Atomic Force Microscopy Probing of Cancer Cells and Tumor Microenvironment Components

Cancer cells have different characteristics from normal cells in terms of morphology, cell–cell interactions, cytoskeleton organization, cell growth rates, and cell-extracellular matrix interactions. Although the clarification of these characteristics is crucial for the effective development of new therapeutic strategies, they remain poorly identified. Furthermore, tumor microenvironment plays a crucial role in cancer progression and metastasis, and thus, it is of utmost importance to decode and fully understand the crosstalk and the interactions between its different components. Atomic force microscopy (AFM) is an established and multifunctional tool in biomedical sciences with many applications in cancer research. In this review, we discuss the use of AFM in the study of cancer cells and the tumor microenvironment. We first provide a brief overview of AFM operating principles highlighting its contribution to the study of cancer and stromal cells. Next, we present research on the use of AFM in the study of cell-tumor microenvironment interactions during cancer cell migration and invasion. Finally, we discuss the development of coupled AFM modalities and the combination of different kinds of imaging with AFM that have been applied to cancer research.


Introduction
Cancer cells differ from normal cells and differences may include alterations in cell morphology, reproduction, communication, adhesion, cell-to-cell or cell-to-extracellular matrix (ECM) interactions, cell invasion/metastasis, and even cell death. Furthermore, recently, it has been recognized that alterations in the mechanical properties of cells play a critical role during malignant transformation and in cancer progression [1]. The mechanical properties of the cells, such as its stiffness, play an important role in cell growth, division, motility, metastasis, and epithelial-mesenchymal transition in the initiation of cancer [2,3]. Although the different characteristics and properties of cancer cells are crucial for the effective development of new therapeutic strategies, they remain poorly identified or understood.
Apart from cancer cells, alterations in tumor microenvironment have been identified as key players in cancer progression [4]. The tumor microenvironment consists of the tumor blood and lymphatic vessels, the stromal cells, the ECM, and a myriad of soluble factors in the extracellular milieu whose importance in cancer progression and metastasis is indisputable. Stromal cells include fibroblasts and myofibroblasts or cancer associated fibroblasts (CAFs) [5]. The most common molecules of the tumor ECM are several types of collagen (e.g., collagen type I) and hyaluronic acid. Evidently, the interactions between constituents of the tumor microenvironment and cancer cells determine cancer cells' properties, such as cell proliferation, apoptosis, migration, and cell invasion through surrounding tissues. Since tumors tend to alter their microenvironment, and at the same time, the microenvironment is known to affect tumor growth and progression, a better understanding of the crosstalk and the interactions between the different components of tumor microenvironment is of utmost importance, in order to develop novel, and more effective anti-cancer therapies [6]. In many types of cancer, including pancreatic, colon, and breast cancers [7], the complex interplay between tumor microenvironment components leads to remodeling and accumulation of ECM fibers, resulting in a desmoplastic reaction. Desmoplasia is a cancer-specific type of fibrosis, characterized by the presence of myofibroblasts and accumulation of ECM components, such as collagen type I [8]. As a result, desmoplasia stiffens the tumor tissue increasing the compressive mechanical forces in the tumor interior [9][10][11]. Mechanical compression of the tumor microenvironment promotes tumor progression both directly by being applied to cancer cells and indirectly by compressing intratumoral blood and lymphatic vessels [12,13]. Direct compression of cancer cells can reduce the proliferation rate of cancer cells and induce cancer cell apoptosis [14][15][16], but it also increases their invasive and metastatic potential [17,18]. Compression of tumor vessels causes hypo-perfusion and hypoxia, which in turn promote tumor progression through multiple ways (e.g., enhance angiogenesis, inflammation, immunosuppression, induction of cancer "stem cell" phenotype, metastasis, resistance to apoptosis) and inhibit the delivery of drugs as well as the efficacy of radiotherapy and immunotherapy [19][20][21].
In order to enhance our understanding of cancer cell properties such as growth, invasion, and metastasis and tumor microenvironment alterations, such as those during desmoplasia, the use of atomic force microscopy (AFM) can been employed. AFM belongs to the scanning probe microscopy (SPM) family and was developed in order to extend scanning tunneling microscopy (the first member of the SPM family) to investigate electrically non-conductive materials at extremely high (nanometer) resolution [22]. AFM operates by measuring forces between a probe (normally a sharp tip) and the sample by scanning a small area of the sample surface. AFM is an established technique that can operate under different conditions [23] without special treatment or vacuum conditions of the specimen [24]. AFM permits quantitative, high-resolution, non-destructive imaging of surfaces, including biological ones [25,26]. In addition, it operates in many different modes, offering a vast amount of information about bio-samples [27][28][29] ranging from topography to mechanical properties characterization [30][31][32][33]. AFM has been used in the field of cancer research, mainly due to its ability to work on live cells with high resolution, and its potential to provide quantitative information (e.g., Young's modulus or elastic modulus, which is a measure of stiffness of an elastic material). It has also been used in cancer cell imaging providing information concerning specific cancer cell markers or properties (e.g., expression of cancer cell surface receptors) [34,35]. Therefore, it offers new possibilities for characterization of cancer-related tissues and biopsies [36].
In this review, we first introduce the AFM apparatus and working principle and then we focus on its applications in the study of cancer and stromal cells. We next discuss current literature on the use of AFM in studies of cell-tumor microenvironment interactions during cancer cell migration and invasion. Finally, we present how the development of AFM coupled modalities and the combination of AFM images with other imaging modalities (e.g., optical microscopy, second harmonic generation, and scanning electron microscopy imaging) can be used in the area of cancer research.

Atomic Force Microscopy: Basic Principles
The molecular or submolecular resolution of AFM is achieved due to the combination of the sharp tip, the careful control of the tip-surface forces, the optical lever, and the high-precision movement of the AFM scanner (Fig. 1a). Piezoelectric elements are used for the accurate movement of the tip over the sample surface, whereas the optical lever is used to amplify the distance of a movement. More specifically, AFM operates by scanning the sharp tip, which is mounted on a cantilever, over the sample surface. Due to the tip-surface forces, the cantilever bends vertically and a laser beam that is focused on the back of the cantilever is deflected and detected by a photodetector. As a result, the movement of the tip is monitored by the alterations of the laser spot position. The tip deflection is then translated into a three-dimensional image of the surface [22]. AFM can operate in several modes, three of which include the following: (a) the contact mode, (b) the intermittent contact (also known as tapping) mode, and (c) the non-contact mode. In the contact mode, the tip is dragged across the sample surface. During contact mode, AFM can also acquire lateral force microscopy (LFM) images. LFM images provide information concerning surface friction that arise from inhomogeneities of the surface of the material and are enhanced by edge deflection (slope variations) of surface features. In the non-contact mode, the AFM tip does not come in contact with the sample surface and AFM does not suffer from tip degradation effects. The most widely used method in biological samples is the tapping mode, which is a 'gentle' imaging mode particularly useful for imaging soft and fragile samples. In this mode, the lateral forces in the sample are circumvented and the frictional forces are minimized [25,31]. In the tapping mode, topography and phase images are simultaneously acquired so as to obtain information on different properties of the sample [24]. AFM phase imaging goes beyond topographical features and it is the appropriate method for detecting variations in composition, friction, and viscoelasticity enhancing contrast on heterogeneous samples (see Fig. 1b) [37,38].
Apart from the above-mentioned modes, AFM can also operate in force spectroscopy mode, which measures individual molecular interactions, the nanoindentation mode employed to measure hardness or softness of a sample, as well as the force scanning mode, used for the calculation of the Young's modulus at every point of an image, allowing the generation of a Young's modulus map [39].
Due to the previously mentioned unique capabilities of AFM, it is increasingly being used for nanomechanical characterization of cells [40][41][42]. Specifically, nanoindentation is used in testing mechanical properties of materials, including biomaterials and biological samples [40,43,44]. The AFM nanoindentation method is based on recording the elastic response of a material using an indenter (i.e., the AFM tip) [44] and enables the determination of the Young's modulus of the sample of interest in different nanoregions. More specifically, AFM generates indentation-force curves which can be modeled in order to calculate the Young's modulus of the sample. These curves can be fitted using mathematical expressions, such as the Hertz model, which considers the sample to be an isotropic and linear elastic solid and describes the elastic deformation of two surfaces in contact under load [24]. The possibility of real-time indentation and measurement of localized mechanical properties on the material's surface has rendered nanoindentation, one of the most common techniques for the characterization of the mechanical response of biological samples including proteins, cells, viruses, and bacteria [42,[45][46][47]. Furthermore, research in the development of new AFM modes is ongoing and new techniques emerge such as the multiharmonic mode, which develops quantitative local property maps by using changes in the amplitude, the phase of the oscillator, and other relevant harmonics [48,49].
Hence, AFM can be used for quantitative multiparametric characterization of complex cellular and biomolecular systems [50] and can be applied for studying basic structural, biophysical, and mechanical properties of cells in health and disease. Moreover, AFM offers a number of additional advantages that make it unique for biological applications: (a) the characterization of samples by AFM does not require dehydration, sample labeling with fluorescent dyes or antibodies, or surface coating (e.g., sputter coating commonly used in electron microscopy), (b) AFM operates well in both air and liquid enabling live cell imaging, (c) AFM imaging is inherently made in 3D, where the horizontal X-Y plane is accompanied by the Z (vertical) dimension, (d) state-of-the-art AFM modes can combine qualitative and quantitative information, such as high resolution imaging with Young's modulus, adhesion, and deformation data, rendering AFM a multifunctional tool in the researchers' quiver. The AFM modes, advantages and limitations, that are relevant to cancer research, are summarized in Fig. 2.

AFM Use in Cancer Research
In the area of cancer research, AFM was initially used to study cancer cells in vitro. Nowadays, however, AFM is increasingly being used for getting more insights on the characteristics of cancer cells in culture with ECM components and in tumor tissue sections (Fig. 3). AFM can provide characteristics of cancer and stomal cells' surface, their pericellular activity at nanoscale level, as well as nanomechanical information, such as cell and ECM stiffness, and evaluation of cell-ECM forces under close to physiological conditions. Furthermore, AFM can be used to study the proteolytic activity of cells, exosomes, cell- Fig. 1 a Atomic force microscopy operating principle. The schematic illustration represents a typical AFM apparatus consisting of a laser source, a cantilever with the mounted sharp tip, the piezoelectric scanner with the sample holder, and a photodetector. The sharp tip, with the help of the piezoelectric scanner, scans the sample surface. Due to the tip-surface interactions, the cantilever deflects the laser beam. The photodetector tracks the deflection, allowing an image of the sample topography to be obtained. b Topography (top) and AFM phase (down) images from fixed primary human fibroblast cultured on mica discs (AFM tapping mode with a CPII Veeco microscope). Each type of image offers different information: Topography image offers a distinct presentation of the boundaries, and the surface texture of the cells. The bright protrusions on the cell surface can either correspond to microvilli (solid arrow) or pyramidal structures (dashed arrow) that are known to be artificial features generated due to close tip-sample interactions. In the phase image method, the contrast is generated due to the heterogeneous mechanical properties of the cell cell or cell-ECM interactions and specific cell characteristics such as invadopodia that play a significant role in cancer cell invasion and metastasis.

Study of Cancer Cells Using AFM
AFM can offer significant information on cancer cell mechanics as it provides extremely high precision on the force applied to the sample; it can be used on live or fixed cells and offers quantitative measurements of cell mechanical properties along with imaging. In the region of interest in the cell surface, AFM applies a force and records force-deformation curves which are used to acquire the stress-strain response and the Young's modulus of the cell. AFM has been used to quantify and depict cancer cell morphology and cell stiffness in different cancer cell lines, such as lung and embryonal carcinoma cells, also providing comparative studies of normal and malignant cells of the same type [51][52][53][54]. So far, a wide range of cancer cell lines have been studied in vitro, including bladder carcinoma cells [55], H-ras transformed fibroblasts [56] and epithelial breast cancer cells [57]. Moreover, using the AFM nanoindentation mode measuring stiffness from tissue sections and biopsies was made possible, providing novel insights into the understanding of the different mechanical properties between healthy and tumor tissue sections facilitating the discovery of novel biomarkers for early diagnosis of cancer or monitoring of its progression [36,58,59]. Lekka et al., for instance, employed AFM to measure the elastic properties of tumors using tissue sections  from patients with various types of cancer at advanced stages [36]. Their results demonstrated that there were significant differences between healthy and tumor areas, since AFM was able to measure changes in tissue mechanical properties and detect cancer cells through their elastic properties. Additionally, Plodinec et al. used AFM nanoindentation mode on breast biopsies and demonstrated that malignant tissues had different stiffness profiles than normal ones [59].
The results, from the majority of the previously mentioned in vitro and ex vivo (i.e., in tissue sections) studies, have shown that cancer cells are softer than normal ones. Although this fact is widely accepted, the underlying mechanisms are not well understood and further research is needed. Additionally, not all cancer cell lines have the same mechanical behavior. For instance, in a recent study, Park et al. used AFM and nanoscaffolds to study how cancer cells (prostate and breast) mechanically respond to their microenvironment [60]. Their findings demonstrated that different cancer cell lines have distinct response to microenvironment characteristics. Specifically, in prostate cancer cells, proliferation and mechanical stiffness were correlated with the size of nanoislands (which are nanometric sized regions that bear different geometrical characteristic compared to the rest scaffold) on nanoscaffolds, but this was not the case for breast cancer cells, which demonstrated no dependence of mechanical responses on the geometric properties of nanoscaffolds.
Furthermore, the application of atomic force spectroscopy provides information on the adhesion forces required to disrupt specific molecules from samples [61], ligand-receptor interactions on cells, [61,62], as well as interactions between cells. Hoffmann et al. investigated the interaction forces between natural killer cells and tumor cells [63], showing an increment in the required force to separate natural killer cells from tumor cells under the engagement of the activating natural killer cell receptor 2B4.
Interestingly, the use of AFM in cancer research is not limited to the previously-mentioned areas, as it can be also used in many other cases, such as in acquisition of cell morphology images, in the study of substrate erosion by cancer cells [64,65], and for detection of cytotoxic events [66]. Krisenko et al. used a multiharmonic AFM method to map nanomechanical properties of live breast cancer cells either lacking or expressing the protein tyrosine kinase Syk [48], which is known to be a modulator of immune recognition receptor signaling, playing an important role in tumor progression. Their results offered new insights into the underlying mechanisms of Syk-dependent cell changes since it was demonstrated that cancer cells that expressed Syk had reduced cell height, increased elasticity, increased viscosity, and a more substantial microtubule network.
Despite its extensive applicability, studying cell mechanics and adhesion forces with AFM possess some limitations such as the need to control a wide range of possible parameters that influence the measurements and the difficulty to acquire a large number of measurements from different cells to get statistical significant data. Furthermore, in the majority of the previous mentioned studies, Young's modulus was measured from force indentation data using the Hertzian model. Although this model is widely used, it contains a significant number of assumptions and many researchers are trying to develop new models. For example, Hayashi and Iwata suggested a new model to overcome this limitation [2]. According to their data, they confirmed that cancer cells are softer than normal cells and they could not identify any differences in the stiffness of cancer cells between the central and the peripheral regions of the tumor. Finally, it must be highlighted that force scanning and force spectroscopy modes, as almost all AFM modes, are time-and labor-consuming methodologies and in cases that statistical analysis is demanded, other techniques, like immunofluorescent staining, may offer better results [61].

Study of Stromal Cells
Stromal cells are among the most important components of the tumor microenvironment, and their interactions with cancer cells, as well as their interactions with other components of the microenvironment, have been suggested to play a crucial role in cancer growth and progression [67]. The most common types of stromal cells involve fibroblasts and myofibroblasts or CAFs. CAFs are fibroblasts that are transformed due to a desmoplastic response, present a spindle shape [68], and are identified by specific markers, (such as α-smooth muscle actin) [69]. Although CAFs can promote tumor progression in a multitude of ways, the underlying mechanisms remain unclear and AFM can significantly contribute toward this direction. In a study by Castella et al., AFM was used to investigate how changes in the intracellular calcium concentration contribute to myofibroblast contraction [70]. They used AFM to assess contractile events by tracking stress-fiber-linked microbeads, and the contribution of calcium signaling to the regulation of myofibroblast contraction was assessed. Fluorescence microscopy was applied simultaneously with AFM so as to evaluate micromechanical cell events during changes in calcium. Myofibroblasts exhibited periodic calcium oscillations, which induced weak contractions, while depletion of calcium reduced these microcontractions. In a different approach, Kloxin et al. used AFM and photodegradable hydrogels, to investigate the effect of Young's modulus on valvular interstitial cell activation into myofibroblasts [71]. Hydrogels with tunable mechanical properties were used to screen the effect of microenvironment elasticity on cell differentiation without affecting cell migration. Their results showed that myofibroblasts can be de-activated solely by changing the stiffness of the underlying substrate. In a similar approach, Saums et al. developed a cell culture system that allowed the study of ECM protein mixtures controlling their mechanical stiffness [72]. This mechanically tunable system consisted of layer-by-layer assembled films of native matrix proteins on which hepatic stellate cells, known to differentiate to myofibroblasts in liver fibrosis, were cultured. AFM nanoindentation was applied to mechanically characterize the system showing that this approach can be used for an extensive research for the effect of matrix proteins and mechanical forces on cell phenotype.
Last but not least, AFM has been utilized extensively to visualize the effect of treatment with growth factors such as nerve growth factor [73], hepatocyte growth factor [74], vascular endothelial growth factor [74,75] and the matrix remodeling tumor growth factor-β (TGF-β) [76][77][78][79] on cell morphology, and cell properties that are related to cancer progression and metastasis.

Study of Cancer Cell-Tumor Microenvironment Interactions During Migration and Invasion
Cancer cell metastasis is a complex process involving cellcell interactions, cell-ECM interactions, and cellular architecture alterations. AFM can be used to study specific cell characteristics, activities, and interactions that are deemed to be important and might affect metastasis.
(a) Studies on proteolytic activity: At the early stages of cancer cell metastasis, tumor cells lose their contact with the ECM and their neighboring cells and acquire pericellular proteolytic activity in order to remodel their microenvironment so as to invade surrounding tissues [80]. While the exact mechanism is unknown, the proteolytic activity and biophysical parameters can be evaluated by AFM. Investigating cancer cell invasion can be challenging, however, and AFM can contribute significantly to the field because of its capability to detect differences in average height, volume, and molecular weight distribution of pericellular matrix proteins in the microenvironment of cancer cells or non-invasive cells [80][81][82]. Consequently, AFM is capable of morphometric imaging and quantification of nanoscale ECM alterations in the cellular microenvironment [80,82]. (b) Studies on exosomes: AFM can be also used for the nanoscale characterization of single exosomes [83], known to play important role as mediators of cellular functions such as tumor proliferation, pre-metastasis, and tumor microenvironment modulation [84][85][86]. In a recent publication, Sharma et al. used AFM to show quantitative differences between cancer cell-linederived and normal cell-line-derived exosomes [87]. They found that glioblastoma and malignant melanoma-derived exosomes possess surface nanofilaments, which may help in cell-cell communication, while the full clarification of their structure at the nanoscale level can provide insights into the mechanisms of exosome-mediated cell-surface interactions. (c) Studies on cell-cell interactions: Puech et al. demonstrated the ability of AFM force spectroscopy to study the binding effects of melanoma cell line (WM115) to human umbilical vein endothelial cells [88]. In another recent study, Laurent et al. used AFM to determine the adhesion strength between an endothelial cell monolayer and tumor cells of different metastatic potential [89]. Their results showed that the more invasive cell lines (T24, J82) formed the strongest bonds with endothelial cells and that ICAM-1 served as a key receptor on endothelial cells, while ICAM-1 did not play any role in the adhesion process for the less invasive cancer cells (RT112 They found that R cells were 2-3 times softer, than the E cells and the results demonstrated differences in mechanical properties, as well as molecular and phenotypical signatures between the two cell types. In another approach, Park and Lee [90] performed AFM force spectroscopy to characterize dual mechanical properties-the cell-substrate adhesion as well as the mechanical compliance of prostate and breast cancer cells. The authors concluded that mechanical compliance alone fails to serve as a universal indicator for metastatic progression and different therapeutic approaches should be considered for each type of cancer in order to protect the patients from metastasis. Guo et al. used AFM to determine the mechanical properties of individual human mammary epithelial cells in different microenvironments as related to cell-cell contacts (e.g., isolated cells; cells residing on the periphery of a contiguous cell monolayer; and cells on the inside of a contiguous cell monolayer) [91]. Their results showed a distinct contrast in the effect of cells' microenvironment on cells' stiffness, where normal mammary epithelial cells were found to be stiffer than peripheral cells and cancer cells were found to be less sensitive to their microenvironment than normal cells. These results demonstrated that the microenvironment plays an important role in determining cell stiffness and AFM can be used in order to investigate cell stiffness in different microenvironments or under specific modifications [2]. The role of cell stiffness in metastasis is still not adequately studied and further research is needed to reach safer and more consistent conclusions.  [64,65], the cellular processes extended by tumor cells enabling them to detect environmental cues and respond accordingly mainly by migrating and invading through surrounding tissues. The researchers used AFM to map the surface topographies and the fine details of cells and to investigate the nanodynamics of human glioblastoma brain tumor cells (T98) on collagen substrates, showing invadopodia of unusual nanomorphology, and describing their response to cell-to-cell signaling mechanisms.

Combined Imaging and Coupled AFM Systems
Coupling AFM techniques with other imaging methods (Fig. 2), such as optical microscopy, including fluorescence microscopy, confocal laser scanning microscopy, multiphoton microscopy (e.g., Second Harmonic Generation), and scanning electron microscopy can further expand its capabilities. Taking into account the fact that several research fields in life sciences are built on correlative imaging, the combination of AFM with other modalities is very promising. Furthermore, a significant number of manufacturers have started manufacturing of microscopes that combine AFM with one or more modalities.

AFM-Optical Microscopy
The combination of AFM with optical microscopy techniques, such as phase contrast, DIC, epifluorescence, and laser scanning confocal microscopy is important for the effective and thorough investigation of cells and their microenvironment [80]. Although, optical microscopy is generally faster, and superior to AFM in terms of imaging individual cells or cells' populations, providing better statistics, and application of fluorescent tags to label specific cell components [61], AFM has certain unique features, such as capability to measure mechanical properties, probing surface interactions, and nanoscale imaging. Consequently, the integration of AFM and optical microscopy offers the advantage of combining nanoscale topographical imaging with broader imaging field and biochemical capabilities with the optical methods. Furthermore, the combination of optical microscopy with AFM enables precise positioning of the AFM tip. Traditionally in the AFM systems, the tip is positioned on the areas of interest by the guidance of an AFM system camera. In some cases, finder and locator grids are used in order to map the sample surface and make some measurements repeatable [95,96]. The use of an optical microscope could offer more accurate positioning so as to perform high resolution imaging or force spectroscopy on a specific region of interest. The combination of the two modalities can be performed either by acquiring images from two separated systems or by unifying an inverted light microscope with an AFM system and using transparent sample supports. It is obvious that the first technique is challenging because the accurate characterization of the exact same area is not straightforward, but the second method can provide unique and accurate results if an appropriate and effective overlay of the two modalities is achieved. Interestingly, apart from the manufacturers who have initiated the development of such imaging modalities, researchers have already started to design combined microscopes, such as AFM and fluorescence microscope that can perform simultaneous optical/fluorescence and topographic measurements [97]. A number of research teams have combined AFM with optical microscopy for several research purposes such as for the study of focal adhesion complexes and identification of actin-related structures within cells [98,99], as well as for the study of the force transmitted from the apical to the basal cell membrane [100]. In fact, several researchers have applied these techniques for investigating issues closely related to cancer cells and tumor microenvironment. Kusick et al. used fluorescence microscopy in combination with AFM to measure the proteolytic activity in the microenvironment of tumor cells and evaluate their invasiveness and metastatic potential [82]. Li et al. combined AFM with confocal fluorescence microscopy in order to study the elasticity of normal and cancer epithelial cells in relation to their sub-membrane cytoskeletal structures [57]. They found that malignant cells have a significant lower Young's modulus, while significantly different sub-membrane actin structures were also observed, contributing to different mechanical properties and facilitating cell migration and invasion. Furthermore, Lopez et al. investigated the origin of tissue stiffening that is associated with mammary tumor development by using AFM in live and snap-frozen fluorescently labeled mammary tissues [101]. In situ labeling techniques were also applied to identify epithelial cells, vasculature, and ECM, while also assessing mechanical properties of several components of the breast tumor tissue. Their results demonstrated that during mammary gland transformation, the mammary gland epithelium, the tumor-associated vasculature, and the adjacent ECM stiffen significantly.
Finally, Fuhrann et al. used the combination of AFM with a scanning confocal fluorescence microscope to perform mechanical testing of single adherent cells from three different cell lines (normal, metaplastic, and dysplastic) [3], showing distinct differences between the mechanical properties of the studied cell lines.

AFM-Second Harmonic Generation
Another promising combination is the correlation of AFM images with multiphoton microscopy and especially with second harmonic generation (SHG). SHG is a nonlinear optical phenomenon during which new photons with twice the frequency and half the wavelength of the initial photons are generated when photons with the same frequency interact with a non-centrosymmetric material (also known as a nonlinear material) [102]. The process is more effective when the initial photons originate from laser sources, providing high intensities, monochrome photons, and highly focused beams. SHG is emerging as a powerful technology in biomedical studies with applications ranging from microscopy to molecular orientation probing due to its unique features that include ability of non-invasive imaging, selectivity to specific tissue components and also it does not require use of staining [103][104][105][106]. The investigation of SHG signals from biological tissues can contribute to the development of new biomarkers such as SHG from collagen [105,[107][108][109][110][111], which is a natural non-centrosymmetric biomaterial and has the ability to produce extremely bright and robust optical SHG signals. Detecting changes in collagen organization is crucial as its concentration, structure, and function are different in various pathological conditions [112]. Moreover, in the tumor microenvironment field of research, there is a strong interest to study collagen concentration, architecture, stiffness, and its associated modifications since its effect on tumor progression is rather crucial [113]. For instance, SHG microscopy has been used to show that collagen undergoes significant morphological modifications during tumor progression and this ECM remodeling was demonstrated to be due to collagen lysyl oxidase (LOX)-crosslinking, showing a close relationship among ECM stiffness, tissue fibrosis, and breast tumorigenesis [114]. Consequently, the combination of AFM and SHG microscope for collagen samples can contribute significantly to the field, as it can facilitate the study of collagen structure/function and the study of the biochemical composition, architectural features, and mechanical properties of collagen while simultaneously monitoring cancer cell behavior [113]. Interestingly, Maller et al. investigated changes in the mammary epithelium during early-age pregnancy as part of research for the reduction of breast cancer risk [115]. They used AFM and SHG methods to evaluate whether ECM contributes to parity-induced tumor suppression and studied the function, composition, and spatial organization of mammary ECM in nulliparous and parous rats. Their findings indicated abundance of collagen which was less linearized and associated with a decrease in stromal stiffness. They suggested that the organization of fibrillar collagen, and not density, is associated with tumor cell invasion [115]. In another work, the role of Caveolin-1, (Cav1, the major component of endocytic caveolae plasma membrane) was investigated [116]. SHG was applied for imaging tumor architecture and AFM for characterizing its mechanical properties. Cav1 was found to promote force-dependent contraction, matrix alignment, and microenvironment stiffening, while the remodeling of the microenvironment by Cav1 forced cell elongation, clearly indicating that Cav1 plays a significant role in microenvironment remodeling essential for tumor progression and the regulation of matrix-dependent cell behavior. Pickup et al. used the combination of AFM (for measuring ECM stiffness) with SHG (for collagen fiber diameter measurements and fiber quantification) to demonstrate that stromal-derived LOX promotes metastasis of TGF-β deficient mouse mammary carcinomas [117]. Finally, in a very recent publication, Navab et al. studied the role of integrin α11β1 (a stromal cell-specific receptor for fibrillar collagens that is overexpressed in CAFs) in cancer progression [118]. AFM was again applied for the quantification of Young's modulus of the cancer-associated stroma and SHG was used for imaging and quantitation of collagen organization. Their results demonstrated that α11β1 integrin expression was associated with collagen cross-linking whereas loss of its expression was correlated with decreased collagen reorganization and stiffness.

AFM-Scanning Electron Microscopy AFM and Scanning
Electron Microscopy (SEM) are two of the most commonly used techniques for high resolution surface imaging and characterization. Both can produce nanoscale topographic images of biological samples that are well suited for investigating changes in cell and tissue characteristics [119]. However, their principles are very different and as a result, they can offer different, overlapping, and complementary types of information about tissue/cell structure. AFM can provide information regarding mechanical properties of the samples in all three dimensions. On the other hand, SEM can image rougher specimens; it has larger depth of field and can also offer elemental analysis using X-ray detection. That is why, in many laboratories, AFM and SEM are used side-by-side, while coupled hybrid systems have started to be developed.
In the case of cancer and tumor microenvironment research, the combination of the two modalities can offer unique information. A first area where AFM-SEM can be applied is the investigation of the in vitro effect of ECM on cells. Grabb et al. used the two modalities in order to investigate the composition and microstructure of the ECM deposited by fibroblasts as a function of time, geometry, and matrix microstructure [120]. Stromal fibroblasts were seeded onto collagen matrices (e.g., films, sponge) whose characteristics were fully evaluated by AFM and SEM. Their results demonstrated an increase in the expression of myofibroblast phenotype markers with time for cells cultured on a collagen sponge, while changes in phenotype were minimized for cells cultured on collagen films. This suggests that the microstructure and not the geometry of the ECM matrix affect cellular phenotype as well as the newly synthesized ECM. In another study, Li et al. studied the behavior of giant cell tumor of bone stromal cells when cultured on self-assembled monolayers of alkanethiols on gold with different terminal chemical groups [121]. AFM was used to characterize the density of the different surfaces and SEM (as well as fluorescence microscopy) was applied to analyze the morphological and biological changes of the cells [122]. Their results confirmed that surface chemistry affected cell morphology, adhesion, proliferation, and apoptosis. In another study, Cassereau et al. used synthetic collagen-based hydrogel to study the impact of ECM composition, density, mechanics, and topography on transformed mammary epithelial cells [122]. AFM was applied to measure the Young's modulus of the collagen hydrogels and SEM to determine the hydrogel porosity and organization. The authors demonstrated that their system can be used with tunable stiffness (while maintaining consistent composition and pore size) that can be used for further investigation of the role of ECM stiffness.
A second area where AFM-SEM combination can be used is the characterization of tissue and cells from biopsies/sections. Kaul-Ghanekar et al. investigated the effect of the tumor suppressor scaffold/matrix associated region binding protein 1 (SMAR1) expression on cell lines, as well as tumor sections from human and mouse breast cancer tissue of different grades [119] using AFM and SEM to study cell morphology and surface roughness. They reported the morphological differences between cancer cell lines and cells overexpressing SMAR1 and suggested that SMAR1 could be used as a phenotypic differentiation marker between cancer and normal cells. Finally, Plodinec et al. used SEM to evaluate tissue architecture and surface integrity of human breast biopsies and AFM for measuring their stiffness profiles [59]. Their novel results demonstrated that AFM measurements of stiffness can be used as indicators in clinical diagnostics of breast cancer with translational significance. Last but not least, the combination of the two modalities could be applied in order to investigate cell characteristics after treatment with specific factors. For instance, Volakis et al. used AFM to evaluate how myoferlin depletion affected cell stiffness and the integrity of the cytoskeletal network and SEM to determine cell surface area and the number and length of lamellipodia and filopodia [123].

Conclusions
This review demonstrated the importance of AFM in the fields of cancer cell and tumor microenvironment research. Other applications of AFM not covered here include its use as a molecular toolbox in nanobiotechnology (e.g., use of AFM probe as a 'lab-on-a-tip' that can probe simultaneously the structure and specific parameters of the cell's machinery) [124] or for investigating the action of drugs on cells, as well as the effectiveness of nanoparticle delivery for cancer therapeutic strategies [125][126][127]. Furthermore, the research in the development of new coupled AFM systems, or modes, is ongoing and new modalities emerge such as the combination of AFM with Raman imaging/spectroscopy [128][129][130][131] or with infrared (IR) nanoimaging [70]. The AFM-Raman combination can additionally offer a label-free chemical analyses of surfaces and cells [130]. The integration of AFM with IR can produce subcellular chemical density and complexity maps of both healthy and cancer cell lines without staining and has already been applied to map biochemical heterogeneity of cancer cells [70].
Although, AFM has revolutionized the way we study cells and tumor microenvironment components in the nanoscale, there are technological challenges to be solved and there is still more work to be done concerning AFM instrumentation, experimental setup, and procedures [83,124,125,129,130,132]. A significant issue concerns the technical difficulties in AFM live cell imaging and nanocharacterization. Novel techniques are demanding in order to improve the spatial resolution of AFM, especially in the case of live cell imaging [129]. Also, the time required to take an AFM image of a live cell is often too long to address dynamic processes of cells [129] and the dynamic movement of the cell membrane may destroy the high resolution [125]. Consequently, methods that can improve lateral resolution, reliability, time acquisition and inhibit the dynamics of cell surfaces face certain limitations. Furthermore, the AFM procedures are time consuming, require intensive labor, and lack of automation [125]. Improvements in these directions will enable the performance of large number of repeats per experiment so as to achieve statistical significant results. Additionally, the reproducibility of the experiments and the minimization of the time will be benefited by the development of simple and standardized protocols, especially for the functionalization of AFM tips and the measurements of the spring constant of the cantilever which are demanded for AFM force spectroscopy and nanoindentation modes. Furthermore, the limitations of the existing mathematical models that are used for the acquisition of quantitative data with AFM modes [133] suggest further research toward the development of new mathematical expressions and models with higher accuracy. Regarding the research in the area of cancer and tumor microenvironment, further efforts need to be made toward the nanocharacterization of the malignancy of tissues from biopsies with AFM. The research so far suggests that AFM can provide unique clinical diagnostic biomarkers to be used as early diagnosis markers or as markers of tumor progression, but so far, only few studies have been performed toward this direction. Furthermore, the influence of numerous components of the tumor microenvironment (e.g., growth factors, ECM proteins) on cancer cell behavior in the nanoscale remains unclear and AFM can contribute to this kind of research in order to clarify the underlying mechanisms. Finally, a promising future direction that could significantly contribute to the field is the achievement of real in vivo AFM characterization of tissues for cancer research. The operation principles of AFM render the in vivo use of AFM challenging, but there are a few studies which have already demonstrated that under appropriate conditions and in combination with other procedures (e.g., surgery), in vivo AFM imaging and nanocharacterization can be achieved [134,135]. In conclusion, research concerning cancer cells and tumor microenvironment is crucial in order to better understand cancer pathophysiology and develop more effective therapeutic approaches. AFM is emerging as a multifunctional tool with a good potential to bridge significant gaps in our understanding of cancer cell behavior and the interactions with the tumor microenvironment. AFM offers a wide range of qualitative and quantitative information, from nanoscale imaging to mechanical properties characterization. In this review, we highlighted the use of AFM on the mechanical properties and nanoimaging of cancer and stromal cells. Furthermore, the review describes the potential use of AFM to study cell-microenvironment interactions during cell migration and invasion. Finally, we summarized the novelty and the applications of using AFM in combination with other microscopes/modalities in tumor microenvironment research.