1. Introduction

Fluorescence is widely used in biological immunoassays, optical devices, cell and tissue imaging, and medical diagnosis. In total internal reflection fluorescence (TIRF) microscopy, an incident light ray with an incident angle greater than the critical angle, is used to generate an evanescent field to excite and image fluorophores on or very near (i.e. within 100 nm) the liquid/solid interface. Experimental evidence suggests that the molecular interactions that occur on or near the surfaces of cell membranes (typically within 50 nm) have different properties from those which occur in bulk solution [1]. To visualize these events within the cell-substrate contact region more clearly, it is desirable to use observation techniques that suppress the interference arising from background fluorescence [1]. Fluorophores located within or close to the plasma membrane are excited by the shallow evanescent field, resulting in images with very low background fluorescence and no out-of-focus fluorescence. TIRF microscopy is particularly suited to the investigation of living cells since it causes minimal photodamage. Furthermore, TIRF microscopy provides a complementary approach that can be readily combined with other microscopy techniques [2]. Therefore, TIRF microscopy has been widely used in studies of cell-substrate contact regions [3], protein dynamics [4], endocytosis or exocytosis [5], and membrane-associated photosensitizers [6]. However, the scattering of one-photon TIRF for bio-imaging applications causes excitation photons to leak into deeper cell regions, resulting in the fluorescence image being excited by both near- and far-field components [7]. The confinement of one-photon TIRF excitation is gradually loosed in its propagation direction, and therefore the biological fluorescent image is generated by the excitation of the superposition of an evanescent field and a scattered light component [8].

In recent investigation, a two-photon TIRF image close to a cell-substrate interface excited by a femtosecond infrared laser has been demonstrated [8]. Benefiting from the nonlinearity and the longer excitation wavelength of multiphoton excitation and the spatial inhomogeneity of the evanescent field, the two-photon TIRF excitation can largely decrease the out-of-focus fluorescence excited by scattered light. Furthermore, two-photon fluorescence excitation represents the quadratic intensity dependence. Even though the two-photon fluorescence excitation utilizes a longer wavelength, the effective evanescent field penetration depths of one and two-photon TIRF excitation are comparable. Therefore, compared with one-photon TIRF excitation, two-photon TIRF excitation obviously increased the signal-to-noise ratio (SNR) by repressing background fluorescence [9]. In addition to constraining the background fluorescence from scattered excitation and restricting the fluorescence excitation to a confined excitation volume, two-photon TIRF excitation also offers the advantages of the excitation of several fluorophores simultaneously as well as better rejection of excitation light [9]. Recent investigations have also included pattern photobleaching [10], enhanced fluorescence excitation by optical waveguide modes [11,12], evanescent wave excited second harmonic generation [13,14], and two-photon fluorescence scanning near-field microscopy based on a focused evanescent field [15]. These studies demonstrate the experimental feasibility of two-photon TIRF imaging for chemical and biological samples. However, the fluorescence signal in live cell imaging is still in need of being enhanced if dynamic images of the molecular interactions, which take place on or near the surface of the cell membrane, are to be acquired. As of this point, improving fluorescence sensitivity to reduce the detection limit is still required to eventually achieve single molecular detection [16,17].

Aside from utilizing two-photon TIRF excitation for suppressing background noise [17],the SNR improvement of fluorescent signals can be achieved with surface plasmon (SP) enhancement. Our previous study has shown that the SP-enhanced one-photon TIRF microscopy can increase the brightness and acquisition frame rate of live cell membrane images [18]. In recent investigations for plasmon-enhanced one-photon fluorescence, metallic surfaces or particles have been exploited to enhance the emission of fluorophores in order to develop more efficient fluorescence immunoassays [19–21] and light scattering of gold nanoparticles for cellular-based diagnostic imaging [22,23]. It has also been reported that the surface plasmon-enhanced (SPE) technique can be employed for enhanced two-photon fluorescence, directional two-photon induced SP-coupled emission, and local second-harmonic generation enhancement [24–27]. In these studies, SPs or localized SPs on the metallic surfaces or nanoparticles enhance the local electromagnetic (EM) field around the fluorophores, and therefore increase the intensity of the detected fluorescence signal [17].

SPs are oscillations of the free electrons located on the surface of a metal film and can be excited by incident light based on the prism-coupled attenuated total reflection (ATR) or grating-coupled diffraction methods. When the wave number of an incident transverse magnetic (TM) wave matches that of the excited SPs, the so-called surface plasmon resonance (SPR) phenomenon occurs and the local EM field associated with the SPR is greatly enhanced [28]. In the Kretschmann SPR configuration, the ATR method is employed to excite the SPs on a thin metal film deposited on the surface of the base of a prism [29]. In the interaction of the SPs with fluorophores located within a very short metal-fluorophore distance (approximately d < 10 nm), the dipole field is dominated by the near field, the strength of the fluorescence is quenched, and the excitation fluorescent energy dissipates into the metal in the form of heat [30]. For intermediate separation distances, the emission fluorescence can be effectively coupled back to the SPs of the metal surface by matching the momentum of the fluorophores with that of the SPs. Subsequently, the emission SPs are re-radiated into the glass prism with a hollow cone of intense light around angles near the SPR angle [31]. In one previous study, however, it had been shown that the SP-coupled emission did not lead to any improvement, but the metal film actually reduced the sensitivity of fluorescence detection [32]. The result related to the fluorescent emission rate is therefore still inconsistent. For larger separation distances, the fluorescent emission rate can be predicted by considering the classic mechanism of a dipole, which is influenced by the back-reflected field caused by its dipole image [33]. Therefore, through the moderate modifications of a metal film or particles on a substrate and the metal-fluorophore distance, the SPs obviously contribute to the increase in the quantum yield for achieving a brighter fluorescent signal, a reduction in the fluorescence lifetime for attaining better photostability, and a specific orientation in the typically isotropic emission for providing more information. These effects are not due to the reflection of the emitted fluorescence, but rather as the result of the fluorophore dipole interacting with free electrons in the metal [34,35].

Previously reported SPE two-photon fluorescence measurement techniques were designed only for single-spot sensing applications. However, this study proposes a SPE two-photon fluorescence capable of imaging the molecular interactions near the cell membrane in real-time. A candidate tumor suppressor WW domain-containing oxidoreductase, known as marine WOX1 or WWOX, is used in this study [36]. WOX1 proteins play an important role in the regulation of a wide variety of cellular functions such as protein degradation, transcription, and RNA splicing [37]. Therefore, this study employs the proposed SPE two-photon TIRF microscopy to dynamically image the interactions of proteins near the cell membrane. The experimental results demonstrate that the live cell membrane images provided by the SPE two-photon TIRF microscopy not only clearly reveal higher SNR compared to those of one-photon TIRF excitation, but are also approximately 30-fold brighter than those of the two-photon TIRF approach with an equivalent image quality.

2. Materials and methods

2.1. Optical system setup

Figure 1(a) shows the optical configuration of the proposed prism-coupled SPE two-photon TIRF microscope, in which the ATR method is used to excite the SPs in order to enhance the local EM field and to increase the SNR of the fluorescence. In this configuration, a thin silver film is sequentially deposited via a sputtering deposition process, a chemical self-assembly monolayer (SAM), and a biomolecular layer, onto an SF-11 slide in accordance with an optimal design that is known to enhance the intensity of the fluorescence and, therefore, increase the acquisition frame rate [18]. In order to avoid the decay of the local EM field enhancement, an additional chromium layer, that ensures silver adhesion, wasn’t deposited on the SF-11 slide. As shown in Fig. 1(a), the beams of a 10-W solid-state pumped mode-locked femtosecond Ti:Sapphire laser (Tsunami, Spectra-Physics; ~100 fs pulse at 80 MHz, λ = 800 nm) and a diode-pumped solid-state laser (20 mW, λ = 473 nm) pass initially through a half-wave plate and a polarizer to adjust their intensity and polarization. Both beams are then focused by a convergence lens (f = 70 mm), pass through a coupled SF-11 hemispherical prism, a layer of index-matching oil, and then the SF-11 slide, leading finally to be incident directly upon the interface between the slide and the thin silver film. The area of laser illumination is controlled to be approximately 100 × 100 μm², which is comparable to the size of general cells. Using the hemispherical prism, the angle of the incident light can be easily adjusted in order to vary the penetration depth of the evanescent field into the thin silver film (thickness 50.0 ± 1.0 nm). In the current Kretschmann configuration, the SF-11 hemispherical prism is used to increase the wave number of the incident light and to decrease the SPR angle, θ_spr, i.e. the angular position of the dip in the reflectivity spectrum. The fluorescence from the cell membrane excited by the SPs or the evanescent wave is wide-field collected and imaged from the reverse side of the prism by an immersion water objective (100×, N.A. = 1.0, Olympus) dipped into the flow cell, and is then directed through a short-pass filter (SPF, λ < 680nm, Semrock) and a long-pass filter (LPF, λ > 515 nm, Semrock) into a high-speed frame rate CCD camera (iXon DV885, Andor).

 

Fig. 1. (a) Schematic illustration of experimental configuration employed for live cell membrane imaging using conventional and SPE TIRF microscopes with one-photon and two-photon excitations of the same fluorophores. (b) Conventional TIRF chip: cell is cultured on collagen-coated slide modified with silane. (c) SPE TIRF chip: cell is cultured on collagen-coated silver thin film modified with thiol.

Download Full Size | PDF

2.2 Cell culture protocol

In this study, cultured monkey kidney COS-7 fibroblasts were suspended in a serum-free culture medium, containing 2 mg/ml bovine albumin, and electroporated with EGFP-WOX1 and EYFP-MEM constructs (BTX ECM 830 Electroporator, Genetronics; 5 μg DNA/3×10⁶ cells, 220 volt and 50 msec). Albumin enhances both the transfection efficiency and gene expression 3–5 fold [38]. These cells were cultured 24–48hrs, and then observed by the proposed prism-coupled SPE two-photon TIRF microscope.

In order to compare the fluorescence intensity of a conventional TIRF chip with that of the proposed SPE TIRF chip, the cells were added to collagens immobilized by chemical SAMs on a naked SF-11 slide and on a silver thin film SF-11 slide containing a pre-warmed medium, respectively. Figure 1(b) shows cells cultured on a collagen-coated slide modified with a silane SAM. In fabricating this conventional TIRF chip, the naked SF-11 slide was immersed in 20% (3-aminoproply)triethoxysilane solution in order to form a dense SAM on its surface. To immobilize the protein collagen, covalent activation was conducted by immersing the chip in a solution containing EDC[N-(3-dimethylaminopropyl)-N’-ethylcarbodimide hydrochloride, 2 mM] and NHS(N-hydroxysuccinimide, 5 mM) for 6hrs. In Fig. 1(c), cells are cultured on a collagen-coated silver thin film modified with a thiol SAM. In developing this SPE TIRF chip, the metal film was immersed in 1 mM 2-aminoethanethiol hydrochloride solution to form a dense SAM on its surface. As the TIR chip, covalent activation was then performed by immersing the chip in a solution containing EDC[N-(3-dimethylaminopropyl)- N’-ethylcarbodimide hydrochloride, 2 mM] and NHS(N-hydroxysuccinimide, 5 mM) for 6hrs to immobilize the protein collagen [18]. One prior study for detecting fluorescent erythrocytes on metal surfaces to induce the quenching effect with a similar optical system setup and cell culture protocol can be referred to in [39].

2.3 Excitation of surface plasmons by incident light

SPR is an optical phenomenon in which an incident TM wave excites SPs, i.e. a surface charge density oscillation, at the interface between a thin metal film and a dielectric sample medium. SPR occurs when the parallel component of the wave vector of the incident TM wave, k_x , matches the wave vector of the SP traveling over a semi-infinite structure, k _sp ⁰,i.e.

where ε ₀, ε ₁, and ε ₂ are the wavelength-dependent complex dielectric constants of the prism, the metal layer, and the dielectric sample, respectively, and θ is at the SPR angle. As shown in Fig. 1(c), the SPR optical configuration basically consists of a prism, a thin metal film, and an analytic layer on which immobilized analytes are used to probe the target receptors [40]. The SPR angle depends on the dielectric constant and thickness of the nanolayer over the sensing metal layer. Simplifying the SPR configuration to the current three-layer 0/ 1/ 2 system, the optical coupling of the SPE sensor can be analyzed by calculating the reflectivity, R ₀₁₂, using the Fresnel equation, i.e.

where ε_i and $k_{\mathrm{zi}}={\left[{\epsilon }_i{(\omega /c)}^2-k_x^2\right]}^{1/2}$ are the dielectric constants and the wave vector components perpendicular to the interface in medium i, respectively, and d ₁ is the finite thickness of the metal film.

Gold and silver are ideal candidates for the metal film in the visible light region. Gold is sometimes a more suitable choice for bio-applications due to its stability under chemical reaction [40]. However, silver can provide a superior EM field enhancement in the blue light (such as λ = 473 nm for one-photon excitation) and near infrared (λ = 800 nm for two-photon excitation) regions [28]. Moreover, we can’t observe an obvious enhancement of the fluorescent signal by replacing the thin silver film with gold in our experiments. Therefore, silver was adopted to demonstrate the enhancement of SPs in this study. In addition, the silver thickness in the Kretschmann configuration is an important parameter for the optimal EM field enhancement. Figure 2plots the enhancement factor of the electric field intensity at the interface between the collagen and the buffer in the SPE one-photon/two-photon TIRF chip as a function of silver thickness. The solid line and dashed line indicate the excitation wavelengths at 473nm and 800nm, respectively. The silver thicknesses with the highest enhancement for the SPE one-photon and two-photon TIRF excitations are approximately 51 nm and 48 nm, respectively, and the corresponding enhancement factors are 63.7 and 104.1. Therefore, the thickness of the silver film adopted is approximately 50.0 nm, which is suitable for the both SPE one-photon and two-photon TIRF excitations.

 

Fig. 2. Plot of the enhancement factor of the electric field intensity at the interface between the collagen and buffer in the SPE one-photon/two-photon TIRF chip as a function of silver thickness. Solid line: 473 nm excitation; Dashed line: 800 nm excitation.

Download Full Size | PDF

 

Fig. 3. (a) Solid line: reflectivity spectrum of two-photon TIRF chip as function of incident angle; Dashed line: enhancement factor of electric field intensity at interface between collagen and buffer. (b) Enhancement factor distribution is perpendicular at sensing interface at incident angle of 51.72°.

Download Full Size | PDF

 

Fig. 4. (a) Solid line: reflectivity spectrum of SPE two-photon TIRF chip as function of incident angle; Dashed line: enhancement factor of electric field intensity at interface between collagen and buffer. (b) Enhancement factor distribution is perpendicular at sensing interface at incident angle of 51.72°.

Download Full Size | PDF

Figure 3(a) plots the reflectivity spectrum of the two-photon TIRF chip as a function of the incident angle. In this figure, the solid line represents the simulated reflectivity spectrum, while the dashed line indicates the enhancement factor of the electric field intensity at the interface between the collagen and the buffer. In the two-photon TIRF chip, the thicknesses of the collagen and silane layers are approximately 5.5 nm and 1.0 nm, respectively, and the corresponding refractive indexes are 1.46 and 1.44. Figure 3(a) shows that the maximum enhancement factor of the electric field at the interface between the collagen and the buffer is 6.3 and occurs at a critical angle of 49.02°. Meanwhile, Figure 3(b) shows that the distribution of the enhancement factor of the electric field intensity is perpendicular to the interface of the two-photon TIRF chip at an angle of 51.72°. The maximum value of the enhancement factor is 1.9. Figure 4(a) shows the simulated reflectivity spectrum (solid line) and the enhancement factor of the electric field intensity at the interface between the collagen and the buffer (dashed line) in the SPE two-photon TIRF chip. The thickness and refractive index of the thiol layer are 1.1 nm and 1.49, respectively. Also, the thickness and the refractive index of the silver layer are 50.0 nm and 0.14 + j5.29 (λ = 800 nm), respectively. In Fig. 4(a), the angle associated with the maximum enhancement of the electric field intensity, i.e. 51.72°, is slightly lower than the SPR angle at 51.84°. Figure 4(b) shows that the distribution of the enhancement factor of the electric field intensity is perpendicular to the interface of the SPE two-photon TIRF chip at 51.72°. The enhancement factor at the interface between the collagen and the buffer has a value of 103.3. In other words, the local field enhancement of the SPE two-photon TIRF chip (103.3) is theoretically 54.4 times higher than that of the two-photon TIRF chip (1.9). Local EM field enhancement leads to an increased excitation rate, and is therefore expected to enhance the fluorescence intensity. However, the interaction of fluorophores with SPs over a short metal-fluorophore distance occurs, and hence the fluorescence quenching effect is counted [30].

3. Experimental results and discussions

Figures 5(a) and 5(b) present one-photon and two-photon TIRF images of the living COS-7 cell transfected with the EYFP-MEM construct cultured on the collagen-coated TIRF chips. The excitation wavelength of two-photon TIRF is roughly twice compared with one-photon TIRF, but the effective excitation depths of both evanescent fields are comparable due to the two-photon fluorescence according to the quadratic EM field intensity. Moreover, two-photon TIRF has a lower scattering due to its longer excitation wavelength and a less sensitive excitation to scattering photons due to nonlinear excitation; therefore, the absence of scattered excitation for the two-photon TIRF results in a low-background-noise and high-contrast living cell membrane image compared with one-photon TIRF (Fig. 5). Specifically, the strong scattering fluorescence located in the right hand side of Fig. 5(a) decreases the SNR of the one-photon TIRF image according to the wave propagation from left to right.

Figures 6(a) and 6(b) show the SPE one-photon and two-photon TIRF images of the living COS-7 cell transfected with the EYFP-MEM construct cultured on the collagen-coated SPE TIRF chips. The exposure times of the images in Figs. 5(b) and 6(b), i.e. without and with SPE, are 0.5 seconds and 0.1 seconds, the illuminating laser powers are 50 mW and 25 mW, while the corresponding fluorescence intensity scales are 950 and 3100 (arbitrary unit), respectively. In comparison with images from the two-photon TIRF, the SPE two-photon TIRF requires about one fifth of the exposure time and half the illuminating laser power, while producing a 3-fold gain in the fluorescence intensity. Therefore, by enhancing the local EM field excited by using the ATR method, the SPs introduce a gain 30 times higher in the SPE two-photon TIRF image (Fig. 6(b)) than that provided by the two-photon TIRF image (Fig. 5(b)). The greater the fluorescence intensity and the higher the SNR value of the image, the shorter the required exposure time and hence a faster frame rate which can be achieved. In addition to enhancing two-photon fluorescence by utilizing the SPE two-photon TIRF microscopy, the excitation volume can be efficiently confined by the nonlinear two-photon excitation and the SP wave (Fig. 6). The cells were usually observed after 30 minutes of SPE two-photon exposure, which revealed that their morphologies and fluorescence intensities were only slightly changed. Therefore, the cell can be expected to survive under the microscopy. Undoubtedly, the SPE two-photon TIRF microscopy provides live cell membrane images with the highest SNR.

 

Fig. 5. Live cell membrane images by utilizing (a) one-photon and (b) two-photon TIRF. The evanescent waves propagate from left to right for both figures.

Download Full Size | PDF

 

Fig. 6. Surface plasmon-enhanced (a) one-photon and (b) two-photon TIRF live cell membrane images. The SP waves propagate from left to right for both figures.

Download Full Size | PDF

The simulation results presented in Figs. 3(b) and 4(b) suggest that an approximately 50-fold enhancement of the local field intensity at the interface between the collagen and the buffer can be obtained in the SPE two-photon TIRF chip. The two-photon fluorescence intensity of fluorophores in free space is theoretically possible according to the quadratic field intensity [24]. However, the experimental results presented in Figs. 5(b) and 6(b) show that the overall fluorescence intensity enhancement of the fluorophores on the silver surface is only 30 times in practice. The two-photon fluorescence intensity enhancement of the fluorophores on the metal surface is at least based on three factors: 1) the quadratic dependence of the field intensity enhancement via SPs; 2) the emission efficiency of the fluorophores on the metal surface; and, 3) the collection efficiency due to SP coupling. The emission characterization of the fluorophores on the metal surface can be described as dipole vibrations. Due to the thin film structures, molecular vibration will be affected, resulting in the emission characterization being changed [43]. Furthermore, the SPE fluorescent signal can show distance-dependent characteristics; therefore, the quenching efficiency varies as a function of the distance between the metal film and the fluorophores [16]. Moreover, the detection of a fluorophore is usually limited by its quantum yield and photostability [10]. The two-photon fluorescence intensity (quantum yield) is not only determined by the quadratic dependence of the excitation SP field intensity, but also by the quenching influence of the molecular dipole-metal interaction [21]. To compensate for the competing effects, a spacer that is 6.1 nm thick is placed between the silver layer and the fluorophore-containing cell membrane and is used to reduce the effect of metallic surface-induced fluorophore quenching at very short distances (approximately 0~5 nm) [33]. However, the key issue of the detection in fluorescence microscopy is not merely how to increase the quantum yield, but rather how to collect as many photons from fluorophore as possible before photobleaching occurs. Although the quenching effect reduces the fluorescence emission, it shortens its lifetime to enhance fluorophore photostability [28]. Therefore, the quantum yield and photostability of fluorophores should be modified and controlled in order to improve the detection limit [10,28]. This study has demonstrated that the proximity to metallic surfaces within 10 nm can increase the two-photon fluorescence intensity more than 30 times. In contrast to the relatively constant radiative decay rate via an increase of incident laser power, the radiative decay rate can be increased by placing the fluorophores at suitable distances from metallic surfaces. An increase in the radiative decay rate results in an increased fluorescence intensity and a reduction in lifetime [16]. Therefore, the fluorescence intensity is enhanced and the fluorophore photostability is improved by the excitation SPs.

In order to delve more deeply into the local field enhancement via SPs, the power of the femtosecond laser was increased to 100 mW. Figures 7(a) and 7(b) show the images of the living COS-7 cell transfected with the EGFP-WOX1 construct before and after high-power femtosecond laser illumination with 100 mW based on the SPE two-photon TIRF microscopy, respectively. When the power of the femtosecond laser was increased, the photodamage phenomenon became more serious and a nucleus explosion-like disruption of the living COS-7 cell was observed. The cellular material inside the nucleus spread into the cytoplasm of the cell. The high internal pressure of the nucleus is induced by the radiative local EM field and non-radiative thermal heating with the SP enhancement. It has been reported that photo-thermal damage caused by two-photon excitation can be avoided by reducing the laser repetition rate to less than 1 MHz to decrease the thermal accumulation [41]. An acousto-optic modulator (AOM) as a pulse selector was inserted after the Ti:Sapphire femtosecond laser in Fig. 1(a) to choose a fraction of pulses from the pulse train and also adjust the pulse peak power. The preliminary experimental results with the laser repetition rate of 800 kHz reveal that we can maximize the efficiency of two-photon fluorescence excitation and decrease the heat accumulation around the metal film by regulating the AOM. However, a femtosecond laser with high repetition rate could be assisted by the SP enhancement to achieve plasmonic photothermal therapy [42]. More theoretical modeling and experimental trials in the plasmonic heating process should be investigated on the SPE two-photon TIRF microscopy in the near future.

 

Fig. 7. Live cell images: (a) before, and (b) after high-power femtosecond laser illumination.

Download Full Size | PDF

4. Conclusions

This study has developed a SPE two-photon TIRF microscopy for the real-time investigation of live cell membrane images of monkey kidney COS-7 fibroblasts transfected with the EYFP-MEM or EGFP-WOX1 construct. The experimental results, consistent with the simulation, have shown that compared to conventional TIRF microscopy, which is based on one-photon evanescent wave excitation only, the additional excitations provided by the two-photon and the SPs in the SPE two-photon TIRF microscopy not only improve the SNR but also increase the brightness and acquisition frame rate of live cell membrane images. In sum, the approximate 30-fold enhancement of fluorescent intensity and simultaneous achievement of a high quality image are clearly superior to the conventional technique.