WO2017035315A1 - Methods and device for phenotypic classification of cells based on migratory behavior - Google Patents

Abstract

The present invention provides a device and methods that allow time-resolved studies of migration with single-cell resolution on a fibrillary surface that mimics in vivo migration of cells. This platform is utilized for a high-content screen of patient-specific glioblastoma samples to analyze heterogeneities in the phenotype of migratory cells. The information the platform provides has prognostic value as it shows significant differences in time to recurrence and tumor location based on phenotypic classifiers.

Description

METHODS AND DEVICE FOR PHENOTYPIC CLASSIFICATION OF CELLS

BASED ON MIGRATORY BEHAVIOR RELATED APPLICATIONS

[0001] This application claims the benefit under 35 U.S.C. § 119(e) to U.S. Provisional Application No. 62/210,860, filed on Aug. 27, 2015, which is hereby incorporated herein by reference in its entirety.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

[0002] This invention was made in part with government support under Grant Nos. R01NS070024, U01CA155758, and U01CA16359, which were awarded by the National Institutes of Health. The United States government has certain rights in the invention.

BACKGROUND OF THE INVENTION FIELD OF THE INVENTION

[0003] The present invention relates generally to diagnosis and treatment of cancer, and more specifically to a device and methods for determining migratory phenotype signatures in a population of human cells.

BACKGROUND INFORMATION

[0004] Personalized medicine can benefit from patient-specific analysis of cell and tissue properties, especially if such tests are of prognostic value. Aggressive cancers, such as glioblastoma multiforme (GBM), are of particular interest due to the heterogeneous nature of individual tumors and high recurrence following surgical resection. Genomic and proteomic profiling can provide a wealth of information about tumor samples, including cancer-causing abnormalities, specific mutations, and clinically relevant subclasses. However, this information may not be easily interpretable or predictive of certain complex phenotypes, such as invasive growth and enhanced migratory cell dissemination. Furthermore, aggressive cell migration can be a product of multiple and distinct combinations of genetic alterations, particularly in highly heterogeneous tumors, such as GBM. These difficulties indicate a need to develop better methods to predict cancer behavior.

[0005] Aggressive cell migration and dispersal is common to GBM, helping the disease overcome standard treatments, including surgery, radio-, and chemotherapies. As with many other cancers, individual GBM cells can spread from the primary tumor bulk, avoid detection, and reconstitute tumor masses in different areas of the body (e.g., form secondary tumor foci). These migratory and invasive capacities are likely governed by a number of genetic and environmental variables. Therefore, there is a need to develop an experimental platform that would more realistically model the mechano-chemical cellular milieu, yet remain simple and accessible to allow practical, high throughput use.

[0006] Treatment options for aggressive cancers such as glioblastoma are very disappointing. Prognosis remains dismal, and most efforts so far have yielded only minor improvements. One of the issues with existing approaches is that targeted therapies are developed based on data dominated by cells forming the majority of the tumor. In aggressive cancers, however, most of such a majority of cells need not be the culprit of invasion or recurrence, which further complicates interpretation of such data.

[0007] Primary brain cancer, glioblastoma, is a heterogeneous infiltrative disease, making prediction of tumor progression difficult due to the presence of cells with different migratory phenotypes. Cell migration remains challenging to understand and to study in brain cancer and other cancers. The standard approach of studying migration on 2D matrices has severe limitations as cells utilize different cellular mechanisms in contrast to migration through the 3D extracellular matrix in vivo. Therefore, there is a need for new approaches to study cell migration.

SUMMARY OF THE INVENTION

[0008] The present invention is based on the seminal discovery that different migratory phenotypes are present in a tumor, and hence, one is able to provide phenotypic signatures that describe the tumor phenotype much more accurately than previous methods. This information regarding the phenotypes is used for diagnostic purposes to better describe a patient’s disease and prognosis, to better predict disease progression, and to develop better targeted therapies that focus on subpopulations with aggressive phenotypes within the heterogeneous collection of cells that make up a tumor.

[0009] In the context of complex phenotypes, behavior of rare cells can pre-determine the outcome of aggressive cell spread, invasion and metastasis. Despite recent progress, high throughput genomic and proteomic analysis of multiple samples with single cell resolution is still not within reach of clinical applications. Thus, significant value may be offered, as demonstrated herein, by a complementary approach which involves the analysis of phenotypic properties of tumor cells. If properly designed, such an analysis can be rapid, simple, high throughput, and easily conducted on the scale of individual cells. Overall, it is desirable for phenotypic analysis of the tumor samples to recapitulate to a sufficient degree the complexity of the chemical and mechanical features present in the tumor micro- and nano-environment. [0010] Growth factors, e.g., platelet-derived growth factor (PDGF), are potential enhancers of malignant potential in GBM, although this function is not clearly decoupled from their effects on tumor cell proliferation. Various components of the extracellular matrix (ECM) are also likely involved in modulation of cell migration. ECM proteins, such as laminin, are known to regulate cell motility, specifically in neural tissues. In addition, migrating cells can be guided by a variety of mechanical cues presented by the ECM. These stimuli emerge through tissue structures ranging, in size, from nanometers (nm) to microns (µm).

[0011] In particular, not the majority but a small subset of particularly aggressive cells is usually the culprit of invasion and recurrence in cancers. In standard approaches, therefore, the relevant targets in these subpopulations are often masked by the cells of the bulk tumor. Therefore, there is a need to provide a different answer on how to approach these aggressive diseases. The present invention is based on the ability to mimic cell migration through the 3D extracellular matrix in vivo: instead of the standard approach of studying migration on 2D matrices, one mimics the in vivo migration on a platform that biases a 1D migratory mechanism through fibrillary-like structures.

[0012] In one embodiment, provided is a method to classify a cell in a heterogeneous population of cells including providing a surface including a plurality of parallel ridges, wherein the ridges have a depth between about 50 to 1000 nm and a width between about 50 and 1000 nm; applying at least one attachment molecule to the surface, wherein the molecule is selected from the group consisting of collagen, fibronectin, laminin, poly-D-lysine, poly-L- ornithine, proteoglycan, vitronectin, and polysaccharide; contacting a plurality of living cells with the surface for a time and under conditions to allow the cells to attach to the surface, wherein a migration speed or migration direction of the attached cells is recorded over a time period; and classifying the cells based on their migration speed, migration direction, and/or the change in speed or direction over time.

[0013] In various aspects, the cells are cancer cells; the cancer cells are selected from the group consisting of carcinoma, sarcoma, lymphoma, leukemia, germ cell tumor, and glioblastoma; the cells are contacted with at least one agent selected from the group consisting of drugs, natural compounds, toxins, nanoparticles, nucleic acids, viruses, bacteria and other microbes, mammalian cells, hormones, growth factors, and cytokines prior to or simultaneous with the step of contacting the cells with the surface; and/or the growth factor is platelet-derived growth factor. [0014] In additional aspects, the cells are classified as a strong responder or a weak responder based on speed of migration; or more specifically, the cells are classified as a strong responder or a weak responder based on a speed of migration of 50% or less of the cells that have speeds of migration faster than all other cells.

[0015] In an embodiment, provided is a method for determining an effect of an agent on a cell attached to a surface, wherein the surface includes a plurality of parallel ridges including contacting the cells with platelet-derived growth factor; and measuring a change in migration speed and/or migration direction of cells, thereby classifying the cells as a strong or a weak responder based on a speed of migration of 50% or less of the cells that have speeds of migration faster than all other cells.

[0016] In another embodiment, provided is a method for classifying a cancer cell including recording a migration speed or a migration direction of a cell over a period of time; contacting the cell with an agent, wherein the agent is selected from the group consisting of a drug, a natural compound, a toxin, a nanoparticle, a nucleic acid, virus, bacteria, mammalian cell, a biological ligand, a hormone, a growth factor, and a cytokine; and classifying the cell based on migration speed, migration direction, and/or a change of speed or direction over time.

[0017] In some aspects, the cancer cells are glioblastoma cells; the growth factor is platelet-derived growth factor; and/or the cells are classified as a strong responder or a weak responder based on a speed of migration of about 25% of the cells that have speeds of migration faster than other cells.

[0018] In a certain embodiment, provided is a method of phenotyping a cancer cell sample including placing a sample including a cancer cell onto a surface, wherein the surface includes a plurality of parallel ridges; and assessing at least one migratory characteristic of the cancer cell, thereby phenotyping the cancer cell sample based on the migratory characteristic.

[0019] In some aspects, the cancer cell is selected from the group consisting of carcinoma cells, sarcoma cells, lymphoma cells, leukemia cells, germ cell tumor cells, and glioblastoma cells; the cancer cell is a glioblastoma cell; the glioblastoma cell is obtained from a resected brain tissue of a patient having glioblastoma multiforme; the surface further includes a molecule selected from the group consisting of a collagen, fibronectin, laminin, poly-D- lysine, poly-L-ornithine, proteoglycan, vitronectin, and polysaccharide; the parallel ridges are spaced from each other by substantially uniform inter-ridge distances; the parallel ridges have heights of about 500 nanometers, widths of about 350 nanometers, and inter-ridge distances of about 1.5 micrometers; the migratory characteristic of the cancer cell is observed using an optical imaging system in a time-resolved mode; and/or the migratory characteristic is observed at a single-cell resolution.

[0020] In additional aspects, the migratory characteristic includes a speed that corresponds to a ratio of migration distance to migration time; the migratory characteristic includes a directionality that corresponds to a ratio of migration distance parallel with a ridge to migration distance perpendicular to a ridge; the migratory characteristic includes a persistence that corresponds to a ratio of shortest migration distance between a start point and an end point to total migration distance between the start point and the end point; the sample includes a plurality of cancer cells; the cancer cells are classified based on a migratory characteristic of a subset of the cancer cells; the cancer cells are classified based on a speed of about a quarter of the cancer cells that are faster moving than any other cancer cells; the cancer cells are classified based on a change in the migratory characteristic between the migratory characteristic of a subset of a test cancer cell sample contacted with an agent and the migratory characteristic of a subset of a control cancer cell sample not contacted with the agent; the agent is platelet-derived growth factor; the change in the migratory characteristic includes a percentage increase in average migration speed of the subset of the test cancer cell sample; the change in the migratory characteristic includes a percent of time when migration speed is increased in the subset of the test cancer cell sample; the change in the migratory characteristic includes a percent of cells from the test cancer cell sample that are faster than all cells from the control cancer cell sample; and/or the migratory characteristic includes a consensus characteristic obtained by considering each of a percentage increase in average migration speed of the subset of the test cancer cell sample, a percent of time when migration speed is increased in the subset of the test cancer cell sample, and a percent of cells from the test cancer cell sample that are faster than all cells from the control cancer cell sample.

[0021] In an embodiment, provided is a method of identifying an agent which reduces the invasiveness of a cancer cell including contacting a platelet-derived growth factor with a sample containing a cancer cell to obtain a processed cancer cell sample; contacting an agent with the processed cancer cell sample to obtain a treated cancer cell sample; placing the treated cancer cell sample onto a surface including a plurality of parallel ridges, thereby allowing the cancer cell to migrate on the surface; determining at least one migratory characteristic of the cancer cell; and determining that the agent reduces invasiveness of the cancer cell if the migratory characteristic of the cancer cell from the treated sample and a migratory characteristic of a cancer cell from an unprocessed and untreated cancer cell sample are both less than or greater than a migratory characteristic of a cancer cell from a processed but untreated cancer cell sample.

[0022] In some aspects, the cancer cell is selected from the group consisting of a carcinoma cell, sarcoma cell, lymphoma cell, leukemia cell, germ cell tumor cell, and glioblastoma cell; the cancer cell includes a glioblastoma cell; the glioblastoma cell is obtained from a marginal area of a glioblastoma tumor in a brain of a patient having glioblastoma multiforme; the surface further includes a molecule selected from the group consisting of a collagen, fibronectin, laminin, poly-D-lysine, poly-L-ornithine, proteoglycan, vitronectin, and polysaccharide; the parallel ridges are spaced from each other by uniform inter-ridge distances; the parallel ridges have heights of about 500 nanometers, widths of about 350 nanometers, and inter-ridge distances of about 1.5 micrometers; the cells are placed onto the surface through a multi-well dish; the migratory characteristic of the cancer cell is observed using an optical imaging system in a time-resolved mode; the migratory characteristic is observed at a single-cell resolution; the migratory characteristic includes a speed that corresponds to a ratio of total migration distance to total migration time; the migratory characteristic includes a directionality that corresponds to a ratio of migration distance parallel with a ridge to migration distance perpendicular to a ridge; and/or the migratory characteristic includes a persistence that corresponds to a ratio of shortest migration distance between a start point and an end point to total migration distance.

[0023] In one embodiment, provided is a method of identifying an agent that reduces aggressiveness of glioblastoma cells including contacting a platelet-derived growth factor with a sample containing a plurality of glioblastoma cells to obtain a processed glioblastoma cell sample; contacting an agent with the processed glioblastoma cell sample to obtain a treated glioblastoma cell sample; placing the treated glioblastoma cell sample onto a surface including a plurality of parallel ridges having uniform inter-ridge distances, thereby allowing the glioblastoma cells to migrate on the surface; determining an average speed of a subset of the treated glioblastoma cells, wherein the speed corresponds to a ratio of migration distance to migration time, wherein the subset includes at most half of the cells, and wherein the subset includes cells each of which is faster moving than all treated glioblastoma cells not within the subset; and identifying the agent as one that reduces aggressiveness of glioblastoma cells if the average speed determined for the treated cells and an average speed determined for unprocessed and untreated cells are both less than or greater than an average speed determined for processed but untreated cells. [0024] In another embodiment, the disclosure provides a nanopatterned substrate having a plurality of substantially parallel ridges, each ridge having a height of between about 400 to 600 nm and a width of between about 300 to 1000 nm. In some embodiments, each ridge has a height of about 500 nm, a width of about 350 nm, and each ridge is separated by a groove of about 1500 nm. In one embodiment the substrate is planar and composed of silicon, silicon dioxide, a polymer or quartz.

[0025] In still another embodiment, the disclosure provides a system for measuring cell motility. The system includes the substrate of the disclosure and a detector for detecting movement of a cell on the substrate. In various embodiments, the detector is an optical imaging device, such as a microscope. Further, the system may optionally include a computing device having functionality to analyze cell motility data.

[0026] In yet another embodiment, the disclosure provides a kit which includes the substrate of the disclosure, a platelet-derived growth factor receptor alpha (PDGFRα) ligand, and optionally, instructions for performing cellular analysis along with appropriate packaging. In one embodiment, the PDGFRα ligand is platelet-derived growth factor (PDGF).

[0027] In one embodiment, provided is a method for reducing invasiveness of a cancer cell including contacting the cell with an agent suspected of affecting invasiveness of the cell and determining a migration speed of the cell before and after contact with the agent in the substrate of the disclosure, wherein a reduction in migration speed is indicative of an agent that reduces the invasiveness of the cancer cell.

[0028] In another embodiment, provided is a method of reducing the time to recurrence after resection of a tumor containing a cancer cell including determining a migration speed of the cell before and after contact with an agent in the substrate of the disclosure, wherein a reduction in the migration speed is indicative of an agent that reduces the time to recurrence after resection of the tumor.

[0029] Overall, the present invention provides for high-throughput analysis for tumor single-cell migration; is more sensitive and physiologically relevant than classical screening assays; detects glioma cells showing inter- and intra-patient differential sensitivity to platelet- derived growth factor (PDGF); and demonstrates how glioma cell sensitivity to PDGF correlates with tumor recurrence and tumor location.

[0030] Other aspects and advantages of the invention will be apparent from the following description and the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figures 1A-1E: Phenotypic Screening of Heterogeneous Cell Populations Recapitulates the Microenvironment of Migrating Cells. (A) Cells with heterogeneous phenotypes are isolated from a patient’s tumor (MRI of tumor for sample GBM 612). (B) The cells are seeded on a platform that has a multi-well structure, allowing testing of multiple conditions, and is an on-glass technology, allowing direct imaging of migration and morphology with single-cell resolution. (C) Images show that GBM 612 cells migrating on the platform have similar morphology and migration speed compared to GBM 612 cells migrating in ex vivo human brain tissue and 3D Matrigel. In comparison, cells migrating on flat surfaces are not polarized and have reduced migration speeds compared to the other cases (see Figures 8 and 9) (scale bars, 25 μm; duration between each frame, 1 hr). Overall, this shows that the nanopatterned quasi-3D surface of the platform recapitulates important aspects of in vivo migration including migration speed and cell morphology. (D) ROCK inhibitor (Y- 27632) affects migration of GBM 612 cells on tissue-mimetic substrates and flat surfaces (n = 30 cells, mean + SEM,^∗∗∗p < 0.0001,^∗paired against control group, #paired against 3 μM group, Kruskal-Wallis one-way ANOVA on ranks, Dunn’s method). (E) The platform provides important information on the migration response of heterogeneous cell populations. GBM 612 samples show a subpopulation of cells whose migration is fast and stable over time. Additional experiments (not shown) indicate that even for samples having the same average migration speed, a detailed analysis with the platform can reveal that some samples have fast moving outliers, and a timelapse data obtained through the platform can show that there are significant differences at the beginning, which would be masked by averaging of the data. These details can have important implications for the disease.

[0032] Figures 2A-2D: Migratory Response to PDGF Correlates with Tumor Characteristics Both In vitro and In vivo (part 1). (A) RT-PCR analysis of PDGFRα mRNA levels in responding sample GBM 276 and non-responding sample GBM 253 (n = 3,^∗p = 0.006, Student’s t test). (B) Western blot for PDGFRα protein expression in GBM samples grown as adherent or spheroid cultures. (C and D) Quantification of migration speed of cell lines GBM 253 (C) and GBM 276 (D) in the presence of PDGF-AA (50 ng/ml) and imatinib (30 μM) (n≈ 80 cells, mean + SEM,^∗p < 0.05,^∗∗p < 0.01,^∗∗∗p < 0.001,^∗paired against control group, #paired against PDGF group, Kruskal-Wallis one-way ANOVA on ranks, Dunn’s method) (asterisks indicate pairing against the control group, hash marks indicated pairing against the PDGF group). [0033] Figures 3A-3F: Migratory Response to PDGF Correlates with Tumor Characteristics Both In vitro and In vivo (part 2). (A and B) Migration speed of GBM 276 for the slowest (A) and fastest (B) quartile of the cells (i.e., 25% of the slowest- and fastest- moving cells, respectively), showing that only a subpopulation responds to PDGF. (C) Quantification of migration measured by alignment of GBM 276 cells (^∗p < 0.05, Kruskal- Wallis one-way ANOVA on ranks, Dunn’s method) (asterisks indicate comparisons to all other conditions). (D) Survival curves of mice injected with GBM 276 cells cultured in control spheroid conditions or in the presence of PDGF-AA (n = 4 mice per group,^∗p = 0.0097, Gehan-Breslow-Wilcoxon test). (E and F) Kaplan-Meier plots based on clinical TCGA data of GBM patients, comparing survival between high and low expression of PDGF-AA (E) and PDGFRα (F), respectively. The cohorts were divided at the median of the expression level of the respective gene.

[0034] Figures 4A-4C: Information on Migration Speed Reveals Important Differences among Patient Samples in Response to PDGF (part 1). (A) Analyzing the fastest quartile (GBM 499) reveals that the subpopulations display a significant response to PDGF. In contrast, for the whole population, there is no significant response. (B) Migration speed time lapse demonstrates that sample GBM 501 does not respond to PDGF at all times (compared to GBM 609); on average, however, both samples respond significantly to PDGF. (C) Both GBM 630 and GBM 544 samples display a significant increase in average migration speed (^∗p < 0.05,^∗∗p < 0.01,^∗∗∗p < 0.001, Wilcoxon rank-sum test). However, GBM 630 has a significantly larger number of fast outliers, while GBM 544 displays a uniform increase in speed.

[0035] Figure 5: Information on Migration Speed Reveals Important Differences among Patient Samples in Response to PDGF (part 2). The platform allows patient sample classification based on multiple characteristics, permitting a better description of the heterogeneity of the samples. Compared are 14 patients’ GBM cell lines. The samples were grouped based on whether there is a significant increase (p < 0.05, Wilcoxon rank-sum test) in average migration speed in response to PDGF (group I) and whether this significant increase is persistent over time (group II). Furthermore, samples are grouped based on the number of outliers (cells faster than the fastest cells in the control group) with a threshold of 4 cells (5%, for a total of 80 cells) (group III). For reference, the PDGFRα protein expression level (group IV) and the subclass of the GBM cells (group V) are also provided.

[0036] Figures 6A-6B: GBM Migratory Response to PDGF Correlates with Patient Tumor Characteristics (part 1). (A and B) Kaplan-Meier plots comparing recurrence between PDGF-responsive and PDGF-unresponsive groups (n = 11 patient tumors, consensus group) (A) and based on the criteria in Figure 5 (n = 14 patient tumors, weak and strong responders) (B). The p values were calculated using a two-tailed log-rank (Mantel-Cox) test.

[0037] Figures 7A-7F: GBM Migratory Response to PDGF Correlates with Patient Tumor Characteristics (part 2). (A–D) Magnetic Resonance Imaging (MRI) scans of patients with tumors from the unresponsive group (A and B) and the responsive group (C and D). (E) Distribution of PDGF-responsive and PDGF-unresponsive tumors that formed in specific locations in the brain (n = 11 patient tumors, Barnard’s exact test). (F) Time to recurrence for GBM samples separated into low-directionality (<3.25) and high-directionality (≥3.25) groups (n≥ 4, mean + SEM, Wilcoxon rank-sum test) (the threshold of 3.25 was determined using linear discriminant analysis).

[0038] Figures 8A-8G: Multi-well, nanopatterned platform induces changes in cell morphology and migration (part 1). (A) SEM images of topographic pattern with parallel ridges 350 nm wide, 350 nm high, spaced 1.5 µm apart. Box (left) indicates higher magnification image (right) (B) SEM images of human GBM sample 318 cells cultured on nanogroove pattern for 24 hours in vitro. Box as in A. (C) Schematic describing construction of multi-well, nanopatterned device, wherein multiple conditions (red and yellow colors) can be observed simultaneously. (D) Bright-field images comparing cells migrating on a smooth surface vs. a nanopatterned substrate of the present invention. (E) Quantitative comparison of average cell area and spindle shape (n≥ 50 cells, mean + s.e.m., *P < .05, Mann-Whitney Rank Sum Test). (F) Quantitative comparison of migration measured by speed, alignment, and persistence (n≥ 60 cells, mean + s.e.m., *P as in E). (G) Instantaneous migration speed quantified as a function of time (n≥ 60 cells).

[0039] Figures 9A-9G: Multi-well, nanopatterned platform induces changes in cell morphology and migration (part 2). (A-B) Quantitative comparison of average cell area (A) and spindle shape (B) of GBM 253 (n≥ 45 cells, mean + s.e.m., *P < .05, Mann-Whitney Rank Sum Test). (C) Bright-field images of GBM 276 cells cultured on smooth PUA surface (left) and nanogroove pattern (right) for approximately 24 hours in vitro. Complete trajectories of individual cells are displayed as yellow tracks (D) Instantaneous migration speed of GBM 276 quantified as a function of time (n≥ 70 cells) (E) Quantitative comparison of migration measured by speed, alignment, and persistence (n≥ 70 cells, mean + s.e.m., *P as in H-I). (F-G) Effect of the ROCK inhibitor Y-27632 to the migration of GBM965 cells on tissue-mimetic substrates (F) and flat surfaces (G), respectively. [0040] Figures 10A-10B: Nanopatterned platform enables higher sensitivity to soluble and immobilized factors (part 1). (A-B) Quantification of migration measured by speed, alignment, and persistence of GBM 318 cells cultured on a smooth surface (A) or a nanopatterned surface of the present invention (B) coated with varying laminin concentrations. Values normalized to 5 µg/mL condition. (n≥ 50 cells, mean + s.e.m., *P < .05, Kruskal-Wallis One Way ANOVA on Ranks, Dunn’s Method. Brackets denote pairwise comparisons. Asterisks without brackets indicate comparisons to all other conditions).

[0041] Figures 11A-11B: Nanopatterned platform enables higher sensitivity to soluble and immobilized factors (part 2). (A-B) Quantification of migration measured by speed and persistence of GBM 318 cells cultured on a smooth surface (A) or a nanopatterned surface of the present invention (B) in the presence of varying PDGF concentrations. Values normalized to control condition. (n≥ 75 cells, mean + s.e.m., *P as in A-B).

[0042] Figures 12A-12D: Nanopatterned platform enables higher sensitivity to soluble and immobilized factors (part 3). (A-B) Quantification of migration speed of GBM 276 cells cultured on smooth surface (A) or nanopatterned surface (B). Comparisons made for varying laminin concentrations (n≥ 75 cells, mean + s.e.m., *P < .05, Kruskal-Wallis One Way ANOVA on Ranks, Dunn’s Method). (C) Quantification of migration speed of GBM 630 cells (n≥ 60 cells, mean + s.e.m., *P < .05, Kruskal-Wallis One Way ANOVA on Ranks, Dunn’s Method. Brackets denote pairwise comparisons. Asterisks without brackets indicate comparisons to all other conditions). (D) Instantaneous migration speed of GBM 630 quantified as a function of time shows that the increase is not time dependent (n≥ 60 cells).

[0043] Figures 13A-13E: Effect of PDGF to migratory response and in vivo tumor formation (part 1). (A) RT-PCR analysis of PDGFRα mRNA levels in responding samples GBM 609 and GBM 630 and non-responding sample GBM 549 (n = 3, *P < .05, One Way ANOVA, Holm-Sidak method). (B) Western blot for PDGFRα protein expression in GBM samples 549, 609, and 630. (C) Western blot for PDGFRα protein expression in additional responding and non-responding GBM samples. (D) Migration speed of GBM 276 cells cultured on nanopatterned platform in the presence of Imatinib. Values normalized to control condition. (n≥ 40 cells, mean + s.e.m., *P < .05, Kruskal-Wallis One Way ANOVA on Ranks, Dunn’s Method. Brackets denote pairwise comparisons. Asterisks without brackets indicate comparisons to all other conditions). (E) Western blot for p-Akt protein expression in GBM 276 exposed to increasing concentrations of PDGF-AA. Alkaline phosphatase (AP, New England Biolabs) was added to serve as a negative control. [0044] Figures 14A-14F: Effect of PDGF to migratory response and in vivo tumor formation (part 2). (A-B) Quantification of average tumor area per brain section (A) and average tumor cell dispersion (B) in mice receiving GBM 276 xenografts. Brains were subsequently infused with saline or PDGF-AA. Values normalized to control condition (n = 2 mice, mean + s.e.m.). (C) Quantification of Ki67 expressing cells shows increases following PDGF-AA (50 ng/mL) exposure when grown as astrocytes on the nanopatterned platform (n ≥ 20 frames, mean + s.e.m., *P = 0.014, Mann-Whitney Rank Sum Test) (D) Quantification of Ki67 expressing cells shows increases following PDGF-AA (20 ng/mL) exposure when grown as floating neurospheres (n = 4 neurospheres, mean + s.e.m., *P = 0.015, Student’s t- test). (E) Quantification of cell marker expression in GBM 276 cells following approximately 3 weeks of exposure to PDGF-AA (50 ng/mL) or control medium (n = 20 random fields, mean + s.e.m., *P = 0.001, Mann-Whitney Rank Sum Test). (F) Flow cytometry of GB 276 cells following exposure to PDGF-AA (50 ng/mL) and Imatinib (10 µM) (n=3, mean + s.e.m., *P < .001)

[0045] Figures 15A-15B: Genotypic information does not correlate with time to recurrence or tumor location (part 1). (A) Scatterplot displaying moderate correlation between alignment of GBM samples and time to recurrence of the source tumor. A linear regression was used to model the data (n= 10, R2 = 0.6732). (B) Kaplan-Meier curve depicting survival with respect to PDGF responsiveness.

[0046] Figures 16A-16D: Genotypic information does not correlate with time to recurrence or tumor location (part 2). (A-B) Kaplan-Meier curves depicting time to recurrence among patients with tumors with High and Low PDGFR expression (A) and Mesenchymal or Proneural Sub-type (B). (C) Analysis of patient tumor information from Rembrandt database. Quantification of PDGFRα expression in numerous patient glioma samples organized by tumor location (n = 15 temporal lobe tumors, 32 frontal lobe tumors, mean + s.e.m.). (D) Distribution of tumors with low-expression (< 10.58) and high- expression (≥ 10.58) of PDGFRα that formed in specific locations in the brain (n = 47 patient tumors. Threshold of 10.58 was determined using linear discriminant analysis).

[0047] Figure 17: Flow Chart disclosing classifying tumors for diagnostic or prognostics using cell-migration data.

[0048] Figure 18: Flow Chart disclosing determining the effect of an agent. DETAILED DESCRIPTION OF THE INVENTION

[0049] Standard methods for assessing a cancer such as glioblastoma suffer from at least two problems: (1) low signal to noise ratio, due to majority of cancer cells not being determinative of overall cancer properties, and (2) incorrect signal, due to the commonly used 2D surfaces not sufficiently mimicking the natural 3D environment. The present invention solves, inter alia, these problems by mimicking the 3D environment with a platform that allows migration on 1D fibrillary surfaces and by allowing for detection of information at a single-cell level. In addition, various embodiments provide for assessments using agents (e.g., platelet-derived growth factor), and for improved prognostic/diagnostic methods that rely on a defined subset of cancer cells characterized by a variety of metrics gleaned from the disclosed platforms and processes.

[0050] The present invention employs engineered nano-scale cell adhesion substrata to construct fibrillar surfaces that mimic topographical signals of natural ECM. In comparison to flat surfaces, the provided platform provides a more representative model for studying the migration of glioma cells. This platform achieves far greater resolution and sensitivity in migration analyses than do commonly used methods. More importantly it can provide highly informative, patient specific results regarding tumor progression in vivo. Previous studies using nano-fabricated platforms have not evidenced such enhancements over existing approaches or obtained similar, medically relevant information.

[0051] Invasive nature of glioblastoma and other aggressive cancers highlights the importance of assaying cell migration as a phenotypic feature potentially predictive of clinical outcomes. Provided herein is a relatively simple but information-rich experimental method aimed at the analysis of primary patient samples on a single cell level, which allows high throughput screening of the effects of variable extracellular milieu. Using this method on a range of patient derived samples and contrasting the results of the analysis with respective clinical information revealed substantial predictive power, particularly when cell migration was examined in conjunction with the effects of platelet-derived growth factor (PDGF). This result indicates that cell migration, as examined in structured, mechanically- defined culture conditions, can be predictive of more complex in vivo invasion processes and can thus be a powerful phenotypic analysis tool with strong clinical implications.

[0052] Examinations of glioma cell migration and its relationship to tumor progression are not without precedent (Friedlander, D.R., Zagzag, D., Shiff, B., Cohen, H., Allen, J.C., Kelly, P.J., and Grumet, M., Migration of brain tumor cells on extracellular matrix proteins in vitro correlates with tumor type and grade and involves alphaV and b1 integrins, Cancer Res. 56, 1939–1947 (1996), which is hereby incorporated herein by reference in its entirety). However, previous studies that monitored average migration speeds on 2D surfaces have not achieved such direct predictions of patient specific tumor features as embodiments of the present invention do, emphasizing the benefits of analyzing heterogeneities within samples and utilizing surfaces that better mimic in vivo conditions. The significant relationships that were found of migratory behavior with time to recurrence and tumor location provide crucial insight into this disease. Recurrence of glioblastoma after tumor resection is the primary cause of death in patients and one of the most important predictors of future patient outcome. The present invention provides a relatively simple method to glean information about these phenomena. Predictions of the tumor location can hold critical relevance to tumor prognosis based on future tumor progression and its effect on eloquent areas of the brain. This information might also provide clues to the origins of glioma which have yet to be fully understood. The disclosed finding that PDGF non-responding tumors were predominantly derived from the temporal lobe could suggest that PDGF signaling is less critical to tumor progression in certain regions of the brain. Moreover, the information provided through the presented platform can be particularly useful in prognostic analyses of tumor samples at the time of surgery. Direct access to individual cell migration analysis would potentially gain critical importance for future treatment modalities.

[0053] The present invention demonstrates that high-throughput, single cell-resolution phenotypic screening is of great value in the diagnostic and prognostic analysis of human cancer samples. Furthermore, the present invention demonstrates that this analysis can be achieved by using a highly versatile, and convenient platform for interrogating migration phenotypes of primary GBM cells in the context of diverse environmental parameters, including different medium and ECM compositions. Using this assay, surprisingly, it was found that the clinical outcome of GBM tumors strongly correlated with the combined responsiveness of a subset of the cell population to two environmental inputs: the aligned surface fibers mimicking the ECM nano-topography and a growth factor, PDGF. The ability to observe this responsiveness at the single cell level and thus examining different cell sub- populations was useful for the success of this phenotypic analysis, revealing correlations with such critical prognostic tumor characteristics as the time of recurrence after resection. Overall, the present invention, in addition to providing further insights into the mechanisms of invasive glioblastoma spread, strongly indicate considerable clinical, prognostic uses of the disclosed phenotypic analysis test. [0054] The experimental platform described in the present invention has important advantages over other phenotypic analysis platforms designed to assay cell migration or invasion. Compared to standard 2D migration assays, migration on the quasi-3D fibrillary surface of the platform provides a cellular environment more similar to in vivo conditions, as evidenced in the similarity of several aspects of migration in ex vivo human brain tissue and a 3D hydrogel. Similarly, trans-well migration analysis, another relatively simple method directly assaying cell invasion, for which a multi-well design has also been described, usually requires at least an order of magnitude greater numbers of cells than the method described herein. More importantly, classical trans-well assays fail to yield the information on migration and morphology of each individual cell only showing information at the end-point, not taking in account the heterogeneity of the tumors, high speed outlier cells or time- dependent responses. This missing information can be critical in the analysis of human tumors. Indeed, a substantial degree of heterogeneity was found in the glioblastoma samples analyzed. The increased average migration speed of a given cell population in the presence of PDGF was ascribed to a relatively small sub-population of particularly aggressive cells (e.g., approximately 25%). Knowledge of the degree of population heterogeneity can be critical to the decision making in the clinic. Furthermore, the described results also highlight advantages over traditional protein expression analyses. For example, an incomplete correlation was observed between receptor expression and response to PDGF signaling, possibly due to veiled differences in the signal transduction pathways. Also demonstrated herein is that protein expression analysis was less sensitive and less robust at predicting differences between patient tumor features. These approaches also suffered from lower amounts of raw data, as limited supplies of primary tissues were available for their cumbersome requirements for cellular material. These points support the description of the provided methodology as a better, simpler, and more informative method to gain critical information about patient tumors and cell populations.

[0055] The analysis presented in the present invention reveals the importance of careful engineering of extracellular milieu in cell migration analysis, both chemical and mechanical. Although sensitivity of cell movements to the presence of PDGF was a strong discriminating factor for various in vivo tumor characteristics, this sensitivity was more detectable when cells were cultured on adhesion substrata with fine, nano-scale topography designed to mimic the structure of in vivo extracellular cell micro-environment. This finding suggests that chemical and mechanical cues can strongly synergize both in vivo and in vitro to guide cell responses, necessitating flexible and multi-faceted bio-mimetic screening of multiple conditions in assaying patient samples. Patterning the nano-topographic features in the form of parallel nano-scale ridge arrays had an added advantage of simplifying the analysis of cell migration, as cells moved primarily in one-dimensional paths consistent with the orientation of this mechanical cue. This aspect of the experimental analysis makes the platform described here easy to use in both academic and clinical settings. The present invention can provide an important prognostic tool, with benefits that include high-throughput label-free analysis with single-cell resolution, low demand for precious primary cell samples, and better physiological relevance compared to other migration assays.

[0056] Before the present methods and devices are described, it is to be understood that this invention is not limited to particular devices, compositions, methods, and experimental conditions described, as such devices, compositions, methods, and conditions may vary. It is also to be understood that the terminology used herein is for purposes of describing particular embodiments only, and is not intended to be limiting, since the scope of the present invention will be limited only in the appended claims.

[0057] As used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural references unless the context clearly dictates otherwise. Thus, for example, references to "the method" includes one or more methods, and/or steps of the type described herein which will become apparent to those persons skilled in the art upon reading this disclosure and so forth.

[0058] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the invention, the preferred methods and materials are now described.

[0059] In some embodiments, the present invention provides for methods that classify, phenotype, or determine a property of cancer cells. In various embodiments, the methods detect an effect of an agent (e.g., drugs, natural compounds, herbal compounds, plant-based compounds, mineral-based compounds, marine-based compounds, chemically synthesized compounds, toxins, nanoparticles, nucleic acids, viruses, bacteria, archaea, eukaryotic microbes, mammalian cells, hormones, enzymes, growth factors, cytokines, chemokines, antibiotics, antibodies, antibody fragments, synthetic antibodies, vaccines, genetically- engineered biologics) on cancer cells. Unless explicitly indicated otherwise, the different embodiments recited herein are not necessarily intended to be mutually-exclusive. In other words, even when an embodiment is recited as an“other embodiment,” features of that embodiment may be combined with those of another embodiment. For example, some methods encompassed by the present invention both detect an effect of an agent and also classify cancer cells.

[0060] In certain embodiments, the methods allow determination of the locations in the brain (e.g., frontal lobe, temporal lobe) of the tumors from which the cells had originated. In some embodiments, the methods allow prediction of time to recurrence of the cancer after a surgical resection. In other embodiments, disclosed methods allow identifying an agent that reduces the invasiveness or aggressiveness of a cancer cell. Some of the disclosed methods allow observing cells at a single-cell resolution. Particular methods allow observing cells (e.g., at a single-cell resolution) using an optical imaging system (e.g., bright-field microscopy, fluorescence microscopy, dark-field microscopy, confocal microscopy) in a time-resolved mode (e.g., in a way that generates a time vs. location trajectory for an observed cell). Various device setups allow observing multiple groups of cells at the same time (e.g., via using a multi-well dish over the surface having the cells).

[0061] The methods encompassed by the present invention are applicable to various cancers, such as aggressive cancers, which may be heterogeneous and infiltrative. An example cancer to which some of the methods are applicable is glioblastoma multiforme (GBM). Other examples include carcinoma, sarcoma, lymphoma, leukemia, and germ cell tumor.

[0062] Some of the embodiments rely on a platform that has multiple parallel ridges. The ridges may be separated from each other by substantially uniform inter-ridge distances. In some embodiments, the ridges have heights of about 500 nanometers, widths of about 350 nanometers, and inter-ridge distances of about 1.5 micrometers. Ridge heights can take on other values, such as greater than or equal to 250, 300, 350, 400, 450, 550, 600, or 650 nanometers. Similarly, ridge widths can take on other values, such as greater than or equal to 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, or 600 nanometers. The inter-ridge distances can be varied as well. For example, they may be greater than or equal to 1.0, 1.1, 1.2, 1.3, 1.4, 1.6, 1.7, 1.8, 1.9, or 2.0 micrometers.

[0063] In some embodiments, the surface of the platform or device is coated with a molecule. The molecule can be an attachment molecule. An example molecule that can be used for this purpose is laminin. Other such examples include collagens, fibronectins, poly- D-lysine, poly-L-ornithine, proteoglycans, vitronectin, polysaccharides, heparan sulfate, keratan sulfate, chondroitin sulfate, keratin sulfate, integrins, cadherins, selectins, immunoglobulins, glycosaminoglyclans, hyaluronic acid, elastin. In some embodiments, the molecule may be hemicellulose, pectin, extension, cellulose, and/or a biofilm component. Multiple molecules can be used together on the surface in some embodiments.

[0064] The present invention provides for methods that include detection of a migratory characteristic. Migratory characteristic is a property derived from an observation of cells that migrate on a platform. The migratory characteristic, which may ultimately be used to estimate a time to recurrence, in some embodiments is directionality (also referred to as alignment). Directionality can be measured by taking a ratio of a distance travelled parallel to a ridge and a distance travelled perpendicular to a ridge. The distances, in this or other embodiments, may be scalar distances (integrated along the path) or vectorial distances (the shortest straight line distance between the start and end points). Additionally, the distances, in this or other embodiments, can be measured for a defined stretch of the platform, or for a defined time segment. In some embodiments, an average directionality of a first sample that is higher than that of a second sample indicates that the first sample has a longer time to recurrence than the second sample.

[0065] The migratory characteristic, in various embodiments, can be obtained as an average of a population of cells. For example, it may be an average of the full group of cells, of half the cells, or a quarter of the cells. In particular, the migratory characteristic may be obtained as the average migratory characteristic of about 100%, 95%, 90%, 85%, 80%, 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10%, or 5% of the cells. Other incremental percentages of cells between these values are also encompassed within embodiments of the present invention.

[0066] In some embodiments, the migratory characteristic is persistence. Persistence can be obtained as a ratio of the shortest migration distance between the start and end points and the total migration distance. It can also be defined with respect to a limited stretch of the platform or a limited time segment. In certain embodiments, a higher persistence of a first sample compared to that of a second sample indicates that the first sample has a shorter time to recurrence than the second sample. As in assessments of directionality, persistence can be obtained as an average of a group of cells.

[0067] In various embodiments, the migratory characteristic is speed or is speed-based. Cells may be classified into groups (e.g., groups of 2, 3, 4, or more) based on a migratory characteristic that is based on speed. An assessment of speed can take many forms.

[0068] In some embodiments, the speed-based migratory characteristic is obtained in the form of a change with respect to an agent. For example, a speed-based characteristic may be based on an average speed obtained from cells treated with PDGF as compared to an average speed obtained from cells not treated with PDGF. In certain embodiments, the migratory characteristic is a percentage increase due to an agent in average migration speed of a group of cells. As in other embodiments, groups of cells may range from 100% of the cells to a subset, such as about 25% of the cells. In other embodiments, the migratory characteristic is the percent of time when migration speed is increased due to an agent in the group of cells. In other embodiments, the migratory characteristic is the percent of cells in an agent-treated sample that are faster than a set of cells (e.g., 100%, 95%, 90%, 85%, 80%, 75%) in a sample not treated with the agent. In various embodiments, a consensus is taken by considering more than one of these migratory characteristics. As an example, one may bin (e.g., separate into two groups, such as agent-responsive and agent-nonresponsive) cancer cells by classifying them as below/above a threshold migratory characteristic, and then classify those that consistently are responsive to an agent according to different migratory characteristics as responsive, and those that are not responsive to an agent according to different migratory characteristics as non-responsive. The threshold for a percentage increase in average migration speed can take many values, such as 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, and 25%. Similarly, the threshold for a percent of time when migration speed is increased can take many values, such as 75%, 70%, 65%, 60%, 55%, 50%, 45%, 40%, 35%, 30%, and 25%. The threshold for the percent of outliers faster than a group of cells in the control group can also take multiple values, such as 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, and 2%.

[0069] In certain embodiments, a speed-based migratory characteristic can also be used to determine the location in the brain of the tumor from which the cells originated. For example, using the methods explained above, a consensus grouping (e.g., binning) can classify cells into agent-responsive and agent-nonresponsive groups. In an embodiment, the responsive cells would be determined to have been located at the frontal lobe of a brain. In an embodiments, the nonresponsive cells would be determined to substantially have been located at the temporal lobe of a brain.

[0070] In some embodiments, provided are methods that identify agents that reduce the invasiveness and/or aggressiveness and/or time to recurrence of a tumor. In one such embodiment, three migratory trajectories are recorded: a first one for cancer cells that are neither processed with a growth factor nor treated with a candidate agent; a second one for cancer cells that are processed (e.g., mixed in a solution together) with a growth factor (e.g., PDGF); and a third one for cancer cells that are both processed with a growth factor and treated with the candidate agent (e.g., a drug that is suspected of reducing the invasiveness of cancer cells). The candidate agent would be identified as one that reduces the invasiveness (or aggressiveness or time to recurrence) of the tumor if it reverses the effects of the growth factor. For example, for an aggressive tumor of glioblastoma cells, speed-based metrics are expected to increase in processed cells (i.e., the processed cells would have higher speeds than unprocessed untreated cells). In that scenario, if an agent reduces the speed of the processed cells (i.e., if the migration speeds of both the unprocessed untreated cells and the processed treated cells are less than the migration speeds of the processed cells), one may infer that the agent reduces the invasiveness of the cancer cell. For other migratory characteristics, such as directionality and persistence, one may compare the characteristics between treated and untreated samples, without a need for processing the cells with a growth factor. For an alternative metric, such as directionality, that is reduced in aggressive cells, one would look for an increase in the agent-treated cells as opposed to a decrease. In embodiments that employ both a growth factor such as PDGF and a candidate agent, one may add, in different embodiments, the growth factor and the agent in any order, including simultaneously, to the cells.

[0071] The cancer cells may be obtained from a resected brain tissue of a patient having a form of cancer (e.g., an aggressive cancer, such as glioblastoma). In some embodiments, the cells are obtained from a marginal area of a glioblastoma tumor in the brain of a patient having GBM.

[0072] Furthermore, Figures 17 and 18 illustrate some of the methods described herein that make use of the migration of cells on the disclosed platforms. Various details of the present invention can be understood more clearly in light of the following examples.

[0073] The following example is provided to further illustrate the advantages and features of the present invention, but are not intended to limit the scope of the invention. While they are typical of those that might be used, other procedures, methodologies, or techniques known to those skilled in the art may alternatively be used.

EXAMPLE 1

EXPERIMENTAL PROCEDURES

Cells

[0074] Human tissues were obtained at Johns Hopkins medical institutions and used with approval of the Institutional Review Board. Glioblastoma pathologically confirmed tumor samples (GBM 221, GBM 253, GBM 276, GBM 318, GBM 499, GBM 501, GBM 544, GBM 549, GBM 567, GBM 609, GBM 612, GBM 626, GBM 630, and GBM 854) were derived from primary intraoperative tissues of patients undergoing surgery. Tissue donors received no treatment before surgery.

[0075] Glioblastoma (GBM) samples were donated by patients at Johns Hopkins Medical Institutions. Human tissues were obtained and utilized with approval of the Institutional Review Board (IRB). Brain tumor samples GBM 221, GBM 253, GBM 276, GBM 318, GBM 499, GBM 501, GBM 544, GBM 549, GBM 567, GBM 609, GBM 612, GBM 626, GBM 630, and GBM 854 were derived from primary intraoperative tissues of patients undergoing surgery for glioblastoma. Tissue donors received no treatment prior to surgery. All tissue samples were pathologically confirmed as glioblastoma. As described in previous studies (Garzon-Muvdi, T., Schiapparelli, P., ap Rhys, C., Guerrero-Cazares, H., Smith, C., Kim, D.-H., Kone, L., Farber, H., Lee, D.Y., An, S.S., et al., Regulation of brain tumor dispersal by NKCC1 through a novel role in focal adhesion regulation, PLoS Biol. 10, e1001320 (2012); Chaichana, K., Zamora-Berridi, G., Camara-Quintana, J., Quinones- Hinojosa, A., Neurosphere assays: growth factors and hormone differences in tumor and nontumor studies, Stem cells 24, 2851 (2006); Guerrero-Cazares, H., Chaichana, K. L., Quinones-Hinojosa, A., Neurosphere culture and human organotypic model to evaluate brain tumor stem cells, Methods in molecular biology 568, 73 (2009), all of which are hereby incorporated herein by reference in their entirety), thorough purification and culturing methods were used to maintain these samples and avoid cross-contamination of other cell types. Primary GBM cells were cultured as either adherent or spheroids, as specified by experiment and as previously described (Garzon-Muvdi, T., Schiapparelli, P., ap Rhys, C., Guerrero-Cazares, H., Smith, C., Kim, D.-H., Kone, L., Farber, H., Lee, D.Y., An, S.S., et al., Regulation of brain tumor dispersal by NKCC1 through a novel role in focal adhesion regu- lation, PLoS Biol. 10, e1001320 (2012); Chaichana, K., Zamora-Berridi, G., Camara- Quintana, J., Quinones-Hinojosa, A., Neurosphere assays: growth factors and hormone differences in tumor and nontumor studies, Stem cells 24, 2851 (2006); Guerrero-Cazares, H., Chaichana, K. L., Quinones-Hinojosa, A., Neurosphere culture and human organotypic model to evaluate brain tumor stem cells, Methods in molecular biology 568, 73 (2009), all of which are hereby incorporated herein by reference in their entirety). Adherent GBM cells were cultured in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 containing 2 mM L-glutamine, with added 50 U mL^-1 penicillin, 50 mg mL^-1 streptomycin, and 10% fetal bovine serum (Invitrogen). Spheroids were cultured in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 containing 2 mM L-glutamine, with added 50 U mL^-1 penicillin, 50 mg mL^-1 streptomycin, supplemented with B27, 20 ng mL^-1 endothelial growth factor (EGF), and 20 ng mL^-1 fibroblast growth factor (FGF).

Construction of a Multi-well Nanopatterned Device

[0076] The topographic nanopatterned substratum, consisting of parallel ridges 350 nm wide, 500 nm high, spaced 1.5 µm apart, was fabricated onto glass coverslips as previously described (Kim, D.-H., Han, K., Gupta, K., Kwon, K.W., Suh, K.-Y., and Levchenko, A., Mechanosensitivity of fibroblast cell shape and movement to anisotropic substratum topography gradients, Biomaterials 30, 5433–5444 (2009); Kim, D.-H., Seo, C.-H., Han, K., Kwon, K.W., Levchenko, A., and Suh, K.-Y., Guided cell migration on microtextured substrates with variable local density and anisotropy, Adv. Funct. Mater. 19, 1579–1586 (2009), all of which are hereby incorporated herein by reference in their entirety), using UV- assisted capillary molding techniques. Prior to application of the poly(urethane acrylate) (PUA) mold, glass substrates were cleaned with isopropyl alcohol, rinsed in distilled-- deionized water, and dried in a nitrogen stream. Afterwards, a thin layer (~100 nm) of an adhesive agent (phosphoric acrylate: propylene glycol monomethyl ether acetate1/41:10, volume ratio) was spin-coated onto the glass substrate for 30 s at 3000 rpm. Subsequently the PUA precursor was dispensed onto the substrate, and a previously-constructed PUA mold was directly placed onto the surface. The PUA precursor spontaneously absorbed into the cavities of the mold via capillary action and was cured by exposure to UV light (X = 250–400 nm) for ~30 s (dose = 100 mJ cm-2). Glioma cells migrate adjacent to elongated ECM fibers and blood vessels. These ECM-rich structures can range from 20 nm in diameter, e.g., collagen fibrils, to several microns across, e.g., myelinated axons. In this study, an intermediate ridge size of a few hundred nanometers was used to capture these different length scales. To construct the multi-well device, nanopattern-coated glass coverslips were irreversibly bonded to modified Nunc®uLabTek®eII Chamber Slide (cat. no. 154534) using biocompatible medical adhesive. Prior to attachment, pattern-coated glass coverslips were washed with 70% and 100% ethanol (EtOH), and allowed to air dry in a sterile environment. During and after construction, multi-well, nanopattern devices were maintained under sterile conditions.

Culture of cells in multi-well, nanopatterned device

[0077] Cells were cultured in the multi-well, nanopatterned device for approximately 48 hours in the course of each experiment. Prior to plating cells, nanoridged substrata were coated with poly-D-lysine (10 µg ml^-1) for 15 minutes and mouse laminin (from 10 to 140 µg ml^-1) for 1 hour. The laminin was based on Engelbreth-Holm-Swarm murine sarcoma (basement membrane), Sigma (cat. L2020). These topographically patterned cell substrata, caused cells to align with and move along the direction of the nanoridges. Both prior to, and during experiments cells were maintained at 37 °C and 5% CO2 in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 containing 2 mM L-glutamine, with added 50 U mL^-1 penicillin, 50 mg mL^-1 streptomycin. Where indicated, media contained 10% fetal bovine serum (Invitrogen) or alternatively, Platelet-Derive Growth Factor-AA (PDGF-AA) (LC Laboratories) at specified concentrations.

Quantitative analysis of cell morphology and migration

[0078] Because cell-cell contact is known to affect the extent of cell spreading and migration, cells were plated at low density (~4 x 10⁴ cells mL ^-1) to allow isolated movements. A custom-made MATLAB® script was used to allow manual tracking and measurement of cells frame by frame. Analysis was stopped in the event of cell death, division, collision with non-cellular debris, or movement out of the field of view. The area and spindle shape factor of individual cells were measured in each frame, and averaged over the entire duration of the experiment. The spindle shape factor was defined as the ratio of the maximum cell length (long axis) to the maximum cell width in the direction perpendicular to long axis, regardless of the orientation with respect to nanoridge pattern. Averages of cell populations were calculated from at least 60 cells. Cells movements were tracked by centroid position or by the approximate center of the cell body. Individual cell trajectories were used to calculate the mean squared displacement (MSD) at each interval. Instantaneous speeds of individual cells were calculated from the MSD and the duration of the image acquisition time interval. Average speeds of individual cells were calculated from the total distance moved throughout the entire cell trajectory and the total time the cell was tracked. Persistence was obtained as previously described (Pankov R. et al., A Rac switch regulates random versus directionally persistent cell migration, The Journal of cell biology 170, 793–802 (2005), which is hereby incorporated herein by reference in its entirety), by calculating the ratio of the shortest distance between starting point and end point, divided by the total distanced moved. Alignment to the nanoridge pattern was calculated by dividing the distance moved parallel to the ridges, by the distance moved perpendicular to the ridges. Averages of cell populations were calculated from at least 60 cells. Persistence distinguishes random, exploratory motility from continuous motion in a particular direction, a critical migratory mode for tumor dispersal. Alignment describes how strongly cells interact with the underlying substrate. Both spindle shape factor and alignment correlate with the structure and strength of cell-substrate adhesion complexes, which are critical regulators of cell motility and morphology.

PDGF-AA Ligand and Imatinib Methanesulfonate Salt

[0079] Platelet-derived growth factor-AA ligand (PDGF-AA) was purchased from R&D Systems (10 μl) and reconstituted in 500 μl of 0.1% BSA. Imatinib, Methanesulfonate Salt was purchased from LC Laboratories. A 10 mM stock solution was dissolved in distilled water and stored at -20 ºC, protected from light. Dilutions of the stock for both PDGF and Imatinib were prepared for use in cell culture medium and added directly to the cells when needed.

Reverse transcription- polymerase chain reaction analysis of PDGF Receptor

[0080] Total RNA was isolated from cell lysates by homogenizing cells with 1 mL of Trizol (Invitrogen) and incubating in 200 pl of Chloroform. Cells were incubated overnight in isopropyl alcohol and RNA was extracted using the RNeasy® Mini Kit according to the manufacturer’s instructions (Qiagen). A total of 1 pg of total RNA was reverse-transcribed into cDNA (Superscript III®; Invitrogen). Polymerase chain reaction (PCR) amplifications were performed with a Platinum® Pfx DNA Polymerase (Invitrogen) in a PCR thermal cycler (Geneamp® PCR System 9700, Applied Biosystems). After PCR amplification (5 minutes at 94 °C initial step, followed by 40 cycles of 15 seconds at 94°C, 40 seconds at 55 °C, 30 seconds at 68 °C followed by 7 minutes at 68°C), PCR products were analyzed on a 1.5% agarose gel (Invitrogen) containing SYBR® Safe DNA gel stain (Invitrogen) and imaged with Gel Logic® 100 Imaging System (Kodak). Quantitative RT-PCR was performed using SYBR® Green PCR Master Mix (Applied Biosystems) and 7300 Real Time® PCR Systems (Applied Biosystems). The thermal cycling conditions were as follows: 50 °C for 2 minutes, 95 °C for 10 minutes followed by 40 cycles of 95 °C for 15 seconds, 60°C for 30 seconds, 72 °C for 30 seconds and finalized with 72 °C for 10 minutes. GAPDH was amplified as endogenous control. The sequence of PDGF Receptor-α primers employed is: sense, 5’- CCT GGT CTT AGG CTG TCT TCT -3’ (SEQ ID NO: 1); antisense, 5’- GCC AGC TCA CTT CAC TCT CC -3’ (SEQ ID NO: 2). The GAPDH primers’ sequence is: sense, 5’- CAT GAG AAG TAT GAC AAC AGC CT -3’ (SEQ ID NO: 3); antisense, 5’- AGT CCT TCC ACG ATA CCA AAG T -3’ (SEQ ID NO: 4).

Immunocytochemistry

[0081] Spheroid and adherent cells were grown on cover slips. The cells were fixed with 4% paraformaldehyde for 30 minutes at room temperature and permeabilized with PBS containing 0.1% Triton X-100® for 5 minutes. The cells were incubated overnight with primary antibodies for PDGF Receptor alpha (1:100; Santa Cruz) and then incubated with the appropriate secondary antibody conjugated with fluorescent dye (1:500) for one hour. Cells were subsequently stained against DAPI (1:200). Coverslips were mounted with Aquamount®. 15-20 spheroids were placed in DMEM/F12 without growth factors for 18 hours and then exposed to PDGF-AA ligand for 24 hours. Whole spheroids were fixed with 4% paraformaldehyde for thirty minutes at room temperature and stained against Ki67 (1:200; Thermo) as previously described (Guerrero-Cazares, H., Chaichana, K. L., Quinones- Hinojosa, A., Neurosphere culture and human organotypic model to evaluate brain tumor stem cells, Methods in molecular biology 568, 73 (2009), which is hereby incorporated herein by reference in its entirety).

Western Blots

[0082] Total cellular protein was extracted using NE-PER Nuclear and Cytoplasmic Extraction Reagents kit according to the manufacturer’s instructions (Thermo Scientific) containing protease (Roche) and phosphatase inhibitor (Thermo). Protein concentration was determined using the Bradford protein quantification method (Biorad Protein Assay, Biorad). SDS-PAGE was performed with 25 pg total cellular protein per lane using 4-12 % gradient Tris-glycine gels. The primary antibodies used were as follows: anti PDGFR-alpha (1:200; Santa Cruz); phospho-PDGFR alpha (1:1000; Cell Signaling); Akt (1:1000; Cell Signaling); phospho-Akt (1:1000; Cell Signaling).

Proliferation Assay

[0083] Edu incorporation was used as a measure of proliferation. For EdU incorporation experiments, cells (6 x 10E5) were cultured in DMEM/F12 medium without growth factors for 18 hours in a six well plate. Cells were exposed to EdU (10 µM Click-iT® EdU Flow Cytometry Assay Kit, Invitrogen) and PDGF-AA ligand (20 ng/mL) or a combination of PDGF-AA (20 ng/mL) and Imatinib (10uM) for 24 hours. After incubation cells were centrifuged, the supernatant was discarded and the pellet suspended in 100 µl of 4% paraformaldehyde for 15 minutes. Cells were washed in 1% BSA and incubated with 100 µl of saponin-based permeabilization buffer for 10 minutes. After an additional wash in 1% BSA cells were incubated in 250 µl of Click-iT® Reaction Cocktail for 30 minutes. All procedures were performed according to manufacturer’s instructions except for the volumes used to prepare the Click-iT® Reaction buffer. Half of the recommended volume was used for all reagents. Flow Cytometry

[0084] Flow cytometry was performed using a FACSCaliber^TM Flow Cytometer (BD Biosciences) and data was analyzed with Kaluza® Flow Cytometry Software (Beckman Coulter). Analysis of 30,000 total events was performed after exclusion of dead cells by FSC/SSC gating. Fluorescence was measured in the FL4 channel.

Time-Lapse Microscopy of Live Cells and Quantitative Analysis of Cell Morphology and Migration

[0085] Cell migration was observed using time-lapse microscopy (data not shown). To enable long-term observation, the multi-well, nanopatterned device was mounted on the stage of a motorized inverted microscope (Olympus IX81) equipped with a Photometrics hCascade® 512B II charge-coupled device camera and temperature- and gas-controlling environmental chamber. Phase-contrast cell images were automatically recorded under a 10× objective (numerical aperture = 0.30) using SlideBook® 4.1 software (Intelligent Imaging Innovations) for 10–15 hr at 10 or 20 min intervals. Because cell-cell contact is known to affect the extent of cell spreading and migration, cells were plated at low density (∼4 × 104 cells ml−1) to allow isolated movements. A custom-made MATLAB® script was used to allow manual tracking and measurement of cells frame by frame.

Tumor Xenografts

[0086] Animal protocols were approved by the Johns Hopkins School of Medicine Animal Care and Use Committee. For intracranial xenografts, severe combined immunodeficiency mice received 100,000 viable cells in 1 μl of DMEM/F12 serum media without growth factors by stereotactic injection into the right striatum. Cells were cultured in DMEM/F12 serum media with epidermal growth factor, fibroblast growth factor, and PDGF ligand for 3 weeks before injections were performed. Cell viability was determined by trypan blue dye exclusion. Mice were perfused with 4% paraformaldehyde at the indicated times, and the brains were removed for histological analysis. Patient Clinical Information Used in the Study

[0087] The following table provides information about the dataset (e.g., as it relates to Figures 4A-7F). The table shows the patients clinical data (from which the primary GBM cell lines were derived). The table contains over 35 factors related to each patient's tumor, general health, and demographics, including tumor size, tumor shape, therapeutic regimen, age and are sorted according to their response to PDGF.

Table 1

Statistical A nalysis

[0088] Results are presented as mean + SEM. The Mann-Whitney rank-sum test was for pairwise comparisons; Dunn's test (rank-based ANOVA) was used in multiple group comparisons. When noted, Student's t test or standard ANOVA (the Holm-Sidak method) was used. Univariate Cox analysis was used to identify correlations among tumor characteristics. To group data, thresholds were determined using linear discriminant analysis as previously described (Lin, B., et al., Synthetic spatially graded Rac activation drives cell polarization and movement, Proc. Natl. Acad. Sci. USA 109, E3668– E3677 (2012)). Statistics were analyzed using Sigmaplot®, GraphPad® Prism, and MATLAB® software.

EXAMPLE 2

CONSTRUCTION AND APPLICATION OF A PHENOTYPIC-SCREENING

PLATFORM

[0089] To create 1D fibrillar surfaces that mimic nanometer-scale features of the 3D ECM microenvironment, topographic patterns consisting of regular, parallel ridges (Figures 1A, 1B, and 8A–8C) similar in size to those found in the brain tissue ECM were fabricated (Kim, D.-H., et al., Guided cell migration on microtextured substrates with variable local density and anisotropy, Adv. Funct. Mater. 19, 1579–1586 (2009); Bellail, A.C., et al.,, Micro- regional extracellular matrix heterogeneity in brain modulates glioma cell invasion. Int. J. Biochem, Cell Biol. 36, 1046–1069 (2004); Ottani, V., et al., Collagen structure and func- tional implications, Micron 32, 251–260 (2001), all of which are hereby incorporated herein by reference in their entirety). How well cell migration on the quasi-3D, fibrillar topography approximated cell migration in a true 3D ECM environment was explored. Experiments were performed that compared the morphology, directionality, and speed of GBM 612 cells on the quasi-3D platform and cells migrating in distinct 3D settings: Matrigel matrix and organotypic human brain slice cultures (Figure 1C). It was found that cell morphology on fibrillar, but not flat, surfaces mirrored the characteristic cell shapes found in 3D environments. Cell morphology in collagen 3D matrices was similar to that observed in the quasi-3D tissue mimetic platform. Furthermore, in contrast to 2D surfaces, migration speed was enhanced to levels similar to those observed in 3D settings (Figure 1C), displaying essentially 1D migration patterns. The same observations were replicated using other patient- derived cell lines (GBM 318 and GBM 276) (Figures 8B, 8D, and 9C). Consistent with a tendency of mobile cells to align parallel to oriented topographic structures, the direction of cell migration was strongly biased along the axis of the ridge pattern (Figures 8D and 9C). Compared to smooth surfaces, cells cultured on the platform showed increased cell area and spindle shape factor (Figures 8E, 9A, and 9B), and their migration was enhanced based on the three metrics scored: average speed, alignment, and persistence (Figures 8F, 8G, 9D, and 9E). These results suggested a high degree of similarity between the quasi-3D platform and the true 3D microenvironments, possibly due to similar molecular mechanisms observed in cells migrating on 1D fibrillar surfaces and 3D matrices but distinct from those observed in cells cultured on 2D surfaces. To test the similarity between 1D fibrillar surfaces and 3D microenvironments, the effect of the Rho-associated protein kinase (ROCK) inhibitor Y- 27632 on the migration of GBM 612 cells was analyzed (Figure 1D). It was found that Y- 27632 inhibited migration on fibrillar surfaces but not on flat surfaces, suggesting that myosin II plays a role in migration on the fibrillar surfaces, similar to findings in vivo. These experiments were replicated with GBM 965 (Figures 9F and 9G). Using the multi-well setup of the device, experiments were performed that explored the influence of two critical environmental cues: ECM density and growth factor (Figures 10A–12D). On the fibrillar surfaces, but not on the flat surfaces, a strong dependence of cell velocity values on the surface density of laminin was observed (Figures 10A, 10B, 12A, and 12B). The density at which cell migration speed was maximized was used in all subsequent experiments. Next, the distributions of cell velocities were observed. It was found that cell migration was highly heterogeneous, displaying a substantial number of outliers, some exceeding the average cell migration 3- to 4-fold (Figure 1E). These results confirm that the heterogeneity of cell behavior can be assessed in this platform, enabling the identification and characterization of rare cells with extreme properties.

EXAMPLE 3

MIGRATORY BEHAVIOR OF GBM CELLS CAN BE ALTERED IN RESPONSE TO A COMBINATION OF PDGF AND NANOTOPOGRAPHIC CUES

[0090] Further experiments explored whether 3D-like cell migration might be differentially sensitive to the effects of growth factors implicated in the onset and progression of glioma, e.g., PDGF-AA (PDGF). PDGF can control glioma cell proliferation, but its effect on GBM migration and invasion is less clear. The effects of PDGF on cell speed and persistence at different doses were tested. On the quasi-3D platform, what was found was a dose-dependent response, with maximal motility achieved at intermediate PDGF concentrations (Figures 11B, 12C, and 12D). Similar experiments on flat substrata showed more limited response to PDGF (Figure 11A). Subsequent experiments explored whether the effect of PDGF could be ascribed to the activation of PDGF receptor alpha (PDGFRα). PDGFRα is thought to be the exclusive receptor for the PDGF-AA isoform employed in this study. First, the expression levels of PDGFRα in patient-derived cell lines by RT-PCR and immunoblotting were examined (Figures 2A, 12B, and 13A–13C). Then, the migratory behavior of cells with low and high PDGFRα expression levels was analyzed (GBM 253 and GBM 276, respectively). Following exposure to PDGF, what was found was an increase in cell speed and directionality of GBM 276 cells (high PDGFRα expression) but no response in GBM 253 (low PDGFRα expression) (Figures 2C, 2D, and 3A–3C). To further ascertain that PDGFRα was functionally involved, the tyrosine kinase inhibitor imatinib was used, which led to the finding that migration was attenuated to similar levels in both cell lines (Figures 2C, 2D, 13D, and 13E). However, a detailed analysis of the GBM 276 subpopulations showed a highly heterogeneous single-cell response to PDGF. In particular, only the fastest 25% quartile of cells responded to PDGF, with the response abrogated by the inhibitor (Figures 3B and 3C). Furthermore, the response of the slowest 25% quartile of GBM 276 cells was analogous to that of the PDGF-unresponsive cell line GBM 253 (Figures 3A and 3B). These results suggest that migration of only a subset of cells is responsive to PDGF stimulation and that this subset represents the fastest cell subpopulation.

EXAMPLE 4

PDGF ENHANCES INVASIVENESS OF PATIENT-DERIVED CELLS IN VIVO

[0091] Using orthotopic human GBM tumor models in mice (Garzon-Muvdi, T., et al., Regulation of brain tumor dispersal by NKCC1 through a novel role in focal adhesion regu- lation, PLoS Biol. 10, e1001320 (2012); Gonzalez-Perez, O., Guerrero-Cazares, H., and Quin^˜ones-Hinojosa, A., Targeting of deep brain structures with microinjections for delivery of drugs, viral vectors, or cell transplants, J. Vis. Exp. (46), 2082 (2010), all of which are hereby incorporated herein by reference in their entirety), the role of PDGF in tumor growth and survival was explored. First, the behavior of the xenografts using GBM 276 cells, cultured in the presence or absence of PDGF for 3 weeks before injection was examined. This PDGF preculture selects for PDGF-responsive cells by stimulating their growth and enriching this cell subpopulation. In vitro analysis of GBM 276 cells showed that increased proliferation correlated with the PDGF-induced migratory response (some data not shown; Figures 14C–14F). Significantly reduced survival of mice injected with PDGF- preconditioned cells was observed (n = 4 each group) (Figure 3D). Although this result suggested the importance of highly PDGF-responsive cells for tumor aggressiveness, other effects of prolonged PDGF exposure could be excluded, e.g., transdifferentiation (Figure 14E). Thus, the putative role of PDGF in tumor spreading was tested by supplying this factor exogenously in vivo to existing tumor xenografts via infusion pump. Quantification of tumor size and qualitative analysis by a blinded neuropathologist suggested that continuous exposure of tumor xenografts to PDGF generated larger, more invasive tumors with more eccentric shapes (Figure 14A). These samples displayed features indicative of migration along fiber tracts. Also observed was increased dispersion of GBM cells beyond the tumor margins (Figure 14B). These findings confirm the role of this growth factor in tumor induction and progression, and they are consistent with clinical data from The Cancer Genome Atlas (TCGA) (Goswami, C.P., and Nakshatri, H., PROGgene: gene expression based survival analysis web application for multiple cancers. J. Clin. Bioinforma, 3, 22 (2013), which is hereby incorporated herein by reference in its entirety) suggesting significant correlation between PDGF expression and survival (Figure 3E). TCGA data did not show a similar correlation for PDGFRα expression (Figure 3F). Because the genomic data did not support the correlation between average PDGFR expression and patient outcomes, the correlation between the tumor characteristics and the cell migration data obtained in the tissue-mimetic platform was examined.

EXAMPLE 5

SCREENING HETEROGENEITY WITHIN AND BETWEEN PATIENT-SPECIFIC

TUMOR SAMPLES

[0092] Heterogeneity of cell properties within the same tumor reflects subpopulations promoting tumor growth, progression, and therapeutic resistance. GBM also has populations with distinct expression profiles of receptor tyrosine kinases, particularly PDGFRα. This heterogeneity can be tackled by analysis on the single-cell level, which is yielded in the quasi-3D platform with less than 1,000 cells (particularly beneficial for screening precious intraoperative human tissue specimens). By taking advantage of the single-cell resolution to quantify the distribution of cell speed in control versus PDGF-exposed conditions, the difference in migratory behaviors among 14 glioblastoma patients was investigated (Figures 4A–4C). Both intra- and inter-patient differences in the cell population behavior were found. When analyzing GBM 499 cells based on their speed, the total population average showed no significant response to PDGF, while analysis of the 25% fastest quartile subpopulation revealed significant differences (Figure 3A). Such masked responses and heterogeneities were also present in the time-domain data. For instance, two patient samples, GBM 501 and GBM 609, responded significantly to PDGF. However, for GBM 501, this response was not persistent throughout the experiment duration, in contrast to the response of GBM 609 (Figure 3B). Finally, the platform allowed investigation of high-speed outliers. For example, for both the GBM 630 and the GBM 544 samples, cells experienced a significant increase in migration speed in response to PDGF. However, a detailed analysis of the speed distribution revealed that only for GBM 630 were fast-moving outliers clearly identifiable; GBM 544 showed a substantially more uniform response (Figure 3C).

[0093] Considerable differences in the cell speed across the spectrum of 14 patient- derived samples, both in the presence and in the absence of PDGF, were observed (Figure 5). Recognizing these patient-specific differences, analyses were performed on the distinct features of single-cell response based on the criteria presented in Figures 4A–4C, namely, based on the average population response (group I), persistence of response over time (group II), and presence of outliers (group III) (Figure 5). For each of these criteria, PDGF- responsive and PDGF-unresponsive groups were identified. There was overlap among strongly PDGF-responsive groups in all groups, which were thus treated as the consensus PDGF-responsive samples (marked in Figure 5). However, several patient samples (GBM 544, GBM 549, and GBM 501) were PDGF responsive by some criteria. To determine whether PDGFRα expression level could serve as a molecular marker predictive of enhanced PDGF responsiveness, the tumor samples were assayed for expression of this receptor. Although all consensus PDGF-responsive samples had high expression of PDGFRα (Figures 2A, 2B, 13B, and 13C), similarly high levels of the receptor were found in several samples in the consensus PDGF-unresponsive group (GBM 626, GBM 612, and GBM 854). The failure of these samples to respond in migration experiments may be due to differences in downstream effectors of PDGFRα, stressing the difficulty of using molecular markers alone in the classification of the aggressive migration phenotype, as is evident from clinical data showing no correlation between PDGFRα expression and survival (Figure 3F). Differential expression of PDGFRα across the groups might also relate to the subtype of the patient tumors. A subclassification study of the used GBM samples only yielded two subtypes: mesenchymal and proneural. Previous studies have observed higher amplification and mutation rates of the PDGFRA gene in the proneural subtype of GBM (Verhaak, R.G.W., Hoadley, K.A., Purdom, E., Wang, V., Qi, Y., Wilkerson, M.D., Miller, C.R., Ding, L., Golub, T., Mesirov, J.P., et al.; Cancer Genome Atlas Research Network, Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1, Cancer Cell 17, 98–110 (2010), which is hereby incorporated herein by reference in its entirety). A clear correlation between expression of this gene and tumor subclass was not observed. High levels of PDGF-triggered cell migration can be achieved in various ways, and they may not be revealed by a simple molecular signature. However, the aggressive migration phenotype can be translated into enhanced invasiveness. These results highlight potential advantages of the disclosed single-cell phenotype analysis.

EXAMPLE 6

MIGRATORY BEHAVIOR CORRELATES WITH CLINICAL TUMOR CHARACTERISTICS

[0094] The degree of cell migration may reflect the propensity for invasive tumor spread. More than 35 factors related to each patient’s tumor, general health, and demographics were examined (Table 1). It was found that the migratory response of GBM samples to PDGF correlated with time to tumor recurrence after surgical resection (Figures 6A and 6B). This correlation was particularly significant when the analysis was focused on the consensus- responsive and consensus-unresponsive groups (Figure 6A). In comparisons to the whole-cell populations, correlations were more significant for the aggressively moving cells: either the fastest 25% of the cells or the outlier population (Figure 6B). Several characteristic tumor features visualized in the MRI of the patients with the responsiveness to PDGF were contrasted (e.g., Figures 6B and 6D). Tumors from the consensus PDGF-responsive subset (e.g., Figure 6D) were larger and more spread out than those from the PDGF-unresponsive subset. Statistically significant differences in the anatomical location of the tumors were observed; all consensus PDGF-responsive samples were in the frontal lobe (Figure 7E), but temporal lobe tumor samples commonly fell into the PDGF-unresponsive subgroup. Migration analysis of GBM cells revealed that higher directionality (i.e., alignment of the migration to the patterns) correlated with longer recurrence times (univariate Cox analysis, p = 0.002) (Figures 7F and 15A). Because alignment of the migration is associated with the strength of cell-substrate adhesion, this result may reflect a higher propensity in more aligned cells to adhere to ECM, leading to retarded migratory response and delayed tumor spread. Blind, qualitative analysis of patient tumor samples by a neuropathologist revealed that the cells were small and that the tumors yielding PDGF-responsive samples had marked microvascular proliferation. The latter feature is commonly associated with advanced progression, receptor tyrosine kinase amplification, and worse prognosis. PDGF-responsive samples emerged from tumors that resulted in shorter recurrence times, after controlling for factors known to be associated with recurrence (age, Karnofsky Performance Scale score, extent of resection, and adjuvant therapy) (p = 0.0009) (Table 1). However, survival times showed a less significant correlation to PDGF responsiveness with the limited number of patient samples tested (Figure 15B).

[0095] These findings were supplemented with comparisons to more traditional protein expression analyses. First, the tumor sample was segregated into groups based on PDGFRα expression, which revealed trends in patient tumor characteristics that were supportive of the performed migration analysis (Figure 5). However, grouping the samples according to PDGFRα expression or molecular sub-classification did not yield significant differences in predicting time to recurrence (Figures 16A–16B). Moreover, the difference in tumor location between high- and low-expression groups was not statistically significant. Second, the public Rembrandt database of glioblastoma patients was mined to investigate the relationship between tumor location and PDGFRα expression. Examination of 47 patients revealed no significant relationship between the two characteristics (Figures 16C–16D). Taken together, these findings suggest that RNA expression is a weaker predictor of patient outcomes than is the disclosed phenotypic migration analysis.

Discussion

[0096] The heterogeneity and invasive nature of glioblastoma and other aggressive cancers highlights the importance of assaying cell migration as a phenotypic feature predictive of clinical outcomes. Herein is described a simple but information-rich experimental platform aimed at the analysis of primary patient samples on a single-cell level. This platform allows high-throughput screening of the effects of variable extracellular milieu. Using this method on a range of patient-derived samples and contrasting the results of the analysis with respective clinical information revealed substantial predictive power, particularly when cell migration was examined in conjunction with the effects of PDGF. This result strongly suggests that cell migration, as examined in structured, mechanically defined culture conditions, can be predictive of more complex in vivo invasion processes and can be a powerful phenotypic analysis tool with strong clinical implications. Prior attempts to examine glioma cell migration and its relationship to tumor progression have used 2D surfaces and have not achieved such direct predictions of patient-specific tumor features as has the present work. This emphasizes the benefits of analyzing heterogeneities within samples and using surfaces that better mimic in vivo conditions.

[0097] The significant correlation of migratory behavior with time to recurrence and tumor location provides crucial insight into this disease. Recurrence of glioblastoma after tumor resection is the primary cause of death in patients and is one of the most important predictors of future patient outcomes. This study provides a simple method to glean information about these phenomena. The described finding that most PDGF-unresponsive tumors are derived from the temporal lobe could suggest that PDGF signaling is less critical to tumor progression in this region of the brain. Direct access to individual cell migration analysis can be important for future treatment modalities. The described experimental platform has important advantages over 2D migration assays, because it provides a cellular environment similar to in vivo conditions (as evidenced in the similarity of several aspects of migration in ex vivo human brain tissue and a 3D hydrogel, e.g., increased cell polarity and migration speed). These factors can be important in migration. Another advantage is the reduced number of cells required when compared to commonly used transwell migration assays. Furthermore, transwell assays fail to yield the information on migration and morphology of individual cells and only originate endpoint information. A substantial degree of heterogeneity was found in the glioblastoma samples analyzed. The increased average migration speed of a cell population in the presence of PDGF was ascribed to a small subpopulation of aggressive cells (approximately 25%). Knowledge of the degree of population heterogeneity can be critical to the decision-making in the clinic. In addition, the quasi-3D tissue mimetic platform can distinguish the effects of cell proliferation and migration phenotypes, which can be a confounding factor in both transwell and in vivo migration studies.

[0098] The described results also highlight advantages of the proposed method over traditional protein expression assays. An incomplete correlation between receptor expression and response to PDGF signaling was observed, possibly due to veiled differences in the signal transduction pathways. The results demonstrated that protein expression analysis was less sensitive and less robust at predicting differences among patient tumor features. Genomic and proteomic approaches also suffer from limited supplies of primary tissues available for their cumbersome requirements of cellular material.

[0099] Overall, the results here support the proposed methodology as a simpler, more biomimetic, and informative method to gain critical information about patient tumors and cell populations. The analysis presented here reveals the importance of careful engineering of chemical and mechanical extracellular milieu in cell migration analysis. This methodology will provide an important prognostic tool, with benefits that include high-throughput, label- free analysis of single-cell resolution; low demand for precious primary cell samples; and better physiological relevance compared to other migration assays.

[00100] Although the invention has been described with reference to the above example, it will be understood that modifications and variations are encompassed within the spirit and scope of the invention. Accordingly, the invention is limited only by the following claims.

Claims

What is claimed is: 1. A method to classify a cell in a heterogeneous population of cells comprising:

providing a surface comprising a plurality of parallel ridges, wherein the ridges have a depth between about 50 to 1000 nm and a width between about 50 and 1000 nm;

applying at least one cell attachment molecule to the surface;

contacting a plurality of living cells with the surface for a time and under conditions to allow the cells to attach to the surface, wherein a migration speed or migration direction of the attached cells is detected; and

classifying the cells based on their migration speed, migration direction, and/or the change in speed or direction over time.

2. The method of claim 1, wherein the cells are cancer cells.

3. The method of claim 2, wherein the cancer cells are selected from the group consisting of carcinoma, sarcoma, lymphoma, leukemia, germ cell tumor, and glioblastoma.

4. The method of claim 1, wherein the cells are contacted with at least one agent selected from the group consisting of drugs, natural compounds, toxins, nanoparticles, nucleic acids, viruses, bacteria and other microbes, mammalian cells, hormones, growth factors, and cytokines prior to or simultaneous with the step of contacting the cells with the surface.

5. The method of claim 4, wherein the growth factor is platelet-derived growth factor.

6. The method of claim 1, wherein the cells are classified as a strong responder or a weak responder based on speed of migration.

7. The method of claim 1, wherein the cells are classified as a strong responder or a weak responder based on a speed of migration of 50% or less of the cells that have speeds of migration faster than all other cells.

8. A method for determining an effect of an agent on a cell attached to a surface, wherein the surface comprises a plurality of parallel ridges comprising:

contacting the cells with platelet-derived growth factor; and

measuring a change in migration speed and/or migration direction of cells, thereby classifying the cells as a strong or a weak responder based on a speed of migration of 50% or less of the cells that have speeds of migration faster than all other cells.

9. A method for classifying a cancer cell comprising:

detecting a migration speed or a migration direction of a cell;

contacting the cell with an agent; and classifying the cell based on migration speed, migration direction, and/or a change of speed or direction.

10. The method of claim 9, wherein the cancer cells are glioblastoma cells.

11. The method of claim 9, wherein the growth factor is platelet-derived growth factor.

12. The method in claim 9, wherein the cells are classified as a strong responder or a weak responder based on a speed of migration.

13. The method of claim 11, wherein the cells are classified as a strong responder or a weak responder based on a speed of migration of about 25% of the cells that have speeds of migration faster than other cells.

14. A method of phenotyping a cancer cell sample comprising:

placing a sample comprising a cancer cell onto a surface, wherein the surface comprises a plurality of parallel ridges; and

assessing at least one migratory characteristic of the cancer cell, thereby phenotyping the cancer cell sample based on the migratory characteristic.

15. The method of claim 14, wherein the cancer cell is selected from the group consisting of carcinoma cells, sarcoma cells, lymphoma cells, leukemia cells, germ cell tumor cells, and glioblastoma cells.

16. The method of claim 15, wherein the cancer cell is a glioblastoma cell.

17. The method of claim 16, wherein the glioblastoma cell is obtained from a resected brain tissue of a patient having glioblastoma multiforme.

18. The method of claim 14, wherein the surface further comprises a molecule selected from the group consisting of a collagen, fibronectin, laminin, poly-D-lysine, poly-L- ornithine, proteoglycan, vitronectin, and polysaccharide.

19. The method of claim 14, wherein the parallel ridges are spaced from each other by substantially uniform inter-ridge distances.

20. The method of claim 19, wherein the parallel ridges have heights of about 500 nanometers, widths of about 350 nanometers, and inter-ridge distances of about 1.5 micrometers.

21. The method of claim 14, wherein the migratory characteristic of the cancer cell is observed using an optical imaging system in a time-resolved mode.

22. The method of claim 14, wherein the migratory characteristic is observed at a single- cell resolution.

23. The method of claim 14, wherein the migratory characteristic comprises a speed that corresponds to a ratio of migration distance to migration time.

24. The method of claim 14, wherein the migratory characteristic comprises a directionality that corresponds to a ratio of migration distance parallel with a ridge to migration distance perpendicular to a ridge.

25. The method of claim 14, wherein the migratory characteristic comprises a persistence that corresponds to a ratio of shortest migration distance between a start point and an end point to total migration distance between the start point and the end point.

26. The method of claim 14, wherein the sample comprises a plurality of cancer cells.

27. The method of claim 26, wherein the cancer cells are classified based on a migratory characteristic of a subset of the cancer cells.

28. The method of claim 27, wherein the cancer cells are classified based on a speed of about a quarter of the cancer cells that are faster moving than any other cancer cells.

29. The method of claim 27, wherein the cancer cells are classified based on a change in the migratory characteristic between the migratory characteristic of a subset of a test cancer cell sample contacted with an agent and the migratory characteristic of a subset of a control cancer cell sample not contacted with the agent.

30. The method of claim 29, wherein the agent is platelet-derived growth factor.

31. The method of claim 30, wherein the change in the migratory characteristic comprises a percentage increase in average migration speed of the subset of the test cancer cell sample.

32. The method of claim 30, wherein the change in the migratory characteristic comprises a percent of time when migration speed is increased in the subset of the test cancer cell sample.

33. The method of claim 30, wherein the change in the migratory characteristic comprises a percent of cells from the test cancer cell sample that are faster than all cells from the control cancer cell sample.

34. The method of claim 30, wherein the migratory characteristic comprises a consensus characteristic obtained by analyzing each of a percentage increase in average migration speed of the subset of the test cancer cell sample, a percent of time when migration speed is increased in the subset of the test cancer cell sample, and a percent of cells from the test cancer cell sample that are faster than all cells from the control cancer cell sample.

35. A method of identifying an agent which reduces the invasiveness of a cancer cell comprising:

contacting a platelet-derived growth factor with a sample containing a cancer cell to obtain a processed cancer cell sample; contacting an agent with the processed cancer cell sample to obtain a treated cancer cell sample;

placing the treated cancer cell sample onto a surface comprising a plurality of parallel ridges, thereby allowing the cancer cell to migrate on the surface;

determining at least one migratory characteristic of the cancer cell; and

determining that the agent reduces invasiveness of the cancer cell if the migratory characteristic of the cancer cell from the treated sample and a migratory characteristic of a cancer cell from an unprocessed and untreated cancer cell sample are both less than or greater than a migratory characteristic of a cancer cell from a processed but untreated cancer cell sample.

36. The method of claim 35, wherein the cancer cell is selected from the group consisting of a carcinoma cell, sarcoma cell, lymphoma cell, leukemia cell, germ cell tumor cell, and glioblastoma cell.

37. The method of claim 36, wherein the cancer cell comprises a glioblastoma cell.

38. The method of claim 37, wherein the glioblastoma cell is obtained from a marginal area of a glioblastoma tumor in a brain of a patient having glioblastoma multiforme.

39. The method of claim 35, wherein the surface further comprises a molecule selected from the group consisting of a collagen, fibronectin, laminin, poly-D-lysine, poly-L- ornithine, proteoglycan, vitronectin, and polysaccharide.

40. The method of claim 35, wherein the parallel ridges are spaced from each other by uniform inter-ridge distances.

41. The method of claim 40, wherein the parallel ridges have heights of about 500 nanometers, widths of about 350 nanometers, and inter-ridge distances of about 1.5 micrometers.

42. The method of claim 35, wherein the cells are placed onto the surface through a multi- well dish.

43. The method of claim 35, wherein the migratory characteristic of the cancer cell is observed using an optical imaging system in a time-resolved mode.

44. The method of claim 35, wherein the migratory characteristic is observed at a single- cell resolution.

45. The method of claim 35, wherein the migratory characteristic comprises a speed that corresponds to a ratio of total migration distance to total migration time.

46. The method of claim 35, wherein the migratory characteristic comprises a directionality that corresponds to a ratio of migration distance parallel with a ridge to migration distance perpendicular to a ridge.

47. The method of claim 35, wherein the migratory characteristic comprises a persistence that corresponds to a ratio of shortest migration distance between a start point and an end point to total migration distance.

48. A method of identifying an agent that reduces aggressiveness of glioblastoma cells comprising:

contacting a platelet-derived growth factor with a sample containing a plurality of glioblastoma cells to obtain a processed glioblastoma cell sample;

contacting an agent with the processed glioblastoma cell sample to obtain a treated glioblastoma cell sample;

placing the treated glioblastoma cell sample onto a surface comprising a plurality of parallel ridges having uniform inter-ridge distances, thereby allowing the glioblastoma cells to migrate on the surface;

determining an average speed of a subset of the treated glioblastoma cells, wherein the speed corresponds to a ratio of migration distance to migration time, wherein the subset comprises at most half of the cells, and wherein the subset comprises cells each of which is faster moving than all treated glioblastoma cells not within the subset; and

identifying the agent as one that reduces aggressiveness of glioblastoma cells if the average speed determined for the treated cells and an average speed determined for unprocessed and untreated cells are both less than or greater than an average speed determined for processed but untreated cells.

49. A nanopatterned substrate having a plurality of substantially parallel ridges, each ridge having a height of between about 400 to 600 nm and a width of between about 300 to 1000 nm.

50. The substrate of claim 49, wherein each ridge has a height of about 500 nm and a width of about 350 nm.

51. The substrate of claim 49, wherein each ridge is separated by a groove having a width of between about 100 to 2500 nm.

52. The substrate of claim 51, wherein each ridge is separated by a groove having a width of about 1500 nm.

53. The substrate of claim 52, wherein each ridge has a height of about 500 nm and a width of about 350 nm.

54. The substrate of claim 49, wherein the substrate is planar.

55. The substrate of claim 49, wherein the substrate is composed of at least one material selected from the group consisting of silicon, silicon dioxide, a polymer and quartz.

56. The substrate of claim 55, wherein the substrate comprises silicon dioxide.

57. The substrate of claim 55, wherein the at least one material is a polymer selected from the group consisting of acrylate, polyethylene, polycarbonate, polyimide, polydimethylsiloxane and cyclo-olefin copolymer.

58. The substrate of claim 49, wherein the ridges are formed by colloidal lithography, polymer demixing, electrospinning, nanoimprinting, dip-pen nanolithography or capillary force lithography.

59. The substrate of claim 58, wherein the ridges are formed by capillary force lithography.

60. The substrate of claim 58, wherein the substrate is coated with at least one cell attachment molecule selected from the group consisting of collagen, fibronectin, laminin, poly-D-lysine, poly-L-ornithine, proteoglycan, vitronectin, and polysaccharide.

61. The substrate of claim 60, wherein the at least one cell attachment molecule is poly- D-lysine and laminin.

62. A system for measuring cell motility comprising:

the substrate of any of claims 49–61; and

a detector for detecting movement of a cell on the substrate.

63. The system of claim 62, wherein the detector is an optical imaging device.

64. The system of claim 63, wherein the optical imaging device is a microscope.

65. The system of claim 62, further comprising a computing device having functionality to analyze cell motility data.

66. The system of claim 62, wherein the system is configured for multi-well cellular analysis.

67. A kit comprising:

the substrate of any of claims 49–61; and

a platelet-derived growth factor receptor alpha (PDGFRα) ligand.

68. The kit of claim 67, further comprising instructions for performing cellular analysis and packaging.

69. The kit of claim 67, further comprising a reagent comprising at least one cell attachment molecule selected from the group consisting of collagen, fibronectin, laminin, poly-D-lysine, poly-L-ornithine, proteoglycan, vitronectin, and polysaccharide.

70. The kit of claim 67, wherein the PDGFRα ligand is platelet-derived growth factor (PDGF).

71. The kit of claim 67, further comprising one or more reagents for culturing a cell.

72. A method of treating glioblastoma comprising administering to a subject having a glioblastoma tumor or prior resected tumor with a therapeutically effective amount of a platelet-derived growth factor receptor alpha (PDGFRα) antagonist, thereby treating the glioblastoma.

73. A method for reducing invasiveness of a cancer cell comprising contacting the cell with an agent suspected of affecting invasiveness of the cell and determining a migration speed of the cell before and after contact with the agent in a system of claim 62, wherein a reduction in migration speed is indicative of an agent that reduces the invasiveness of the cancer cell.

74. A method of reducing the time to recurrence after resection of a tumor containing a cancer cell comprising determining a migration speed of the cell before and after contact with an agent in a system of claim 62, wherein a reduction in the migration speed is indicative of an agent that reduces the time to recurrence after resection of the tumor.

75. The method of claim 1, wherein the molecule, is selected from the group consisting of collagen, fibronectin, laminin, poly-d-lysine, poly-l-ornithine, proteoglycan, vitronectin, and polysaccharide.

76. The method of claim 9, wherein the agent is selected from the group consisting of a drug, a natural compound, a toxin, a nanoparticle, a nucleic acid, virus, bacteria, mammalian cell, a biological ligand, a hormone, a growth factor, and a cytokine.