2.1.

Atomic Force Microscopy

The AFM¹ has become an important tool for studying biological samples due to its ability to combine both structural and functional studies of samples immersed in liquids. The principle of operation of the AFM consists of an atomically sharp cantilever tip, which passively senses the localized forces between the molecules of the scanning tip and cell surface. This “nanosensor” can image live cells with atomic resolution^{24, 25, 26, 27} and probe single molecular events in living cells under physiological conditions.^{28, 29, 30, 31, 32} Currently, this is the only technique available to directly provide both structural and functional information at high resolution.

2.2.

Total Internal Reflection Fluorescence Microscopy

TIRF microscopy is based on the total internal reflection phenomenon that occurs when light passes from a high refractive medium into a low refractive medium. At the interface, the light will bend, traveling along the interface if the incident angle is equal to a so-called critical angle that is determined by the ratio of the refractive indices of the two adjacent media. If the incident angle is higher than the critical angle, the light will turn back into the high refractive medium, and only a short-range electromagnetic disturbance called evanescent wave will pass into the low refractive medium. The evanescent wave intensity decreases exponentially with distance from the interface, confining the eventual fluorescence excitation to the near-field region ( $\sim \lambda ∕4$ to $\lambda ∕5)$ .³³ This imaging method provides high contrast images of the near-membrane area that contains the finest details of cells grown on coverslips, with a useful maximum depth of penetration of $\sim 100\mathrm{nm}$ . Due to minimal exposure of cells to light in any other planes in the sample, TIRF images have very low background with virtually no out-of-focus fluorescence. When coupled with fluorescence labeling of specific molecules, TIRF is able to offer biochemical specificity for the visualization and spectroscopy of single-molecule fluorescence near a surface. Using a thorough calibration of the microscope, the separation distance between cell membrane and substrate can be measured.^{34, 35} Also, any movement of fluorescent-tagged proteins in focal adhesion contacts or at the membrane level can be monitored in real time using time-lapse images. The total internal reflection condition is achieved in the microscope with off-center laser illumination from the periphery of the back focal plane of the objective. The sensitivity of the method is directly related to fluorescence intensity, which is maximal for separation at the interface and decreases exponentially with increase of the separation.

2.3.

Interference Reflection Microscopy

Regular reflection microscopy is difficult to apply to cell studies due to the enormous stray light relative to the very low reflection intensities of cells (the refractive indices of cells and water are very close). In biology, the IRM was used from the beginning for the visualization of attachment sites of living cells due to the contrast obtained from interference in monochromatic polarized light.^{6, 36} Due to interference, the cell-glass interface appears as various shades of gray with increasing darkness, indicating a closer approach of the cell to the coverslip. The darkest areas are the result of destructive interference at the cell-coverslip interface, and the lighter areas represent the result of different degrees of constructive interference.³⁷ IRM is the easiest and fastest method to obtain immediate information about the overall focal adhesion distribution, which represents the cell fingerprint at the coverslip level. Due to the fact that this method does not employ fluorescence labeling, cell-substrate separation distances can be determined even in unstained cells.³⁸

2.4.

Forster Resonance Energy Transfer

FRET is a quantum process of nonradiative transfer of energy from a donor fluorophore to an acceptor fluorophore. Although fluorescence is usually used to measure the energy transfer efficiency, fluorescence by itself is not part of the FRET mechanism.^{39, 40} When a donor fluorophore absorbs light, it changes from the ground state to a vibrational level in an excited state. This fluorophore will decay spontaneously (i.e., within picoseconds) to the lowest of the vibrational levels of the excited state without emitting a photon. The fluorophore can de-excite further to the ground state with the emission of a photon (nanoseconds—fluorescence effect), whose wavelength is longer than the excitation wavelength. Alternatively, if an acceptor molecule (properly oriented and spaced) is adjacent to the donor, the acceptor will absorb that energy undergoing a transition to a higher excitation state, and consequently will de-excite fluorescently with a photon emission at an even higher wavelength (FRET effect) than the donor emission. This energy transfer is the result of long-range dipole-dipole interactions between donor and acceptor, and takes place when the two molecules are close $(<10\mathrm{nm})$ , with the transfer rate varying with the sixth power of the separation distance between the two molecules.^{41, 42} For FRET to occur, not only the distance between the donor and acceptor molecules is important, but also the dipole orientation of the two molecules should be optimal, and there should be a significant overlap between the donor emission and acceptor excitation spectrum, with a high donor quantum yield.^{43, 44, 45} FRET is used to monitor protein interactions^{46, 47} and conformational changes,^{48, 49, 50} or to determine molecular distances between fluorophores.^{51, 52}

4.1.

Microscope and Optics

Our imaging system is assembled around an Olympus IX-81 (New York) inverted microscope that offers a stable platform with multiple camera ports that have simple optical paths from the objective. The microscope stand itself is heavy and stable, providing the option to split the light at two levels inside the stand. Further, this microscope provides a split box on the epiport to facilitate the dual connection of a TIRF illuminator and a custom designed epi-IRM-illuminator. The microscope is motorized and incorporates a revolving back-arm that allows 360-deg rotation for the condenser and a transilluminator with a halogen lamp housing. The revolving back-arm is essential for mounting the AFM on the inverted platform, because the condenser can be offset without disassembly, so that the AFM scanner support can be positioned. The microscope has an electronic focus with a total travel of $9\mathrm{mm}$ and a resolution of $0.01\mathrm{\mu }\mathrm{m}∕\mathrm{step}$ with $1$ - $\mathrm{\mu }\mathrm{m}$ repeatability that can be controlled manually or via software. A Lambda LS (175 W Xe, Sutter Instruments, Novato, California) is used as an alignment-free epifluorescence lamp. It is equipped with a double filter wheel (the internal one is used for excitation filters and the external one is used for neutral density filters, ten positions each) and is fed into the IRM/epi-illuminator through a light guide. All excitation filters are mounted into the filter wheel, unless otherwise specified. A Halogen 100-W lamp with a long working distance condenser ( $\mathrm{NA}=0.55$ , $\mathrm{WD}=27\mathrm{mm}$ ) is used for transillumination. A tunable $\mathrm{Ar}\text{-}\mathrm{Kr}$ laser (Spectra Physics, Mountain View, California), fed into the TIRF illuminator through an optical fiber, is used as an excitation source for TIRF experiments.

The microscope uses infinity-corrected, high-resolution objective lenses. These include an Olympus TIRF objective Plan Apo $60\times \mathrm{NA}=1.45∕\mathrm{oil}$ (Olympus, Japan) and a Zeiss Antiflex objective Plan NeoFluor $63\times \mathrm{NA}=1.25∕\mathrm{oil}$ that has a $\lambda ∕4$ wave plate built into the body of the objective (Zeiss, Germany). Our imaging applications require Plan (flat field) and Apochromatic (four color chromatic aberration, two color spherical aberration) corrections to preserve spatial relationships without distortions throughout the field of view, which contains a single cell. However, the Antiflex Plan NeoFluor is a semiApochromat lens that is corrected for flatness of field and for two-color chromatic aberration and spherical aberration. This objective is better suited for color photomicrography than the Achromats due to the fluorite glass content.⁵³ Further, because cells are attached to a coverslip, only oil immersion lenses are needed (nonfluorescent oil, $n_{\mathrm{oil}}=1.516$ , Olympus, Japan).

All optical components (fluorescence filter sets, mirrors, polarizers, and protection filters) are from Chroma Technology (Rockingham, Vermont) unless otherwise specified. The filter sets (see Table 1) were chosen to match the fluorophores of interest and the wavelengths available from the $\mathrm{Ar}\text{-}\mathrm{Kr}$ laser. Heat reflectors are mounted in front of the excitation filters.

4.1.1.

Total internal reflection fluorescence illuminator

The TIRF optical system has the following features (Fig. 3 )

• 1. The laser beam used for excitation is collimated (parallel beam) and focused at the back focal plane of the objective at a chosen angle of incidence $\theta$ with respect to the optical axis of the microscope.

• 2. The point of focus in the back focal plane is off-axis. There is a one-to-one correspondence between the off-axis radial distance “d” (see Fig. 3) that is dictated by the distance between the optical axis of the microscope and the fiber position at the entrance of the TIRF illuminator (see D in Fig. 3), and the angle of incidence $\theta$ . By moving the micrometer that changes the fiber position in respect with the optical axis, the axial radial distance will increase and the critical angle of incidence can be exceeded to allow supercritical angles to be reached in a reproducible manner. This adjustment is limited by damping the laser light into the edge of the objective.³ The TIRF objective (glass BK-7, $n_{\mathrm{glass}}=1.516$ ) allows a range of incident angles between ${\theta }_c$ $(\arcsin n_{\mathrm{water}}∕n_{\mathrm{glass}}=61\mathrm{deg})$ and ${\theta }_{\max }$ (arcsin $\mathrm{NA}∕n_{\mathrm{glass}}=73\mathrm{deg}$ ).

• 3. The opening angle of the laser beam $\left(\alpha \right)$ at the back focal plane of the objective determines in a proportional manner the size of the illuminated area at the sample plane (“a” in Fig. 3). In our system, the TIRF illuminated field of view has a diameter of $195\mathrm{\mu }\mathrm{m}$ (the epi-illumination field of view is $340\mathrm{\mu }\mathrm{m})$ .

Fig. 3

Schematic representation of the TIRF optical path. The TIRF illuminator provides a micrometer that allows the controlled movement of the optical fiber off-axis to achieve the condition of total internal reflection at the sample level. The total reflecting mirror from the epi-TIRF split box should be set in the optical path to direct the laser into the objective. The reflected light from the sample plane is further sent to the left CCD camera where the image is recorded.

4.1.2.

Internal reflection microscopy illuminator

The IRM uses monochromatic light (usually $546\mathrm{nm}$ ) selected through a narrowband filter and, since the intensity of the reflected light is very low $(<1\%)$ compared with the incident light, a powerful light source must be used.⁵⁴ There are a few important components of the optical system that require attention (Fig. 4 )

• 1. It is advantageous to eliminate stray light from areas outside the field of interest as much as possible by closing the field diaphragm. Usually the field diaphragm is visible in IRM images.³⁸

• 2. The aperture diaphragm has a significant influence on the IRM image. When the aperture diaphragm is closed, to some extent the illuminating numerical aperture (INA) at the sample level is smaller than the NA of the objective.³⁸ There is a direct correspondence between the diameter of the aperture diaphragm and the INA at the sample level. At higher INA, the contribution of deeper structures from relatively thick layers of cytoplasm is much weaker than in the case of normal incidence,⁵⁴ but the image from the cell-coverslip interface is preserved.⁵⁵ The oil immersion objective helps avoid reflections at the glass-air interface between the coverslip and objective lens. Even the weaker reflections inside the objective produce enough stray light to degrade the IRM image.³⁷ Therefore, we use the antiflex principle introduced into IRM by Ploem.⁷

Fig. 4

Schematic representation of the IRM optical path. The monochromatic light from the Xe lamp is fed into the IRM illuminator through a light guide. After passing through the aperture and field diaphragm, the light is linearly polarized and directed into the objective. The light that passes through the objective reaches the quarter-wave plate on its way to the sample and becomes circularly polarized. On its way back, after reflection from the object, it is linearly polarized again but with a $\pi ∕2$ difference in the direction of polarization and passes through the analyzer. The light is further sent to the left CCD camera where the image is recorded.

The incoming IRM light from the epi-illuminator is plane polarized before being reflected into the objective, and the analyzer behind the objective is crossed with the polarizer to cut off all reflections within the microscope body and the objective. As the Antiflex objective has a revolving plane-parallel $\lambda ∕4$ plate mounted on the front lens of the objective, the light beam that forms the image passes through this plate on its way to and back from the sample. If the $\lambda ∕4$ plate is properly oriented (the incident light should make a 45-deg angle with the plate optical axis), the light becomes circularly polarized at the first pass, reaches the object, and is reflected back. At the second pass through the wave plate, the reflected light again becomes linearly polarized but with a $\pi ∕2$ phase difference in the direction of polarization and passes through the analyzer. Thus, only the reflected light that comes from the sample is participating in the image formation.⁵⁴

4.1.3.

Forster resonance energy transfer optical path

In a standard epi-illumination configuration for sensitized FRET experiments, three filter cubes are used: excitation donor/emission donor, excitation acceptor/emission acceptor, and excitation donor/emission acceptor.⁴⁵ Optical misalignment of this combination of filters can easily introduce spatial shifts in the data, making the weak FRET signal difficult to detect.⁵⁶ To minimize the spatial shift, several aspects of filter set design should be considered: the dichroic mirror used for capturing all images should be the same, the donor/acceptor emission filters should be thin and of the same thickness, and heat reflecting filters should be used in front of the excitation filters at the arc lamp. For the dye combination (CFP/YFP, Forster radius $\sim 5\mathrm{nm}$ )^{57, 58} that was selected for use on the NanoFluor system, another problem is presented by the overlap of spectra tails from the two dyes that will generate bleedthrough between channels. This can be minimized using a judicious choice of the filter sets and proper corrections.^{59, 60} Consideration of these facts led to the generation of a simpler optical path design (Fig. 5 ), using a double-band dichroic at the turret level that transmits all channels and uses no filters. The light split is done inside the microscope stand in such a way that the donor channel is acquired with the right camera, while the acceptor and FRET channels are acquired with the left camera. The image acquisition is done sequentially via software control. The first sequence consists of the simultaneous acquisition of donor and FRET signal, followed by the acceptor signal in a different sequence. Using a three-channel FRET algorithm,⁵⁹ the corrected FRET image is calculated on a pixel-by-pixel basis, taking into consideration the bleedthrough between channels. In this way there are no moving parts in the system, and all critical optical components are chosen for a precise match.

Fig. 5

Schematic representation of the FRET optical path. The monochromatic light set on donor or acceptor excitation wavelength from the Xe lamp is fed into the epi-illuminator through a light guide. The light reaches the turret where it is reflected into the objective and reaches the sample. After sample excitation, the light emitted in the process passes undisturbed through the turret double dichroic mirror, and reaches the beamspliter where the donor emission is reflected to the right CCD camera and the acceptor emission is directed toward the left CCD camera. The FRET excitation path is the same with the donor excitation, and the emission path is the same with acceptor emission.

4.2.

Atomic Force Microscope

The Bioscope AFM system, equipped with a $xyz$ closed-loop scanner and Nanoscope IV controller (Digital Instruments, Santa Barbara, California), is mounted on top of the inverted microscope (Fig. 6 ). The AFM stage represents the key element that must be adapted to replace the classical microscope stage. Proper design of this AFM stage facilitates the union of the two instruments. The AFM scanner has a low noise level with maximum $xy$ scanning range of $100\times 100\mathrm{\mu }\mathrm{m}$ , $z$ scanning range of $9\mathrm{\mu }\mathrm{m}$ , and a scan speed of $\sim 40\mathrm{\mu }\mathrm{m}∕\sec$ . The acquisition time required to obtain one image varies, depending on the image size and resolution ( $256\times 256$ pixels up to $1064\times 1064$ pixels).

Fig. 6

Schematic illustration of the AFM system. A flexible cantilever with a tip at the end is rigidly connected with a $xyz$ piezoelectric element. The AFM is operated under fluid having the tip barely touching the cell surface. The optical lever consists of a laser diode beam that is focused on the back of the cantilever and bounces off, reaching the quadrant photodetector. The electrical signals coming from the photodetector are sent to the controller and used to measure the cantilever deflection. The $xyz$ position of the cantilever is given by recording the corresponding position of the piezoelectric element. During AFM experiments (topography or force measurements), the cells can be monitored on a video camera attached to a side port at the trinoc level.

To obtain images of single cells in culture, the AFM is operated under fluid in contact or tapping mode. In the scanning process in contact mode, the instrument is set to scan the cell in the $xy$ plane. During scanning, both height image (piezo element position on the $z$ axis) and the deflection (error signal) image (cantilever deflection on the $z$ axis) are recorded.

In contact mode, the height image is obtained by using a feedback system that keeps the deflection of the cantilever constant by changing the height of the probe while adjusting the voltage applied to the $z$ piezo. The change in $z$ axis corresponds with the true topological height of the sample at each point in the $xy$ raster. The 3-D image is constructed by combining the information of the three axes.

The $z$ -axis translation of the cantilever deflection (nanometers) is translated into an electrical signal (volts) using an optical lever system,^{24, 61, 62} consisting of a laser diode beam that reflects off the back of the cantilever and is projected onto a photodetector. The change in the position of the reflected beam follows the change in the cantilever deflection. The optical lever amplifies the deflection signal, so that even deflections $<1\mathrm{nm}$ can be measured.⁶³ In this way, the “tall” features in the image correspond to regions of greater degree of cantilever bending. There are more fine details visible in the deflection image than in the height image due to the smaller mass displacement of the cantilever when compared with the piezoelectric element (faster response from the control loop for correction of the cantilever deflection than in the correction of piezoelectric element displacement).⁶⁴ The laser diode emission wavelength $\left(690\mathrm{nm}\right)$ was chosen beyond the spectral range used for fluorescence measurements to avoid spectral interference.

For force measurements, the AFM is operated in force mode such that measurements of the relative stiffness of the cell surface (approach curve) and adhesion force measurements between the biologically functionalized AFM tip and the cell surface (retraction curve) are acquired (Fig. 7 ). The piezo is set to drive the cantilever to touch and retract over a predefined distance in the $z$ axis at a fixed $xy$ position. The $z$ -axis movement of the piezo and the deflection signal from the cantilever are recorded in a force curve (Fig. 7). When the probe is extended toward the cell surface (part A), a cell contact point is established (point B), and thereafter the cell surface is indented. Because of the cell stiffness, further probe extension causes an opposing force of increasing magnitude to be generated, along with increasing indentation in the cell membrane (part B to C). The upward (positive $z$ ) deflection of the cantilever as it bends in response to this force results in an increasing deflection signal. The membrane indentation part of the force curve can be analyzed using various models^{65, 66} to determine the local values of the Young modulus of elasticity as a measure of the apparent elasticity of the cell membrane. When the probe retracts from the sample (part C to D), the force between probe and sample gradually decreases until the cantilever returns to the original position, at which time the deflection signal returns to the original value (part A). However, if adhesions occurred between the probe and sample surface, the force causes the cantilever to bend downward (negative $z$ ), and the deflection signal is lower than the original value (parts E, F and G). When all adhesions are broken, the cantilever rapidly returns to the original position, and the deflections (unbinding of adhesion events) are recorded on the retraction force curve. The adhesion force is calculated by multiplying the deflection height associated with the unbinding event and the spring constant of the cantilever.

Fig. 7

Diagram of generic force curves. Approach (black thin line) and retraction force curves with (black thick line) and without (middle line) adhesions are presented. The $x$ axis represents the piezoelectric element displacement and the $y$ axis represents the force calculated as the product between the photodetector signal and spring constant of the cantilever. The labeled probes were set to touch and retract from the cell surface at a fixed speed. When the AFM tip approaches the sample (going from right to left), first there is no deflection (noncontact) regime (A). Moving the tip further toward the sample, there is a moment when the tip reaches the surface and establishes contact with the cell (B). Moving the tip further in the same direction causes deflection of the cantilever (contact regime). At a certain indentation (C), the tip begins to move away from the sample, causing the tip to bend upward, and finally detach from the sample (D) and lose contact (middle curve). If specific adhesion events occurred between the functionalized tip (e.g., fibronectin) and molecules on the cell surface (e.g., ${\alpha }_5{\beta }_1$ integrin) during the approach procedure, the retraction curve (thick black line) will record the appearance of distinct bond ruptures (E, F, G).

4.3.

Cameras

The CCD cameras used for fluorescence and IRM imaging are CoolSnap Monochrome HQ (Roper Scientific/Photometrics, Tucson, Arizona) that are fast, high-resolution imaging systems designed for low-light scientific applications. CoolSnap HQ is Peltier cooled to $-30°\mathrm{C}$ , supplemented with a remote fan for extra cooling. The CCD image sensor is a Sony ICX285 progressive scan interline transfer device with microlenses to provide high quantum efficiency for improved signal-to-noise ratio. The progressive interline transfer design consists of rows of photosensitive and masked pixels that are connected by a readout electrode that, during imaging, transfers the information from each vertical exposed channel to an adjacent masked channel. This design allows electronic shuttering at high speeds and on-chip integration. Incorporation of on-chip microlenses improves the optical collection efficiency of the interline transfer devices by increasing the effective photosensitive area to 70 to 90% of the surface.⁶⁷ This type of camera offers excellent resolution— $1392\times 1040$ array of $6.45\times 6.45\text{-}\mathrm{\mu }\mathrm{m}$ pixels (due to the fact that some pixels are masked off for reference values, the actual number of pixels in the image is $1344\times 1024$ ), with dual readout rates having low-noise characteristics ( $10\mathrm{Hz}$ with $6{\mathrm{e}}^{-}\mathrm{rms}$ or $20\mathrm{MHz}$ with $8{\mathrm{e}}^{-}\mathrm{rms}$ ) and high dynamic range (12 bits/pixel). The typical frame readout is $100\mathrm{ms}$ with a binning that can vary from 1 to 4, and a quantum efficiency of $\sim 60\%$ in the visible range. The dual camera capture system controls both cameras simultaneously for real-time FRET acquisition.

To obtain good resolution of a specimen imaged through the microscope, the Nyquist sampling theorem has to be fulfilled. To correctly reconstruct an analog signal (microscope image) converted to digital signal (CCD camera), the sampling frequency should be at least twice the frequency of the input signal. In this case, the pixel size is the sampling unit and the $x$ - $y$ microscope resolution is given by Raleigh criterion. To fulfill the Nyquist theorem, the necessary objective magnification must be calculated as the ratio between two times the pixel size and $x$ - $y$ microscope resolution. Our cameras and objectives were chosen to match the Nyquist theorem.

The cameras are mounted on the left and right ports of the inverted microscope via c-mount adapters. Both microscope ports are equipped with $xyz$ -adjustable mounts, and the cameras are mounted on $xyz$ micrometric stages, allowing for pixel-by-pixel alignment of the camera chips in hardware. This alignment is critical for dual capture of FRET images, because with almost perfect optically paired camera chips (same pixel register), no adjustment is necessary on captured images.

The microscope is also equipped with a video camera (Hitachi $2∕3^{”}$ B/W) on the trinocular port that allows cell monitoring during AFM experiments and is also used in AFM tip alignment.

4.4.

Software

The microscope control and all imaging data acquisition (3-D time-lapse and ratiometric imaging) are conducted in Slidebook 4.0.8.1 software (Intelligent Imaging Innovations, Denver, Colorado). This software also allows for TTL synchronization and extensive off-line data processing (3-D rendering and constrained iterative deconvolution with measured/calculated point spread function). The TTL board (DAQ 6023E, National Instruments, Austin, Texas) enables rapid, single-pulse communication and synchronization between the computer and other collection hardware such as shutters and cameras. This converter can be used to synchronize additional hardware (AFM controller) in an optimized capture scheme.

AFM acquisition is controlled by Nanoscope 611a software (Digital Instruments, Santa Barbara, California) that acquires images and force curves, and allows multiple options/working regimes for controlling the piezoelectric element. Automatic force curve processing is achieved using NforceR (copyright pending) proprietary software developed in-house. MatLab (Mathworks, Natick, Massachusetts) is used for further processing of the AFM images.

4.5.

Vibration Isolation and Electrical Wiring

External conditions in the laboratory where the microscope system is installed play a major role in the instrument performance, especially when it is used for live-cell imaging in time-lapse fashion over long observation intervals. There are three main sources of vibration that can affect the microscope: building vibration, acoustic noise, and direct disturbances. Also, to minimize the drift of the $z$ microscope focus, the room temperature must be kept constant $(\pm 1^{\circ })$ .

To reduce the effects of building vibrations, the extremely sensitive AFM and optical imaging system was mounted on an $8\times 5\times 1^{\prime }$ research-grade optical table top (Newport, Irvine, California), combined with four I-2000 stabilizer vibration isolators. A self-standing prewired shelf unit ( $20\mathrm{A}$ , $120\mathrm{V}$ ) is mounted above the table, and all heavy wiring leaving the tabletop is routed through a damper. All moving parts (camera fans, Sutter lamp with moving filter wheels, camera power supplies, and AFM controller) are located on the self-standing shelf unit.

To provide optimal conditions for this complex system, the room was wired specifically for this instrument, with separate electrical circuits as follows: AFM, microscope, and Lambda LS lamp are each wired on single-phase ( $20\mathrm{A}$ , $120\mathrm{V}$ ), computers ( $15\mathrm{A}$ , $120\mathrm{V}$ ), and the laser has a wall breaker, three-phase with ground ( $45\mathrm{A}∕\mathrm{phase}$ , $208\mathrm{VAC}$ ). It is highly desirable to have the fluorescence lamp on a separate circuit than the rest of the instrument components to be able to turn it on/off when all the other instruments are turned on. To reduce electrical noise in the AFM measurements, a medical grade line conditioner (Newark Electronics, Omaha, Nebraska) was installed between the electrical system and the very sensitive AFM controller.

4.6.

Laser and Water-Cooling System

An $\mathrm{Ar}\text{-}\mathrm{Kr}$ tunable laser Stabilite 2018-RM (2.5-W total power, Spectra Physics, Mountain View, California) that is fed into the TIRF illuminator through a single-mode optical fiber ( $\mathrm{NA}=0.1$ , coupling efficiency $>65\%$ ) was used as an excitation source for TIRF imaging. The laser can be tuned manually over a range of wavelengths in the visible spectrum for exciting a wide variety of fluorophores: $458\mathrm{nm}$ $\left(40\mathrm{mW}\right)$ , $477\mathrm{nm}$ $\left(185\mathrm{mW}\right)$ , $488\mathrm{nm}$ $\left(300\mathrm{mW}\right)$ , $514\mathrm{nm}$ $\left(300\mathrm{mW}\right)$ , $521\mathrm{nm}$ $\left(120\mathrm{mW}\right)$ , $531\mathrm{nm}$ $\left(240\mathrm{mW}\right)$ , $568\mathrm{nm}$ $\left(180\mathrm{mW}\right)$ , and $647\mathrm{nm}$ $\left(300\mathrm{mW}\right)$ . For stability during the experiments, the laser is operated in a constant power mode. The laser aperture is set to the optimal value to ensure ${\mathrm{TEM}}_{\mathrm{oo}}$ mode for each wavelength. To properly feed the $1.8\mathrm{mm}$ diameter of the laser beam into the optical fiber (core diameter $\sim 5\mathrm{\mu }\mathrm{m}$ ), a downsizing telescope and a user friendly alignment system consisting of an $xy$ mount combined with a KineFlex pitch and yawl mount for the optical fiber (Point Source, United Kingdom) was designed. In front of the laser, an attenuator compensator (Newport, Irvine, California) was placed to allow fine continuous control of the laser power throughout the tunable range.

During operation, the laser head must be water cooled at a flow rate of $2\mathrm{gal}∕\min$ with a water temperature range of 10 to $35°\mathrm{C}$ . To ensure proper cooling, a heat exchanger (Spectra Physics, Mountain View, California) was placed remotely from the instrument room due to the amount of noise that is created. A dedicated closed-loop cooling system was constructed, and a remote control switch was installed in the room. The heat exchanger is coupled to the chilled water system of the building.

The alignment procedure of the laser beam is always performed at low power for safety and to minimize the risk of damage to the optical fiber. The entire open beam path is enclosed in a 1/4-in. black Plexiglas box. During laser alignment at the microscope level, a U-shaped screen of the same material is mounted on top of the microscope, enclosing the exposed laser beam, and several shutters are provided along the beam path. Blocking filters (Chroma Technology, Rockingham, Vermont) for the AFM laser diode light are mounted in front of both CCD cameras and in the eyepieces. In front of the video camera is placed another filter (Digital Instruments, Santa Barbara, California) that partially blocks the laser diode to allow a very dim spot to be seen on the computer screen for AFM alignment purposes. It is of great importance to not allow the light of the laser diode to reach the CCD cameras, because the level of light is much higher than the dim fluorescence coming from the sample. Use of the eyepieces during TIRF or AFM experiments is not recommended.

5.1.

Sample Preparation

5.1.3.

Labeling of atomic force microscopy probes

For imaging and force spectroscopy experiments, unsharpened silicon nitride cantilevers purchased from ThermoMicroscopes (Sunnyvale, California) were used [Fig. 8(a) ]. For both types of measurements, the largest and softest triangular cantilever from a set of five on the cantilever chip was used (spring constant $15\pm 1\mathrm{pN}∕\mathrm{nm}$ ). For adhesion force measurements, the probes were coated⁶⁹ with 1-mg/ml fibronectin (FN). Polyethylene glycol (PEG, Sigma, Saint Louis, Missouri) $10\mathrm{mg}∕\mathrm{ml}$ was used to cross-link fibronectin onto probes at room temperature.⁷⁰ After the tip was mounted on the glass holder and washed, it was incubated with PEG for $5\min$ , washed five times with deionized water, and then incubated for $1\min$ with fibronectin. The tip was then washed again five times with phosphate buffered saline (PBS) and mounted on the AFM scanner. The coating was performed only at the very end of the cantilever to avoid altering its spring constant, assumed to be unchanged after protein labeling.

Fig. 8

Scanning electron microscope images of AFM cantilevers. (a) Typical AFM probe for force measurements (magnification $45\times$ ). The cantilever used in our experiments is the longest one $\left(320\mathrm{\mu }\mathrm{m}\right)$ with a spring constant of $\sim 0.015\mathrm{N}∕\mathrm{m}$ . At the very end of the cantilever is a pyramidal shaped tip with a $4\mathrm{\mu }\mathrm{m}$ base, 35-deg half angle of the tip, and a radius of $50\mathrm{nm}$ (inset, magnification $16,000\times$ ). (b) Typical AFM probe for mechanical stimulation measurements (magnification $10,000\times$ ). The cantilever is V-shaped, with a spring constant of $\sim 0.06\mathrm{N}∕\mathrm{m}$ and a $2\text{-}\mathrm{\mu }\mathrm{m}$ glass bead attached at its end (inset, magnification $10,000\times$ ). Images were taken at the Microscopy and Imaging Center of Texas A&M University, College Station, Texas.

For mechanical cell stimulation, avidin (Sigma, Saint Louis, Missouri) was used to cross-link the Biotin-FN (Pierce, Rockford, Illinois) onto functionalized glass beads attached to silicon nitride cantilevers (spring constant $0.06\mathrm{N}∕\mathrm{m}$ , Novascan Technologies, Ames, Iowa) [Fig. 8(b)]. The probes were first incubated with avidin $(1\mathrm{mg}∕\mathrm{ml})$ for $5\min$ , washed with PBS five times, and then incubated with Biotin-FN for $5\min$ . The tip was washed again with PBS five times.

5.2.

Live Cell Topography and Force Spectroscopy Using the Atomic Force Microscope

One main applications of the AFM is imaging live cells in culture using contact mode operation. In Fig. 9(a) , an AFM image of a cultured VSMC is shown (apical surface). The acquisition of the AFM image took $35\min$ at a speed of $\sim 40\mathrm{\mu }\mathrm{m}∕\sec$ for $512\times 512$ pixel. The image of the cell exhibits many cytoskeletal stress fibers (e.g., actin) that can be visualized beneath the cell surface, and appear to be organized along the axis of the VSMC. These images can be further processed off-line for contrast enhancement of the cytoskeleton features of the cell. An IRM image of the basal surface of the same cell is shown in Fig. 9(b), featuring the focal adhesions where the cell was attached to the substrate. The IRM image was acquired after the AFM image using the Lambda LS lamp in combination with a 545/30 excitation filter and the polarizing cube. The overlap of the two images [Fig. 9(c)], where the AFM image was pseudo-colored in gray and the IRM image was pseudo-colored in rainbow fashion with blue being the darkest areas and red the lightest areas, shows a very good agreement between the focal adhesions (blue spots) and the cytoskeletal actin filaments that are terminating in the focal adhesions anchoring the cell to the substrate.

Fig. 9

A cultured VSMC was scanned by AFM in contact mode over the apical cell surface (a) while the IRM image of the same cell (basal surface) emphasizes focal adhesion areas (b). The overlapped image (c) shows the AFM image pseudo-colored gray and IRM image in rainbow with blue, indicating the darkest areas, and red the lightest areas. One can observe the actin filaments terminating in the focal adhesions. Image size is $46\times 82\mathrm{\mu }\mathrm{m}$ .

The adhesion forces between the $\alpha 5\beta 1$ integrin on the cell surface and fibronectin functionalized AFM tip were characterized by constructing histograms of the number of events detected at various forces (Fig. 10 ). The midpoints in each histogram bar were then connected to approximate the envelope of the distribution. The distributions were analyzed further by fitting the entire envelope with multiple Gaussian density curves using a deconvolution algorithm. The result shows good agreement between the experimental points (black squares) and the fitted envelope (solid black line). Three distinct Gaussian populations are apparent. The first distribution on the left (dashed line) is interpreted to represent the adhesion force of a single integrin ${\alpha }_5{\beta }_1$ -fibronectin bond having a value $28\pm 1\mathrm{pN}$ . This interpretation is suggested by the presence of a second (mid dashed line) and a third (right dashed line) distribution with peaks equally spaced at higher values of $57\pm 1\mathrm{pN}$ and $84\pm 1\mathrm{pN}$ , respectively. We suggest that these peaks correspond to the simultaneous rupture of two and three bonds, respectively. The force values obtained through this experiment are in good agreement with those reported in the literature.^{71, 72}

Fig. 10

Force spectroscopy analysis performed by simultaneous deconvolution of the experimental data (black squares) with Gaussian distributions to resolve the ${\alpha }_5{\beta }_1$ integrin-fibronectin adhesion forces (solid black line is the theoretical envelope of experimental points, left dashed line is the theoretical fit for single bond adhesion, and the middle and right dashed lines are the theoretical fit for simultaneous rupture of double and triple adhesion bonds, respectively). Arrows indicate the peaks positions.

5.3.

Total Internal Reflection Fluorescence and Internal Reflection Microscopy Imaging

Images of VSMC cultured on glass-bottom dishes and transiently transfected with GFP-vinculin are shown in Fig. 11 . As compared to the epifluorescence image [Fig. 11(a)], the TIRF image [Fig. 11(b)] has very low background fluorescence, showing more clearly delimited focal adhesions that exhibit high fluorescence due to vinculin aggregates. The TIRF image was acquired using laser illumination at 70-deg angle of incidence with an exposure time of $10\mathrm{ms}$ and a laser power of $\sim 8\mathrm{mW}$ . The Lambda LS lamp in combination with an FITC excitation filter was used to acquire the epifluorescence image. Both images were acquired using the same dichroic mirror and emitter filter. In the epifluorescence-TIRF overlapped image [Fig. 11(d)], it is easy to see that in some parts of the cell (nuclear area), the fluorescence signal is missing in the epifluorescence image, but is visible in the TIRF image. Based on the simplified TIRF theory,^{34, 73} it is possible to quantify relative separation distances at the cell-coverslip interface (data not shown). The IRM image of the same cell shown in Fig. 11(c) was acquired using the Lambda LS lamp in combination with a 545/30 excitation filter and the polarizing cube. When the IRM and TIRF images are overlapped, the discrete vinculin labeling appears pixilated over the darker spots in the IRM images for the same focal adhesion area [Fig. 11(e)]. The pixilation is due to the fact that only vinculin is labeled, whereas IRM likely shows all proteins in the focal adhesion area.

Fig. 11

VSMC transiently transfected with GFP-vinculin was imaged in (a) epifluorescence, (b) TIRF (pseudo-colored red) and (c) IRM. The TIRF image offers a very good contrast with no background. The IRM image offers an excellent way to visualize the attachment sites at the cell-coverslip interface due to the high contrast obtained from interference in monochromatic polarized light. The TIRF image was overlapped with the epifluorescence (d) to emphasize the differences between the two imaging modes (yellow represents good overlap between the two images). In the TIRF-IRM overlapped image (e), one can see the pixilation of GFP-vinculin over the focal adhesion contact area. Image size is $100\times 74\mathrm{\mu }\mathrm{m}$ .

Figure 12 presents an overlapped time-lapse TIRF image of a stationary VSMC in culture, transiently transfected with GFP-vinculin and color coded as follows: blue-time 0, green- $20\min$ , red- $120\min$ . One can see that during a two-hour interval, the focal adhesions undergo major reorganization. A focal adhesion was considered motile if it moved at least one focal adhesion length within $1\mathrm{h}$ . This set of images was acquired in a different TIRF experiment using a similar experimental setting as described before.

Fig. 12

Focal adhesions are motile in stationary VSMCs in culture. The overlapped image was formed by color-coding images from the time-lapse recording: $0\min$ (blue), $60\min$ (green), and $120\min$ (red). (Yellow represents an overlap of red and green). Image size is $85\times 90\mathrm{\mu }\mathrm{m}$ .

5.4.

The Atomic Force Microscope Mechanical Stimulation of Vascular Smooth Muscle Cells Induces Focal Adhesions Rearrangement Monitored in Real Time

Use of the AFM to mechanically stimulate the apical cell surface induces significant changes in cell shape and focal adhesion reorganization that can be monitored in real time using TIRF and IRM. The AFM is driven in contact mode imaging by fixing the probe at a chosen $xy$ coordinate on the cell surface, and applying a controlled upward movement of the cantilever in discrete steps ( $\sim 1\mathrm{nN}$ every 2 to $5\min$ ) over a period of time $(\sim 1.5\mathrm{h}$ ). The $z$ -axis movement of the piezoelectric element and the deflection signal from the cantilever are recorded in a series of images of $512\times 512$ pixels, each line in the image representing a 1-s unit of time. The “height” of that particular recording represents the height changes at the cell level at that location. Figure 13 shows IRM images of the same cell before [Fig. 13(a)] and after [13(b)] mechanical cell stimulation. The AFM probe consists of a $2\text{-}\mathrm{\mu }\mathrm{m}$ glass bead functionalized with FN that is brought in contact with the cell surface and kept there for $20\min$ . After this time, a strong focal adhesion is formed. On a controlled upward movement of the cantilever $\left(1\mathrm{V}\right)$ , the cell was exposed to $\sim 900\text{-}\mathrm{pN}$ upward force, and it responded by pulling back over $\sim 70\sec$ at a rate of $\sim 0.25\mathrm{nm}∕\sec$ . This response was systematically recorded after repeated upward movements of the cantilever. The IRM image acquired at the end of the experiment shows that the cell significantly modified its shape by retracting to oppose the mechanical stimulation (arrows). Also, focal adhesions were reorganized in a dramatic way: the darkest ones (closer to the substrate, potential higher protein density) organized around the point where the mechanical stimulus was applied at the apical cell surface (the white dashed lines represent the AFM tip position above the cell).

Fig. 13

The IRM images of the same VSMC (a) before and (b) after $\sim 1.5\mathrm{h}$ of AFM mechanical cell stimulation. The VSMC changes its shape and focal adhesions undergo a significant reorganization (arrows). The white dashed lines represent the position of the AFM tip, and the white dot represents the position of the glass bead on the cell body. Image size is $82\times 90\mathrm{\mu }\mathrm{m}$ .

5.5.

Forster Resonance Energy Transfer Imaging

Images of COS fibroblast cells cultured on glass coverslips expressing CFP linked with YFP are shown in Fig. 14 . FRET was detected in sensitized emission by exciting the sample in epi-illumination with light corresponding to the absorption spectrum of the donor, and detecting light emitted at the wavelengths corresponding to the emission of the acceptor. By using a Plan Apo $60\times \mathrm{NA}=1.45$ objective, a set of images was acquired for CFP only, YFP only (images not shown) and CFP-YFP construct (Fig. 14) by changing only the excitation filters in the Lambda LS lamp with the other dichroic mirrors and filters being kept in place. The bleedthrough coefficients for CFP and YFP into the FRET channel were 0.6 and 0.025, respectively. It is known that CFP has low absorption and quantum yield, whereas YFP has high absorption and quantum yield, but is more susceptible to photobleaching than CFP.⁵⁸ Photobleaching can alter the donor-acceptor ratio,⁶⁰ and therefore the value of FRET signal [Fig. 14(c)]. To avoid acceptor photobleaching, the images were acquired at an exposure time optimized for the acceptor.

Fig. 14

Synthesized FRET imaging of COS fibroblast cells expressing CFP - YFP construct. The first row presents (a) the CFP signal into the donor channel, (b) YFP signal into the acceptor channel, and (c) FRET signal into the FRET channel. The second row presents (d) the corrected FRET signal and (e) the normalized FRETc in respect to the donor channel. All FRET images are displayed in rainbow pseudo-color. The color bar is scaled in arbitrary units of fluorescence intensity. Image size is $125\times 155\mathrm{\mu }\mathrm{m}$ .

An algorithm for considering the spectral bleedthrough between channels and a background correction implemented into Slidebook software (Intelligent Imaging Innovations, Denver, Colorado) according to Gordon ⁵⁹ allows for calculating the corrected FRET (FRETc) [Fig. 14(d)]. The intensity of the uncorrected FRET image is less than the corrected one FRETc, because the spectral bleedthrough components and the background were eliminated from the uncorrected FRET. By normalizing FRETc in respect to donor channel, a normalized FRETN image is obtained [Fig. 14(e)]. The set of CFP/YFP samples was a gift from Dr. Scott Young of Leica Microsystems (Exton, Pennsylvania).