Guidelines for the selection of functional assays to evaluate the hallmarks of cancer.

The hallmarks of cancer capture the most essential phenotypic characteristics of malignant transformation and progression. Although numerous factors involved in this multi-step process are still unknown to date, an ever-increasing number of mutated/altered candidate genes are being identified within large-scale cancer genomic projects. Therefore, investigators need to be aware of available and appropriate techniques capable of determining characteristic features of each hallmark. We review the methods tailored to experimental cancer researchers to evaluate cell proliferation, programmed cell death, replicative immortality, induction of angiogenesis, invasion and metastasis, genome instability, and reprogramming of energy metabolism. Selecting the ideal method is based on the investigator's goals, available equipment and also on financial constraints. Multiplexing strategies enable a more in-depth data collection from a single experiment - obtaining several results from a single procedure reduces variability and saves time and relative cost, leading to more robust conclusions compared to a single end point measurement. Each hallmark possesses characteristics that can be analyzed by immunoblot, RT-PCR, immunocytochemistry, immunoprecipitation, RNA microarray or RNA-seq. In general, flow cytometry, fluorescence microscopy, and multiwell readers are extremely versatile tools and, with proper sample preparation, allow the detection of a vast number of hallmark features. Finally, we also provide a list of hallmark-specific genes to be measured in transcriptome-level studies. Although our list is not exhaustive, we provide a snapshot of the most widely used methods, with an emphasis on methods enabling the simultaneous evaluation of multiple hallmark features.


Introduction
Tumorigenesis is a multi-step process that progressively converts normal cells into malignancies. The six hallmarks characterizing neoplasia (sustaining proliferative signaling, evading growth suppressors, resisting cell death, enabling replicative immortality, inducing angiogenesis, and activating invasion and metastasis) were delineated in Hanahan and Weinberg's seminal work compressing decades of cancer research data into a single paper [1]. Since then further features of malignant conversion have been described. Genomic instability generates random mutations, while inflammation nurtures and accelerates tumorigenesis. Reprogramming of energy metabolism and avoiding immune destruction emerged as important supporting features and epigenetics was accepted as a player of equal importance to genetic abnormalities in the initiation and progression of cancer. Finally, development of cancer can only be understood by integrating the dynamic signaling circuitry between cells and the tumor microenvironment [2].
Hallmark capabilities are regulated by partially redundant signaling pathways; the significance of these pathways depends on the tumor's underlying molecular features. Choosing an appropriate in vitro assay to monitor a given hallmark among the available options is a challenging task, especially for concurrent analysis of multiple hallmarks. Combining assays provides more information, but at the same time multiplexing requires compatibility for parallel or sequential measurements. The cost and speed of an assay, a need for a quantitative real-time or endpoint measurement, and the equipment required should be considered carefully to select an optimal solution. Understanding the pros and cons of each assay is crucial.
In this review we attempt to summarize the most frequently used in vitro assays for analyzing each hallmark in cell culture. We particularly emphasize assay combinations to study multiple hallmarks simultaneously. Animal models, e.g. mouse or zebrafish, are popular to reproduce hallmark features; however, the cost and the required special procedures put animal models beyond the scope of this review. Also, hallmarks involving the immune system were excluded because of the difficulty of assessing them in cell culture-based experiments.

Sustaining proliferative signaling and evading growth suppressors
Unlimited proliferation is the predominant feature of all cancer cells. Sustaining proliferative signaling and evading growth suppressors is defined by the number of healthy cells in a sample and/or by the maintenance of proliferative ability over time. Assays are based on biochemical events specific to living cells, such as metabolic activity and cellular ATP, membrane integrity, DNA synthesis, protein content, and the impedance of living cells. An alternative way to measure cells' ability to undergo sufficient proliferation is to assess colony formation by single cells.

Metabolic activity
Reduction of substrates to a final product by intracellular enzyme activity in living cells can be assessed by a colorimetric assay. The degree of color change is not directly proportional to the number of viable cells, but rather to the activity of an integrated set of enzymes of the cellular environment, which utilize the cofactor, NADH, and various other substrates [3]. This method permits only a moderately robust measure of viability; however, the ease of use and potential for high throughput analysis in multiwell plates has widely popularized it. The most frequently used tetrazolium compound (MTT) is reduced to formazan [4]. More recently-developed tetrazolium analogs (XTT, MTS, WST-1,  are used in conjunction with intermediate electron acceptors [3] and are converted to water-soluble formazan [5][6][7][8]. Eliminating the solubilization step simplifies the protocol and allows real-time measurement of metabolic activity. The deep blue Alamar Blue (resazurin) is reduced to the pink, fluorescent resorufin [9]. Acetoxymethyl ester of calcein (Calcein AM) cleaved by esterases within live cells produces green fluorescence, while dead cells exhibit red fluorescence. Quantification with a fluorescence microplate reader permits determination of the ratio of viable/dead cells in a population. Bioluminescence assays (e.g. CellTiter-Glo) measuring cellular ATP by the development of mutant luciferase perform with better reproducibility and sensitivity compared to MTT assays and are particularly useful when assessing cell viability in low density cell populations [10][11][12][13]. Table 1 summarizes the most frequently used enzymatic assays and their limitations.

Protein content
The presence of living cells is assayed by SRB (sulforhodamine B), a red fluorescent aminoxanthene dye. This compound produces a change detectable via colorimetry as its sulfonic group binds to basic amino acid residues in proteins, to give an estimation of total protein mass, which is directly proportional to cell number [14]. In contrast to MTT, SRB staining is not dependent on mitochondrial function; therefore less variation is observed between cell lines [15]. SRB has better linearity and higher sensitivity than other colorimetric assays, and is comparable to those achieved with standard fluorescent dye staining methods [16,17]. The method is very cheap as it requires only a multiwell reader, has a stable endpoint, and the reagent can be stored indefinitely.

Membrane integritydye exclusion assays
Several methods employ dyes which diffuse passively through damaged membranes, while they are excluded by healthy cells. First, the trypan blue exclusion assay [18] is cheap, simple and still widely used. Viable cells are counted with a hemocytometer or an automated electronic particle counter, and the percentage of viable cells is calculated [19]. Second, healthy and early apoptotic cell membranes are impermeable to the red-fluorescent propidium iodide (PI), which intercalates between base pairs of double-stranded DNA. Uptake is assessed by flow cytometry allowing single cell level analysis [20,21]. Finally, SYTOX Green Stain and YO-PRO-1 can provide more reliable results than PI-based tests in flow cytometric assays [22]. The major drawback of the dye exclusion assays is the cytotoxic nature of most probes, preventing long term use, and inaccurate automated readings if the cells tend to cluster.
Cells with a compromised membrane release substances such as lactate dehydrogenase (LDH) [23]. Fluorescence-based detection methods have been developed to perform LDH measurement directly in culture wells with a spectrophotometer containing a mixture of viable and dead cells without damaging the membranes of the viable cell population. Limitations are the low selectivity and the relatively short halflife of LDH [24].

DNA synthesis
Incorporation of radiolabeled DNA precursor 3 H-thymidine into new strands of chromosomal DNA or more recent analogs, BrdU Table 1 Summary of intracellular metabolic assays available to measure cellular proliferation.
Not suitable for cells with lower metabolic activity due to relatively elevated background levels. The presence of DTT, mercaptoethanol, L-cysteine, L-ascorbic acid, cyanide, azide, and changes in environmental oxygen distorts the results [5]. At higher cell density or at metabolically very active cells hydroresorufin is produced distorting the results. Necessary to test crossreactivity with any compound to be tested, without any cells in the medium [268,269]. Calcein AM Acetoxymethyl ester of calcein Is unaffected by pH change, and has a superior cellular retention [270].

Cellular ATP content
Firefly luciferase enzyme Higher reliability and sensitivity compared to MTT, can detect as few as 10 cells/well. Shorter protocol and less handling compared to other methods. Especially useful in low density populations.
(bromoeoxyuridin) and EdU (5-ethynyl-2 deoxyuridine), can be used to assess de novo DNA synthesis [25][26][27]. During the S phase of the cell cycle they are integrated into the nuclear DNA [28]. 3 H-thymidine detection requires a scintillation beta-counter, whereas BrdU can be detected by antibodies, allowing analysis by flow cytometry or by immunohistochemistry, eliminating the need for radioactive labeling. BrdU is toxic and mutagenic, alters cell cycle and the fate of the cells that incorporate it, thus appropriate controls are required [29,30]. In contrast to BrdU, the newest analog EdU does not require DNA denaturation by exposing cells to HCl, heat or DNAse, and its detection is rapid and highly sensitive [27]. BrdU and EdU are thought to provide the most reliable and direct index of proliferation.

Impedance of a living cell
The traditional methods, usually based on end-point analysis, provide limited information about cell physiology. The non-invasive, label-free real-time cell analyzer platforms (RTCA) are capable of continuously monitoring biological processes, such as proliferation, migration and invasion, based on the impedance of living cells grown on surfaces, thereby permitting conductivity measurements [31]. Platforms such as xCELLigence RTCA are commercially available; gold microelectrode arrays line the bottoms of the wells and create an electric field with the introduced cells acting as insulators, such that net cell adhesion can be measured. Cell Index (CI) is an arbitrary unit reflecting the electronic cell-sensor impedance that directly correlates with the number of viable cells. The RTCA is also suitable for co-culture studies. When testing cytotoxic compounds, the effects of toxic insult are often transient. RTCAs are ideal to determine the optimal time points for end-point measurements and the temporal profile of drug responsiveness [32].

Live-cell time-lapse imaging
Commercially available, the IncuCyte™ system allows real-time, automated monitoring of cell proliferation based on live cell time-lapse imaging. The system monitors proliferation by analyzing confluence and allows the separation of living and apoptotic cells by labeling reagents. Moreover, transfecting the nucleus with green or red fluorescent protein (GFP/RFP) tagged constructs allows the quantification of cell number and the monitoring of migration/invasion. A major advantage is the nearly unlimited real-time analysis, but one must consider the high initial cost of the setup. Performance of real-time systems (xCELLigence, IncuCyte™) was outstanding in sub-confluent cell cultures, but these were outperformed by endpoint assays at full confluence [33].

Colony formation assay
The clonogenic assay tests the effects of external stress signals on cell survival, namely the cell's ability to undergo sufficient proliferation to form a colony. This assay is especially useful to assess long-term effects, such as survival after irradiation. It is suitable for use with any type of cell that can be grown in culture. Sufficient colonies generally contain at least 50 cells. Colony counting is usually done manually with a pen or with automated gel count systems, such as the Oxford Optronix. These systems provide additional information, such as colony size, are able to handle high plate throughput and offer consistent counting. However, the method is also time consuming with extended incubation times, plus colony formation ability of cells differs [34]. Cell architecture and conformation influence cellular responsiveness [35,36].
Two-dimensional assays limit intercellular communication; therefore 3D culture and assay methods are under development [37]. The soft agar assay determines anchorage-independent proliferation, a characteristic of cancer cells, as normal cells undergo anoikis when they detach from the extracellular matrix. Cells are grown in soft agar, with a denser agar beneath to prevent adhesion to the culture plate, where transformed, but not normal, cells form colonies. The assay allows the study of tumor suppressive function of signaling molecules and is often predictive of tumorigenicity in vivo [38]. The traditional assay requires weeks for completion and does not allow cell retrieval; however incorporation of fluorimetric dyes allows quantitative, highthroughput colony counting and the use of specialized agars permits viable cell retrieval (e.g. CytoSelect™ multiwell transformation assays).
Assays detecting living cells and proliferation in cell cultures are summarized in Table 2. Sustained proliferation and evasion of growth suppression can also be estimated by analyzing gene expression of key enzymes involved (Table 6) [39,40].

Resisting cell death
Apoptosis is an active form of cell death, an organized process that leads to programmed self-destruction. Activation of apoptotic pathways alters morphology and cell surface, leads to genome fragmentation and mitochondrial dysfunction and fragments the dying cells in the final phases into apoptotic bodies [41]. It is a relatively swift process, from initiation to completion lasting less than 3 h. The complex signaling pathways are strongly regulated, enabling opportunities to detect and evaluate the process, although the timing of analysis must be optimized as some proteins only appear transiently. Two main apoptotic pathways are recognized. The extrinsic pathway is mediated by cell surface death receptors such as Fas or TNF-R, while the intrinsic pathway is activated by internal damage leading to mitochondrial cytochrome c release. The two pathways are linked, mutually influence each other and ultimately converge on a single execution pathway modulated by caspase activation [42].

Morphological changes
Morphological changes such as cell size reduction, nuclear condensation and fragmentation, the presence of large clear vacuoles, and membrane blebbing can be quantified by light microscopy or transmission electron microscopy. Live cell time-lapse imaging allows realtime, automated monitoring of cell proliferation and morphology by high-definition phase-contrast images [43]. Additionally, each of these cellular modifications can be rapidly measured by flow cytometry [44].

Genome fragmentation
During apoptosis the chromatin is fragmented by endonucleases, the nucleus destructs (karyorrhexis), and cells lose oligo-nucleosomal fragments. DNA laddering is a common end result that can be detected as genome fragments by gel electrophoresis [45]. The loss of genome fragments leads to reduced staining of apoptotic cells with DNAintercalating fluorochromes such as propidium iodide, which is measured easily and accurately by flow cytometry [20].
The TUNEL (terminal deoxynucleotidyl transferase (TdT)-mediated dUTP Nick End Labeling) assay is a common method to detect early apoptosis. It is based on the appearance of 3′-OH DNA termini generated by endonucleases [46]. Terminal transferase (TdT) is used to add fluorescently-labeled dUTP to the free DNA double-and singlestranded breaks [47,48]. Such DNA breaks can also be detected in cell suspensions by flow cytometry or fluorescence microscopy by apoptosis enzyme-linked immunoassay based on immunostaining [49].

Plasma membrane alterations
Phospholipids are asymmetrically distributed on the inner and outer surfaces of the plasma membrane. Apoptotic cells are unable to maintain the phospholipid asymmetry, thus phosphatidyl-serine (PS) appears on the outer surface. PS can be detected using PS-binding anticoagulant Annexin V-FITC. FITC or other fluorescent-labeled apoptotic cells then are visualized with fluorescence microscopy or counted by flow cytometry [50,51]. Using Annexin V-FITC in combination with PI permits viable (double negative), early apoptotic (Annexin V-FITC positive/PI negative) and late apoptotic (PI positive) cells to be distinguished [51].

Assays to monitor mitochondrial membrane alterations
Assays to monitor mitochondrial membrane alterations allow the recognition of changes in the early phase of the intrinsic pathway activation. The collapse of the electrochemical gradient across the mitochondrial outer membrane decreases mitochondrial transmembrane potential (ΔΨ m ), leading to increased membrane permeability, detectable by several sensitive cationic dyes (listed in Table 3) [52,53]. Normally accumulating in the mitochondria, during apoptosis these cationic dyes diffuse into the cytoplasm where they can be detected by altered fluorescent emission [54].

Intracellular calcium concentration
Altered intracellular Ca 2+ concentration is one of the first signs that a cell is committed to apoptosis initiated by the Bcl-2 family (intrinsic pathway; see below). Release of the key apoptotic enzyme, cytochrome c, depends on the emission and uptake of Ca 2+ from endoplasmic reticulum and mitochondria, respectively. The Ca 2+ levels of the cytosol or mitochondria can be measured by selective fluorescent dyes (Table 3) [55,56].

Detection of apoptosis regulators
Cytochrome c can be detected in cell extracts by immunoprecipitation or immunoblotting with cytochrome c-specific monoclonal antibodies (mAbs). Subcellular detection of exogenous GFP-tagged cytochrome c is possible by fluorescence microscopy [57,58]. Cytochrome c is also released extracellularly from cells that are committed to apoptosis. In cell cultures it can be detected in the media in the early phase and is specific to apoptosis, but not to necrosis [59].
Members of the CDK (cyclin-dependent kinase) family play a key role in the regulation of both mitosis and apoptosis. Activation of CDK2 is required for apoptosis and is controlled by p53 and Bcl2-family proteins [63]. The detection of CDK2 along with associated proteins is based on immunoprecipitation followed by immunoblots [64].

Detection of caspases
Caspases, belonging to the cysteine protease family, are irreversibly activated during apoptosis and other cellular processes, such as inflammation or differentiation. An intrinsic or extrinsic death signal activates the proteolytic cascade of caspases promoting their aggregation into dimers or macromolecular complexes, resulting in controlled demolition of cellular components. With specific antibodies, caspases can be identified by immunoprecipitation or immunoblot, may be labeled with fluorescent probes, and detected by fluorescence microscopy [65], flow cytometry [66], or laser scanning cytometry [67]. The IncuCyte™ technology allows the direct addition of the caspase 3/7 apoptosis assay reagent to tissue culture wells, preventing cell loss. The inert, non-fluorescent reagent freely crosses the cell membrane, and after cleavage by caspase 3, releases a green DNA-binding dye. The fluorescent staining of the cleaved substrate can be monitored by live cell imaging [43].
Small labeled peptide-based substrates conjugated to fluorogenic or chromogenic groups are released and detected spectrophotometrically when the substrate is cleaved by a particular caspase [68][69][70]. Synthetic caspase inhibitors (Fluorochrome Inhibitor of Caspases; FLICA) bind to functional groups of active sites of caspases [71].
Detecting cleavage of the multiple caspase substrates inside a cell is a common method to assess caspase activity. One of the most frequently used targets, a DNA repair enzyme, Poly(ADP)ribose polymerase (PARP), is identified by specific antibodies (together with mAb for entire, uncleaved PARP) and detected by immunoblot [72], fluorescence microscopy [73] or flow cytometry [74].
A detailed comparison of apoptosis detection assays is included in Table 3. Hallmark features for resisting cell death and their applicable detection methods are presented in Fig. 1A. The occurrence of apoptosis can also be identified by analyzing the gene expression of key enzymes and regulators (Table 6) [75,76].

Enabling replicative immortality
Normal cells pass through a limited number of cell cycles, followed by an irreversible, nonproliferative state (senescence), and finally death. Telomeres protect the termini of chromosomes and are central in preventing uncontrolled proliferation [77]. The size of the telomeric DNA determines the cell's maximum capacity for division before telomeres are eroded and lose their protective function. In the absence of telomeres, unprotected chromosomal termini undergo fusion resulting in karyotypic disarray (aneuploidy) and cell death. Telomerase, a specific DNA polymerase, adds telomere hexanucleotide repeats at the termini of telomeric DNA. It is nearly absent in normal cells, but its expression is up-regulated to significant levels in almost all types of malignant cells.

Measuring the length of telomeres
For many years, terminal restriction fragment analysis (TRF) was the gold standard quantitative method to measure telomere length; it became a reference for techniques developed thereafter. Restriction enzymes fully digest the genomic DNA but specifically avoid telomere repeats, resulting in short genomic fragments, while simultaneously leaving longer uncut telomeres. Fragments are resolved by agarose gel electrophoresis and telomeres are identified through either Southern blotting or in-gel hybridization using a labeled probe specific for telomeric DNA [78][79][80]. TRF requires substantial amounts of genomic DNA, prepared at least from 10 5 cells, and is relatively insensitive to very short telomeres [81].
Single Telomere Length Analysis (STELA) takes advantage of a single-stranded 3′ G-rich overhang present at the end of all telomeres [82]. Using the overhang as a specific template, an oligonucleotide linker is annealed to the 5′ terminus of the telomere. After amplification with polymerase chain reaction (PCR) the products are Southern blotted and probed with the specific subtelomeric sequence [83,84].
Quantitative fluorescence in situ hybridization (qFISH) of telomere repeats directly employs fluorescent-labeled(CCTAA) n peptide nucleic acid (PNA) probes on metaphase chromosomes. Once the PNAs specifically hybridize to denatured telomere DNA repeats [85,86], the detected signal is compared to standards of known telomere length with specific Q-FISH image analysis software [87]. The method allows the quantification of telomere length in distinct cell populations within a single sample by antibody staining [88,89].

Telomerase activity
Telomerase activity is measured by TRAP (telomere repeat amplification protocol) or real-time quantitative PCR. TRAP is a PCR-based assay based on the low substrate specificity of telomerase. During elongation, telomerase from the cell extract adds telomeric repeats to the telomere-imitating oligonucleotide, after which telomerase- synthesized DNA is amplified by using specific primers (telomere-imitating and reverse) [90][91][92]. Different labels are incorporated into the telomerase-synthesized DNA [93]. In the final step the telomerase product is visualized. The TRAP assay can be performed in both semiquantitative and quantitative modes [94]. Multiplexing can be achieved by the microarray TRAP (MTRAP) [95]. In situ TRAP (FISH-TRAP) using fluorescent telomerase primers and in situ PCR can detect telomerase activity in clinical specimens [96].

Senescence
Repeated cell division eventually leads to replicative senescence. Excessive extra-or intracellular stress, such as telomeric dysfunction, mitochondrial retrogression, oxidative stress, DNA damage, or oncogene activation also provoke stable cell-cycle arrest to prevent the spread of damage to the next generation [97,98]. The first identified marker was the lysosomal senescence-associated β-galactosidase (SA-βgal) that cleaves galactosidase from glycoproteins and is only functional at pH 6.0 in senescent cells [99]. By cleaving the chromogenic substrate X-gal at pH 6.0, senescent cells develop a blue stain detected by bright field or phase contrast microscopy. Using substrates that become fluorescent after cleavage, SA-βgal positive cells can be quantified by fluorescence microscopy, by fluorescence-activated cell sorting (FACS) for living cells, or by high throughput screening (HTS) on cell lysates in multiwell plates [100]. Senescent cells are metabolically active and secrete a complex mixture of factors referred to as the SASP (senescence-associated secretory phenotype) [101]. The secretome, regulated by transcription factors such as CEBPb and NF-kB, includes extracellular proteases, matrix components, growth factors, pro-inflammatory cytokines, and chemokines. The secretome can be evaluated by antibody arrays [102]. In some cases, cellular senescence is a potent tumor suppressive mechanism that stops the growth of cells at risk for malignant transformation [103,104]. However, some cytokines and growth factors are pro-tumoral and pro-invasiverather than tumor suppressiveand act by changing the tissue microenvironment and promoting migration and invasion [101].
The most frequently used methods and their impacts on detecting hallmark features are illustrated in Fig. 1B. Cancer cell immortality can be assessed by analyzing expression of key enzymes and regulators as listed in Table 6 [101,102,105-107].

Inducing angiogenesis
Neoplastic cells induce angiogenesis to secure the tumor's requirements for nutrients and oxygen. A trait first attributed solely to rapidly growing macroscopic tumors, angiogenesis has later been identified in microscopic pre-malignant lesions as well, in morphologically atypical tissues with an abnormally high occurrence of cancer. A prototypical angiogenesis inducer is vascular endothelial growth factor-A (VEGF-A), that coordinates vessel growth, homeostatic survival of endothelial cells and wound healing. Up-regulation of VEGF and fibroblast growth factor (FGF) expression in tumors as compared to normal tissue is attributed to oncogenes like KRAS and hypoxia [2,108,109].
Two approaches have been widely used to simulate the complex process of angiogenesis [110,111]. First, the response to exogenous stimulatory or inhibitory agents, proliferation, differentiation, and migration of endothelial cells can be studied in cultured HUVEC (human umbilical vein endothelial cells) and HuDMEC (human dermal microvascular endothelial cells) cell lines. The proliferative ability of these established cell cultures offers various methods to quantify angiogenesis via measuring endothelial proliferation [112].
Endothelial cells respond to angiogenic factors with increased motility, and move along the gradient of angiogenesis-inducing factors (e.g. VEGF), a process called chemotaxis. Migration is assessed with a modified Boyden-chamber assay. Endothelial cells are plated on top of a filter with a defined pore size and migrate through as a response to an attractant in a lower chamber. The assay is highly sensitive to small differences in concentration gradients [113][114][115][116][117][118].
Endothelial cells, including progenitor and immortalized cells, rapidly form capillary-like structures in vitro when plated on top of a basement membrane extracellular matrix [119]. The presence of fibrin, collagen or Matrigel stimulates the attachment, migration and differentiation into capillary-like tubules in a manner that mirrors the in vivo situation, and, to some extent, tissue architecture. Tube formation is quantified from several microscopic fields randomly selected in a given experiment, by measuring the total number of nodes, connected and unconnected tubes as well as their lengths by digital images produced by CCD cameras. These images are processed with a specific digital image-analysis system for automated and quantitative evaluation [120][121][122][123].
The second approach is based on direct detection of known factors effecting tumor cell angiogenesis [124]. It employs monoclonal/polyclonal antibodies for known angiogenesis activators (e.g. members of the VEGF, FGF, EGF family) by immunoblot or ELISA [125][126][127].
The key enzymes and regulators that allow the study of gene expression linked to angiogenesis are summarized in Table 6 [128-132].

Activating invasion and metastasis
The spread of neoplastic disease has been described as a sequential multi-step process, termed the invasion-metastatic cascade. The program by which epithelial cells acquire migratory and invasive capacity is called epithelial-mesenchymal transition (EMT) [133]. The transformation includes the loss of cell adherence proteins, alteration of cell morphology, increased motility, expression of matrix-degrading proteinases and resistance to apoptosis, representing a fundamental reprogramming of every aspect of the cell's biology. EMT, which is central to normal developmental processes such as gastrulation [134], is hijacked by cancer cells for their own dissemination. EMT can be induced by intrinsic factors, such as genetic mutations, or by extrinsic signals, such as growth factors. However, some tumors exhibit incomplete or partial EMT [135], or invade without epithelial-mesenchymal conversion [136]. The tumor microenvironment acquires increasing importance, as the crosstalk between the tumor and the surrounding stroma is capable of inducing metastasis [137,138]. Understanding the complex nature of interactions between cancer cells and the extracellular matrix (ECM) requires a systems-based approach instead of focusing merely on genetic mutations within neoplastic cells.
Migration and invasion are distinct phenomena in experimental cell biology. Migration is the directed movement of cells without passing through obstructive barriers, whereas invasion necessitates the destruction of barriers in order to pass through them and thus is of necessity accompanied by ECM remodeling [139]. However, during both migration and invasion, cells squeeze through tight interstitial spaces, and the nuclear deformation caused by the confining microenvironment requires subsequent restoration of the nuclear envelope and DNA content [140].

Transwell migration and invasion assay (Boyden chamber)
A double chamber is filled with two media, one with an attractant (like FBS) to trigger chemotaxis. Cells are seeded in the upper well, and migrate vertically between the chambers through a porous membrane [141,142]. Migrated cells can be visualized by cytological dyes, stained fluorescent, or lysed and assessed by a plate reader. Due to different cell dimensions, the assay must be optimized for each cell type [139]. By coating the porous filter with ECM components like type I collagen or a basement membrane-like matrix (Matrigel), the device is made suitable for testing invasion. Invasive cells degrade the matrix and move through to the bottom. Parallel measurements with ECMcoated and non-coated assays allow one to calculate an "invasive index": the rate of invasiveness versus migration [143]. Although providing only endpoint measurements, the transwell invasion assay is the most frequently used method due to its easy setup.

Scratch (wound healing) assay
This simple and popular assay allows the observation of twodimensional (2D) cell migration in confluent, monolayer cell cultures. The surface of a cellular monolayer is wounded by scraping, and images are captured by microscopy at regular intervals to track the migration of cells from the edge of the wound until the scratch is closed [116,144,145]. Time-lapse microscopy with image analysis software enables automation. Mechanical scraping is done usually with pipette tips, needles, razors, etc., although more advanced methods produce more controllable wound sizes with clearly defined edges, without damaging the underlying matrix. Electric cell-substrate impedance sensing (ECIS) systems enable automatic monitoring of cell adherence and wound closure while ablating cells with a high electric current [146,147]. Laser ablation is another method to create a well-defined scratch [148,149], however both methods generate leftover cells and debris in the wound area [150]. Dying cells release apoptotic factors, potentially effecting wound closure. The advantage of the scratch assay is its easy setup and low cost, but the experiments usually require relatively large cell and reagent quantities. Scratch assays are appropriate to study regeneration of epithelial or endothelial cells [116], and can be combined with other techniques, such as gene transfections [151].

Cell exclusion zone assays
Cell-free zones are created before seeding the cells, using a physical barrier such as silicone. Later, the barrier is removed, allowing cells to occupy the area [152]. Alternatively, cells are seeded inside of a ring, and migrate outward once the barrier is removed [153]. Barriers can be made of glass, silicone, metal, Teflon, microfabricated soft and elastic "stencils" or agarose gels [139,150,153,154]. Cells are visualized by photomicrography or labeled with fluorescence and measured with a microplate reader [150,152]. Migratory capacity and interaction between two different populations can be compared [139]. These assays avoid the drawbacks of scratch assays, such as uneven and irreproducible scratch areas and remnants of dying cells in the media.

Microcarrier bead and spheroid migration assays
Cells migrate from the surface of confluent cell-coated microcarrier beads to a 2D surface in a cell culture vessel, where cells can be fixed, stained and quantified. The close contact of cells on the bead surface mimics in vivo cell-to-cell contact better than flat monolayers [155]. The original version of the assay measures end-points, but cells can also be traced during migration by light microscopy [139]. Beads are expensive and some beads might not be covered sufficiently. In the spheroid migration assay, tumor cell spheroids are placed on a surface of cultured vessels. Tumor cell spheroids resemble in vivo physiologic conditions by establishing oxygen and nutrient gradients characteristic of tumors, but the method can only be performed on cells that are able to form spheroids [156]. The spheroid invasion assays enable the study of invasive properties of multicellular spheroids embedded into threedimensional (3D) extracellular matrices [157]. Different scenarios including malignant and non-malignant spheroids and invasion of epithelial cells can be investigated [158,159]. In the spheroid confrontation assay two cell types are co-cultivated, allowing them to invade each other in a single cell pattern or collectively as a population [160]. Fluorescent labeling enables an invasion to be followed with microscopy, and flow cytometry allows quantification of invasive cells.

The capillary chamber migration assay
The capillary chamber migration assay consists of two chambers, one containing the cell suspension and the other a chemoattractant, connected by a linking capillary. Migratory behavior of cells is observable, and morphological responses are visualized by time-lapse microscopy in real time [161]. The small cell numbers and reagent volumes make this assay suitable for rare cell types and expensive compounds. Liquid handling and image processing can be fully automated [162].

Vertical assays
Cells are placed on top of an ECM, where after forming a monolayer they move down individually or collectively, as first described with lymphocytes [163]. In skin cancer, cells can also be coated with matrix and move up [164,165]. Alternatively, cell monolayers can be sandwiched between a top and bottom layer of ECM, and additional cell types might be embedded into the ECM to study their effects on cell invasion [166].

Motility of individual cells
Motility of individual cells can be traced by time-lapse video microscopy. New algorithms are available to follow up to a dozen cells simultaneously [167]. 3D tracking is also possible, but requires specialized microscopes and image analyzing software, as well as extensive data processing [168]. The gelatin degradation assay allows analysis of progressive invasion at a subcellular level. The ability to form invadopodia directly correlates with the invasive potential of tumor cells. Cells are seeded on top of a thin, fluorescent dye-labeled gelatin matrix. The actin-rich invadopodia and podosomes extend into the ECM and degrade the gelatin with matrix metalloproteinases, creating spots with decreased fluorescence [169]. Computerized methods allow the analysis of invasive properties of single cells as well as of populations [170].

Mechanical properties
Mechanical properties of individual tumor cells affect the way cells are able to spread. Metastatic cancerous cells are 70% less stiff than normal cells [171], and stiffness inversely correlates with migration and invasion through 3D structures, measured by a magnetic tweezer system combined with a 3D Matrigel invasion assay [172]. A recently developed high throughput microscope system is able to assess cellular mechanical properties in multiwell plates [173].

Real-time monitoring and live-cell imaging
The impedance based xCELLigence System (see also Section 2.1.5) permits label-free real-time monitoring of invasion and migration. Membranes can be coated with ECM to study invasive properties [174].
The IncuCyte™ Chemotaxis System, based on live-cell time-lapse imaging, allows automated visualization of labeled or label-free chemotactic cell migration and invasion even in multiwell format [175,176].
The most frequently used in vitro migration and invasion assays are summarized in Table 4, and the genes linked to migration and invasion are summarized in Table 6 [157,167,177-180].

Genome instability and mutation
Most cancers exhibit chromosomal instability, meaning altered chromosomal structures and high rates of rearrangements compared to normal cells. Microsatellite instability is characterized by the expansion or reduction of short nucleotide repeats present in microsatellite sequences. Nucleotide instability refers to the increased frequency of substitutions, deletions and insertions in one or several nucleotides [181]. The origin of genomic instability is still a subject of intense research [182]. Oncogene activation can induce the collapse of DNA replication forks resulting in double-strand breaks. DNA damage activates TP53, leading to senescence or apoptosis. In contrast, inactivating (loss-of-function) mutations in TP53 eliminate the barrier against tumor progression [183]. In hereditary cancers, loss of a DNA repair gene is considered to drive cancer development by increasing the spontaneous mutation rate [182]. Neoplastic cells incur mutations through increased sensitivity to mutagens, by weakening the genomic maintenance machinery or by eliminating the surveillance system guarding genomic integrity.

Flow cytometry
Flow cytometry is a universal multicellular approach that can detect cellular ploidy and cell cycle distribution with fluorescent dyes that bind to DNA in stoichiometric fashion. Genome instability can be evaluated by comparing ploidy in the G0/G1 fraction for tumor versus normal cells. However, flow cytometry does not provide information about other levels of genome instability [184].

Karyotyping
Karyotyping is a more informative method. A DNA-binding dye intercalates into specific DNA regions of chromosomes (e.g. A/T-rich areas) in metaphase spreads, resulting in a specific banding pattern for each chromosome. Spectral karyotyping is a multicolored whole chromosome-painting assay using FISH probes to visualize each chromosome. Decoding the chromosomes with different fluorochromes allows specification of rearrangements by confocal microscopy. Spectral karyotyping is limited to detecting global changes in chromosomes and as such is unable to evaluate alterations at the sequence level [185,186].

Array of comparative genomic hybridization
Array of comparative genomic hybridization (aCGH) is a multicellular technique used to quantitatively detect and visualize chromosomal alterations. For the detection of gains, losses, amplifications and lossof-heterozygosity, the sample DNA is compared to a normal reference genome. The differentially-labeled reference and sample DNAs are combined, and non-analogous hybridizations are visualized as changes in fluorescence intensity on metaphases. CGH detects unbalanced chromosomal abnormalities affecting copy number, but is unable to distinguish reciprocal translocations, inversions or somatic mutations [187,188]. Single nucleotide polymorphism (SNP) arrays are also hybridization-based. Fragmented nucleic acid sequences labeled with fluorescent dyes hybridize to immobilized, allele-specific oligonucleotide probes. SNP arrays are capable of detecting copy-neutral loss of heterozygosity (so-called uniparental disomy or gene conversion) and distinguish alleles at specific polymorphic sites [189].
To detect microsatellite instability (MSI), the gold standard approach amplifies known microsatellite regions by PCR and quantifies the lengths of PCR products by gel or capillary electrophoresis using autoradiography, silver staining [190] or fluorescent methods [191,192]. The microsatellite length of the sample is compared to normal DNA to determine MSI status. Alternatively, real-time quantitative PCR (qPCR) melting point analysis is used with sequence-specific fluorescentlabeled hybridization probes [193]. MSI status can also be detected with next generation DNA sequencing (NGS) [194]. Microsatellite instability is the results of defects in mismatch repair (MMR) genes such as MSH, MLH and PMS: the expression profile of MMR genes correlates with microsatellite instability [193]. Mutant MMR genes are translated to proteins, and mutant MMR proteins can be studied by immunochemistry using mutation-specific monoclonal antibodies [195].

Whole genome sequencing
Whole genome sequencing is the most comprehensive and informative method to distinguish nucleotide mutations in coding, non-coding and un-annotated regions. It is also able to measure larger genomic rearrangements frequently present in cancer such as copy number variations (CNV), insertions and translocations [196]. Sanger sequencing, developed in 1977, is based on a dideoxy chain-termination method and provides high-quality, long read length, spanning 400-900 base pairs. However it is costly and is unable to provide high throughput. Next-generation sequencing (NGS) systems enable high throughput and accurate analysis along with decreased sequencing cost. The Illumina GS/HiSeq platform implements sequencing by synthesis technology providing high throughput and low reagent cost; nevertheless it is hampered by short read lengths. The Ion Torrent adopts semiconductor sequencing and provides rapid and low-cost sequencing results [197]. Sequencing the transcriptome (RNA-seq) also allows the discovery of special genetic aberrations, such as gene fusions; it could replace microarrays in the long run [198]. Developments in NGS and wholegenome amplification already allow single-cell sequencing [199,200]. Adequate genome coverage is critical, and each area of genome research requires markedly different sequencing depth [201]. Common methods to assess genomic instability and their association with hallmark features are illustrated in Fig. 1C. Genes participating in genome maintenance and related to cancer genome instability are listed in Table 6 [202,203].

Reprogramming energy metabolism
Actively proliferating cancer cells must regulate their energy metabolism to fuel constant cell growth. By simultaneously increasing the rate of glycolysis while limiting oxidative phosphorylation, cells enter a state termed aerobic glycolysis, also known as the Warburg effect [204,205]. Increased glycolysis can allow the diversion of glycolytic intermediates into biosynthetic pathways required for new cell assembly. Altered energy metabolism is so widespread in cancer cells that it has been included as one of the hallmarks of cancer [206].

Glucose uptake
Glucose is rapidly metabolized, so virtually non-metabolizable glucose analogs like FDG (fluoro-deoxyglucose), 2DG (2-deoxy-D-glucose) and 3MG (3-O-methylglucose) are needed to study uptake efficiency. In cell cultures, the most commonly used analogs are radioactive hexoses, quantified from cell lysates by a liquid scintillation counter [207][208][209]. The uptake of nonradioactive, fluorescent glucose analogs is measured by flow cytometry or fluorescence microscopy [210]. Enzymatic assays of glucose uptake are based on a dye reduction reaction connected to the first step of glycolysis and are detected spectrophotometrically [211,212].

Activity of rate-limiting glycolytic enzymes
Hexokinase, phosphofructokinase, and pyruvate-kinase are the main rate-limiting enzymes of glycolysis, controlling the maximum possible carbon flux. Enzyme activity is assessed by coupling them with a NAD + /NADH-dependent enzyme reaction producing visible substrates and measuring changes in absorbance values [213]. Determination of free nuclear NADH can be assessed by two-photon microscopy to quantify the intensity of NAD(P)H autofluorescence.

Metabolomics and inorganic products
Metabolites are quantified by mass spectroscopy (MS) or nuclear magnetic resonance spectrometry (NMR). By using 13 C-or 3 H-labeled glucose, the amount of 13 C-or 3 H-glucose derived isotopes in downstream metabolites is linked to glycolytic activity. 13 C can replace any of the six carbons in the glucose molecule [214]. 3 H is incorporated into H 2 O from a 3 H-labeled glucose [215]. 3 H 2 O is released into the medium by the conversion catalyzed by enolase and its subsequent presence in the media of cultured cells can be measured by liquid scintillation counting [216].

Oxygen consumption
The rate of oxygen consumption (OCR) is a universal indicator of mitochondrial respiration, and can be measured both intra-and extracellularly. Intracellular O 2 -levels provide accurate information about the metabolic activity of a cell. The measurement is based on a cell-penetrating O 2 -sensitive phosphorescent probe [217]. Currently, supramolecular and nanoparticle-based intracellular probes are available with rapid intracellular accumulation allowing direct, contact-free and quantitative oxygen assessment. The assays are typically performed in microplates and measured by real-time fluorescence plate readers [218].
Extracellular oxygen consumption can be measured by polarographic respirometry (Clark's electrode) or with optical sensor probes. A Clark's electrode allows measurement of intact cells or extracted mitochondria in suspension by polarography [219]. Optical sensor probes determine oxygen consumption from the medium over a monolayer of living cells within a microplate reader. The Extracellular Flux Analyzer measures multiple metabolic parameters, such as oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) in real time. Metabolism can also be modified by successive addition of different compounds [220]. During mitochondrial stress, the sequential addition of metabolic modulators to different parts of the electron transport chain ( Fig. 2A) allows the identification of key mitochondrial parameters such as basal respiration, ATP turnover, proton leak and maximal respiration. The difference between maximal and basal respiration represents the respiratory reserve, the capacity of a cell to create ATP via oxidative phosphorylation in response to increased energy demand. Non-mitochondrial respiration results in substrate oxidation and cell surface oxygen consumption (Fig. 2B).

Extracellular acidification
Intensified glycolysis with higher glucose uptake results in excess lactate secretion into the extracellular environment. Commercially available kits are available to quantify lactate levels in cell cultures and are based on colorimetric or fluorometric reactions compatible with standard laboratory equipment, e.g. spectrophotometers [221,222]. Glycolytic activity is also indicated by conversion of pyruvate to lactic acid, determining the extent of extracellular acidification rate (ECAR) as measured by an extracellular flux analyzer [223]. This test is performed by adding glucose, followed by inhibitors of glycolysis and oxidative phosphorylation, and provides information on glycolytic capacity and reserve. Glycolytic reserve indicates the cell's ability to increase glycolytic rate upon increased energy demand (Fig. 2C).

Glutathione (GSH)
Glutathione (GSH) is a tissue antioxidant preventing cell damage from free radicals and peroxidases. Increased levels of oxidative stress generate oxidized glutathione (GSSG), causing an increased ratio of GSSG to GSH. Increased susceptibility to oxidative stress has been implicated in cancer development [224]. Alterations in the GSH molecular pathways in cancer cells can lead to enhanced chemoresistance and increased survival. Tumor glutathione levels tend to be altered, i.e. frequently increased compared to healthy tissue [225]. GSH and GSSG levels can be assessed rapidly and reliably spectrophotometrically in a microplate reader [226].
Assays of energy metabolism are summarized in Table 5, and their connection with hallmark features are illustrated in Fig. 1D. Expression of glucose transporters or other metabolic enzymes is rendered feasible by immunoblotting, FACS or qPCR. The bioenergetics and metabolism of a cell can also be estimated by analyzing gene expression of key enzymes (Table 6) [227,228].

Epigenetics
In recent years, epigenetics has become equally important compared to genetic mechanisms in cancer initiation and progression [229,230]. Promoter hypermethylation leads to transcriptional silencing. CpG methylation represents a key epigenetic mechanism of inactivation of tumor suppressors and MMR genes [231]. Methylation patterns can be detected by methylation-specific PCR (MSP). Treating the DNA with sodium bisulfite converts unmethylated cytosine to uracil, and during a subsequent PCR reaction, primers anneal to either the methylated or the unmethylated sequence [232]. MSP is not quantitative and bears a significant risk of false positive and false negative results. In contrast, Quantitative Methylation-Specific PCR (qMSP) assays utilizing TaqMan probes are highly sensitive to promoter hypermethylation and available in high throughput format (e.g. MethyLight) [233]. Quantitative multiplex methylation-specific PCR (QM-MSP) accurately determines promoter hypermethylation for many genes simultaneously in small samples, is highly sensitive and linear over 5 orders of magnitude [234]. Pyrosequencing is a "sequencing by synthesis" technique, during which nucleotide incorporation releases a pyrophosphate molecule, generating a luciferase-catalyzed luminometric signal. Pyrosequencing can be used for quantification of DNA methylation at specific CpG sites within the target region [235]. The method provides high quality, quantitative results and is appropriate to analyze methylation of CpG sites in close proximity [236]. DNA methylation enrichment methods avoid the limitations of bisulfite-conversion. Methyl-CpG binding domain-based capture (MBDCap) uses beads coated in MBD polypeptide that bind methylated DNA; the eluted DNA is sequenced with NGS [237,238]. In addition, epigenetic studies employ chromatin immunoprecipitation (ChIP) to identify DNA sequences that are associated with a chromatin component of interest. Coupled with high-throughput methods, such as microarrays (ChIP-chip) or massively parallel sequencing (ChIPseq) it is possible to identify binding sites of specific proteins across the entire genome [239,240].
Small (17-24 base), highly conserved, non-coding microRNAs (miRNA) regulate gene expression at the post-translational level by targeting complementary mRNAs and play a regulatory role in all cellular pathways [241]. About half of all miRNA genes are located in close proximity to CpG islands and thus are subject to epigenetic regulation via DNA methylation and chromatin remodeling. Additionally, miRNAs modulate the expression of genes linked to the epigenetic machinery [242]. Disregulated miRNA expression has been connected to cancer initiation and progression [243]. Microarray hybridization, quantitative reverse transcription PCR (RT-qPCR) and small RNA sequencing with NGS are the most common platforms to evaluate miRNA expression, although the small size of active, mature miRNAs renders expression profiling challenging. Comparing reproducibility, specificity, sensitivity and accuracy across 12 commercially available platforms showed qPCR in general to be especially suitable for low input RNA samples  Spectrophotometric method [223] The pH in cell media is measured by solid state fluorophore probes. Oxidative stress Glutathione Measuring GSH/GSSG ratio Spectrophotometric method [226] Enzymatic recycling of GSH from GSSG by glutathione reductase. GSH reacts with DTNB producing TNB chromophore.

Outlook
Selecting the ideal analytic technique is based on the investigator's goals, available equipment and also on financial constraints. Multiplexing strategies enable a more in-depth data collection from a single experiment, yielding several values while controlling for variability, saving time and relative cost and leading to more robust conclusions.
High throughput screening (HTS) allows rapid compound analysis in low volume assay formats, such as in 384-or 1536-well plates and is gaining increasing support with the development of automated optical detection systems [254]. High content screening (HCS) is based on automated microscopy and image analysis and is able to measure multiple cell physiology characteristics in real time. HCS allows screening of enormously large libraries of molecules to discover new inhibitors of cellular physiology, thereby enhancing the drug discovery process.  ), TIMP2  Filopodia  BAIAP2, CDC42, DIAPH1, EGFR, ENAH, EZR, MSN, RDX, SVIL, VASP,  Lamellopodia  CTTN, DPP4, EGFR, ENAH, FAP, PIK3CA (p110a), PLD1, PTK2, PXN, RDX, SVIL, VASP, VCL, WASF1, WASF2, WASL  Invasive projections  ACTR2, ACTR3, ARF6, CDC42, CFL1, CTTN, DPP4, EGF, EZR,  HCS provides benefits over, or complements, existing methods to assess proliferation [255][256][257], migration [258,259] and metastases [260,261]. Cytometers based on automated fluorescence microscopy are sensitive enough to detect single cells based on the strength of a fluorescent signal [256]. Proliferation can be assessed by counting DAPI-labeled nuclei and BrdU-incorporating cells separately, then evaluating the proportion of cells in the S-phase based on the dual emission-images [258,262]. Viability and proliferation assays are relevant to evaluate additional hallmarks that involve the analysis of similar physiological processes within living cells. For example, the loss of membrane integrity is a signal of cell death, measured by dye exclusion or LDH release after a cytotoxic insult. Metabolic assays provide information not only about cell viability but also about the extent of overall metabolic change. Dyes incorporated into DNA, such as PI or BrdU, also help to gain insight into DNA content changes when evaluating genome instability.
Each and every hallmark possesses properties that can be analyzed by immunoblot, RT-PCR, immunochemistry, immunoprecipitation, RNA microarray or RNA-seq (see Table 6). In general, flow cytometry, fluorescence microscopy, and multiwell readers are also extremely versatile tools that with proper sample preparation, allow the detection of a vast number of hallmark features (summarized in Table 7).
In summary, here we have provided a link between available techniques and the recognized hallmarks of cancer. Although our list is not exhaustive, our aim is to include the most widely used methods. We encourage the reader to make an informed decision regarding assay selection to gain the maximum benefit from existing tools.

Transparency document
The Transparency document associated with this article can be found, in online version.