison, WI, USA). PCR reactions were performed on the genomic DNA as the template with primers against human Alu (amplifies DNA segment from circulating human BCCs) and mouse GAPDH (glyceraldehyde 3-phosphate dehydrogenase) sequences (amplifies DNA segment from mouse blood cells) (see Supplementary Table S1 for PCR details). PCR products were run on a 2% agarose gel, and BCB was determined by normalizing the intensity of human Alu to that of mouse GAPDH (serves as a loading control) PCR bands. Metastases in lung were identified by either histological staining of paraffin sections with hematoxylin and eosin or fluorescence imaging of red fluorescent protein-positive tumor cells on whole lung surface. At least two 10 × fields/histosection/lobe of the lung in each animal were analyzed for quantification of metastasis. All animal experiments were performed in compliance with an approved protocol by the Institutional Animal Care Committee of the University of Pittsburgh.

Cells, plasmids and siRNA

MDA-231 and MDA-468 cells were cultured in Eagle’s minimal essential medium and Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum and antibiotics, respectively. Details of siRNA-resistant GFP-Pfn1 expression vector, shRNA vectors targeting Pfn1 and luciferase (control) and stable transfection of MDA-231 cells have been previously described.17,35 Plasmids encoding GFP-PH-TAPP1 and GFP-INPP4B were kind gifts of Dr Tamas Balla (National Institutes of Health) and Dr Christina Mitchell (Monash University), respectively. For VEGF silencing, cells were transfected with 100 nM of either VEGF siRNA (Santa Cruz Biotechnology, Dallas, TX, USA) or smart-pool control siRNA (ThermoFisher, Hudson, NH, USA), and subsequent experiments were performed 48 h after transfection.

Protein extraction/immunoblotting

Total cell lysate was extracted by modified RIPA buffer (50 mM Tris–HCl-pH 7.5, 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 0.3% sodium dodecyl sulfate, 2 mM EDTA) supplemented with 50 mM NaF, 1 mM sodium pervanadate and protease inhibitors. For immunoblotting, antibodies were used at the following concentrations: Pfn1 (1:500—monoclonal (Santa Cruz Biotechnology)), GAPDH (1:2000—polyclonal (Santa Cruz Biotechnology)), Lpd (1:1000—polyclonal (Santa Cruz Biotechnology)), vimentin (1:1000—monoclonal (Pharmingen, San Jose, CA, USA)) and MMP1 (1:1000—monoclonal (Millipore, Billerica, MA, USA)).

Pfn1 immunohistochemistry

Conventional histosections or tissue microarrays (0.6 mm core—each sample was spotted in triplicates) were deparaffinized in clearite and rehydrated. After blocking endogenous peroxidase and performing antigen retrieval (10 mM citrate buffer, pH 6.0 at 95 °C for 25 min), tissue slides were blocked in goat serum for 20 min and stained with Pfn1 antibody (Novus Biologicals, Littleton, CO, USA) at 1:100 for 30 min followed by biotinylated goat-anti-rabbit secondary antibody (Jackson ImmunoResearch, West Grove, PA, USA) for 30 min. Following incubation of slides with ABC kit (Vector laboratories, Brulingame, CA, USA) for 45 min, chromogenic reaction was performed with diaminobenzidine for 60–70 s. All tissue sections were counterstained with hematoxylin before mounting. Pfn1 immunoreactivity was semi-quantitatively evaluated by a pathologist using the standard H-scoring technique with the conventional intensity categories: 0 (no detectable staining), 1 + (weak staining), 2 + (moderate staining), and 3 + (intense staining). For each sample, the H-score was equal to Σi × Pi, (where i is the intensity value ranging from 0 to 3 and Pi is the percentage of cells stained at intensity i).

Transendothelial migration assay

Early passage (within three passages) HUVEC were grown to confluence on the upper side of collagen-coated 3 μm-pore transwell inserts (BD Biosciences). Fifty thousand MDA-231 cells were seeded on top of HUVEC monolayer and allowed to migrate toward a serum gradient for 24 h. At the end of the experiment, cells on the upper side of transwells were removed, and transmigrated MDA-231 cells located on the bottom side of the transwells were fixed and imaged at multiple fields (at least 5 fields/transwell from 2–3 replicate transwells/group). Immunostaining of β-catenin in HUVEC/MDA-231 co-culture was performed according to the published protocol.36 Images of stained cells were acquired with a ×20 objective on an Olympus IX-71 epifluorescence microscope (Olympus America, Center Valley, PA, USA).

Antibody array

CM of serum-starved MDA-231 cells was collected and concentrated in centrifugation filter units (molecular weight cutoff: 3 kDa; Millipore) at 2500 g for 40 min before analyzing on a commercially available human angiogenesis antibody array (R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s protocol. Chemiluminescence signals on the array spots were quantified using ImageJ software (NIH Image, Bethesda, MD, USA).

Collagen invasion assay

Random and nested collagen gel invasion assays were carried out according to the published protocols.9,37

Reverse transcriptase (RT)–PCR

Total RNA was extracted from cells using the RNeasy Mini Kit (Qiagen). RT–PCR was performed with 1.5 μg of RNA using the OneStep RT–PCR Kit (Qiagen, Valencia, CA, USA). PCR products were run on a 1% agarose gel.

MMP assays

Cells, initially cultured in complete growth media for 48 h, were serum-starved for 21 h before harvesting and concentrating the CM (the seeding density was optimized to ensure sub-confluent culture at the day of experiment). MMP1 secretion was determined by SDS–PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis)/immunoblot analyses of CM. To measure MMP2/9 secretion, gelatin zymography of CM was performed according to our published protocol.9 For assessing MMP9 promoter activity, cells were co-transfected with a reporter construct expressing GFP under MMP9 promoter (this was created by replacing the luciferace reporter by enhanced GFP sequence in the original 2.1 kb pGL3-MMP9 promoter vector gifted by Dr Chunhong Yan, Albany Medical College) and pCFP vector (plasmid cyan fluorescent protein; transfection control). GFP and CFP fluorescence values of cells, recorded 48 h after transfection, were normalized (a measure of the promoter activity) on a cell-by-cell basis and then averaged for statistical comparison.

Gelatin degradation assay

This assay was performed with some modification from the published protocol.38 Briefly, glass coverslips were washed and coated with 0.1% fluorescein isothiocyanate-gelatin and 0.2% porcine gelatin in a 1:9 ratio before incubation with 2 ×104 of either control or Pfn1 KD MDA-231 cells in 24-well plate wells. After 20 h incubation, each well was fixed with 3.7% formaldehyde and mounted onto glass slides before acquiring images for quantification.

Statistics and data representation

All statistical tests were performed with analysis of variance followed by Tukey–Kramer post-hoc test for multiple comparisons whenever applicable, and a P-value <0.05 was considered to be statistically significant.

Supplementary Material

Ding Supplementray data

Acknowledgments

This work was supported by grants from the National Cancer Institute of the National Institute of Health (2R01CA108607-07) and the Magee Women’s Research Institute to PR.

Footnotes

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

References

  • 1.Gronborg M, Kristiansen TZ, Iwahori A, Chang R, Reddy R, Sato N, et al. Biomarker discovery from pancreatic cancer secretome using a differential proteomic approach. Mol Cell Proteomics. 2006;5:157–171. doi: 10.1074/mcp.M500178-MCP200. [DOI] [PubMed] [Google Scholar]
  • 2.Janke J, Schluter K, Jandrig B, Theile M, Kolble K, Arnold W, et al. Suppression of tumorigeniciy in breast cancer cells by the microfilament protein profilin 1. J Exp Med. 2000;191:1675–1685. doi: 10.1084/jem.191.10.1675. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Wu N, Zhang W, Yang Y, Liang YL, Wang LY, Jin JW, et al. Profilin 1 obtained by proteomic analysis in all-trans retinoic acid-treated hepatocarcinoma cell lines is involved in inhibition of cell proliferation and migration. Proteomics. 2006;6:6095–6106. doi: 10.1002/pmic.200500321. [DOI] [PubMed] [Google Scholar]
  • 4.Zoidakis J, Makridakis M, Zerefos PG, Bitsika V, Esteban S, Frantzi M, et al. Profilin 1 is a potential biomarker for bladder cancer aggressiveness. Mol Cell Proteomics. 2012;11:M111 009449. doi: 10.1074/mcp.M111.009449. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Jockusch BM, Murk K, Rothkegel M. The profile of profilins. Rev Physiol Biochem Pharmacol. 2007;159:131–149. doi: 10.1007/112_2007_704. [DOI] [PubMed] [Google Scholar]
  • 6.Karlsson R, Lindberg U. Profilin, an essential control element for actin polymerization. In: Lapplainen P, editor. Actin Monomer Binding Proteins. Chapter 3. Landes Biosciences and Springer; Georgetown, USA: 2006. pp. 29–44. [Google Scholar]
  • 7.Witke W. The role of profilin complexes in cell motility and other cellular processes. Trends Cell Biol. 2004;14:461–469. doi: 10.1016/j.tcb.2004.07.003. [DOI] [PubMed] [Google Scholar]
  • 8.Bottcher RT, Wiesner S, Braun A, Wimmer R, Berna A, Elad N, et al. Profilin 1 is required for abscission during late cytokinesis of chondrocytes. EMBO J. 2009;28:1157–1169. doi: 10.1038/emboj.2009.58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Ding Z, Gau D, Deasy B, Wells A, Roy P. Both actin and polyproline interactions of profilin-1 are required for migration, invasion and capillary morphogenesis of vascular endothelial cells. Exp Cell Res. 2009;315:2963–2973. doi: 10.1016/j.yexcr.2009.07.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Ding Z, Lambrechts A, Parepally M, Roy P. Silencing profilin-1 inhibits endothelial cell proliferation, migration and cord morphogenesis. J Cell Sci. 2006;119(Pt 19):4127–4137. doi: 10.1242/jcs.03178. [DOI] [PubMed] [Google Scholar]
  • 11.Haugwitz M, Noegel AA, Karakesisoglou J, Schleicher M. Dictyostelium amoebae that lack G-actin-sequestering profilins show defects in F-actin content, cytokinesis, and development. Cell. 1994;79:303–314. doi: 10.1016/0092-8674(94)90199-6. [DOI] [PubMed] [Google Scholar]
  • 12.Khadka DK, Liu W, Habas R. Non-redundant roles for Profilin2 and Profilin1 during vertebrate gastrulation. Dev Biol. 2009;332:396–406. doi: 10.1016/j.ydbio.2009.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kullmann JA, Neumeyer A, Gurniak CB, Friauf E, Witke W, Rust MB. Profilin1 is required for glial cell adhesion and radial migration of cerebellar granule neurons. EMBO Rep. 2011;13:75–82. doi: 10.1038/embor.2011.211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Sato A, Khadka DK, Liu W, Bharti R, Runnels LW, Dawid IB, et al. Profilin is an effector for Daam1 in non-canonical Wnt signaling and is required for vertebrate gastrulation. Development. 2006;133:4219–4231. doi: 10.1242/dev.02590. [DOI] [PubMed] [Google Scholar]
  • 15.Severson AF, Baillie DL, Bowerman BA. Formin homology protein and a profilin are required for cytokinesis and Arp2/3-independent assembly of cortical microfilaments in C. elegans. Curr Biol. 2002;12:2066–2075. doi: 10.1016/s0960-9822(02)01355-6. [DOI] [PubMed] [Google Scholar]
  • 16.Verheyen EM, Cooley L. Profilin mutations disrupt multiple actin-dependent processes during drosophila development. Development. 1994;120:717–728. doi: 10.1242/dev.120.4.717. [DOI] [PubMed] [Google Scholar]
  • 17.Bae YH, Ding Z, Das T, Wells A, Gertler F, Roy P. Profilin1 regulates PI(3,4)P2 and lamellipodin accumulation at the leading edge thus influencing motility of MDA-MB-231 cells. Proc Natl Acad Sci USA. 2010;107:21547–21552. doi: 10.1073/pnas.1002309107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Bae YH, Ding Z, Zou L, Wells A, Gertler F, Roy P. Loss of profilin-1 expression enhances breast cancer cell motility by Ena/VASP proteins. J Cell Physiol. 2009;219:354–364. doi: 10.1002/jcp.21677. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Poincloux R, Collin O, Lizarraga F, Romao M, Debray M, Piel M, et al. Contractility of the cell rear drives invasion of breast tumor cells in 3D Matrigel. Proc Natl Acad Sci USA. 2011;108:1943–1948. doi: 10.1073/pnas.1010396108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Zou L, Jaramillo M, Whaley D, Wells A, Panchapakesa V, Das T, et al. Profilin-1 is a negative regulator of mammary carcinoma aggressiveness. Br J Cancer. 2007;97:1361–1371. doi: 10.1038/sj.bjc.6604038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Ho MY, Tang SJ, Chuang MJ, Cha TL, Li JY, Sun GH, et al. TNF-alpha induces epithelial-mesenchymal transition of renal cell carcinoma cells via a GSK3beta-dependent mechanism. Mol Cancer Res. 2012;10:1109–1119. doi: 10.1158/1541-7786.MCR-12-0160. [DOI] [PubMed] [Google Scholar]
  • 22.Krause M, Leslie JD, Stewart M, Lafuente EM, Valderrama F, Jagannathan R, et al. Lamellipodin, an Ena/VASP ligand, is implicated in the regulation of lamellipodial dynamics. Dev Cell. 2004;7:571–583. doi: 10.1016/j.devcel.2004.07.024. [DOI] [PubMed] [Google Scholar]
  • 23.Smith K, Humphreys D, Hume PJ, Koronakis V. Enteropathogenic Escherichia coli recruits the cellular inositol phosphatase SHIP2 to regulate actin-pedestal formation. Cell Host Microbe. 2010;7:13–24. doi: 10.1016/j.chom.2009.12.004. [DOI] [PubMed] [Google Scholar]
  • 24.Fedele CG, Ooms LM, Ho M, Vieusseux J, O’Toole SA, Millar EK, et al. Inositol polyphosphate 4-phosphatase II regulates PI3K/Akt signaling and is lost in human basal-like breast cancers. Proc Natl Acad Sci USA. 2010;107:22231–22236. doi: 10.1073/pnas.1015245107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Gewinner C, Wang ZC, Richardson A, Teruya-Feldstein J, Etemadmoghadam D, Bowtell D, et al. Evidence that inositol polyphosphate 4-phosphatase type II is a tumor suppressor that inhibits PI3K signaling. Cancer Cell. 2009;16:115–125. doi: 10.1016/j.ccr.2009.06.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Gorges TM, Tinhofer I, Drosch M, Rose L, Zollner TM, Krahn T, et al. Circulating tumour cells escape from EpCAM-based detection due to epithelial-to-mesenchymal transition. BMC Cancer. 2012;12:178. doi: 10.1186/1471-2407-12-178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Chao Y, Wu Q, Acquafondata M, Dhir R, Wells A. Partial mesenchymal to epithelial reverting transition in breast and prostate cancer metastases. Cancer Microenviron. 2012;5:19–28. doi: 10.1007/s12307-011-0085-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Chao YL, Shepard CR, Wells A. Breast carcinoma cells re-express E-cadherin during mesenchymal to epithelial reverting transition. Mol Cancer. 2010;9:179. doi: 10.1186/1476-4598-9-179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Goldschmidt-Clermont PJ, Kim JW, Machesky LM, Rhee SG, Pollard TD. Regulation of phospholipase C-gamma 1 by profilin and tyrosine phosphorylation. Science. 1991;251:1231–1233. doi: 10.1126/science.1848725. [DOI] [PubMed] [Google Scholar]
  • 30.Goldschmidt-Clermont PJ, Machesky LM, Baldassare JJ, Pollard TD. The actin-binding protein profilin binds to PIP2 and inhibits its hydrolysis by phospholipase C. Science. 1990;247:1575–1578. doi: 10.1126/science.2157283. [DOI] [PubMed] [Google Scholar]
  • 31.Lederer M, Jockusch BM, Rothkegel M. Profilin regulates the activity of p42POP, a novel Myb-related transcription factor. J Cell Sci. 2005;118(Pt 2):331–341. doi: 10.1242/jcs.01618. [DOI] [PubMed] [Google Scholar]
  • 32.Barkan D, Kleinman H, Simmons JL, Asmussen H, Kamaraju AK, Hoenorhoff MJ, et al. Inhibition of metastatic outgrowth from single dormant tumor cells by targeting the cytoskeleton. Cancer Res. 2008;68:6241–6250. doi: 10.1158/0008-5472.CAN-07-6849. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Shibue T, Weinberg RA. Integrin beta1-focal adhesion kinase signaling directs the proliferation of metastatic cancer cells disseminated in the lungs. Proc Natl Acad Sci USA. 2009;106:10290–10295. doi: 10.1073/pnas.0904227106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Roy P, Jacobson K. Overexpression of profilin reduces the migration of invasive breast cancer cells. Cell Motil Cytoskeleton. 2004;57:84–95. doi: 10.1002/cm.10160. [DOI] [PubMed] [Google Scholar]
  • 35.Gau D, Ding Z, Baty C, Roy P. Fluorescence resonance energy transfer (FRET)-based detection of profilin-VASP interaction. Cell Mol Bioeng. 2011;4:1–8. doi: 10.1007/s12195-010-0133-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Zou L, Hazan R, Roy P. Profilin-1 overexpression restores adherens junctions in MDA-MB-231 breast cancer cells in R-cadherin-dependent manner. Cell Motil Cytoskeleton. 2009;66:1048–1056. doi: 10.1002/cm.20407. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Kim A, Lakshman N, Karamichos D, Petroll WM. Growth factor regulation of corneal keratocyte differentiation and migration in compressed collagen matrices. Invest Ophthalmol Vis Sci. 2010;51:864–875. doi: 10.1167/iovs.09-4200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Artym VV, Yamada KM, Mueller SC. ECM degradation assays for analyzing local cell invasion. Methods Mol Biol. 2009;522:211–219. doi: 10.1007/978-1-59745-413-1_15. [DOI] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Ding Supplementray data
NIHMS501421-supplement-Ding_Supplementray_data.pdf (419.2KB, pdf)

RESOURCES