1. Introduction

Since the first reports by Ashkin [1], optical trapping has demonstrated the potential to revolutionize live cell imaging. However, despite the advances in optical trapping techniques, there remain limitations to incumbent trapping methods which unfortunately restrict the range of possible applications. Conventional optical trapping requires a high numerical aperture (N.A.) lens (typically N.A.>0.9) to generate the necessarily high-magnitude optical force gradient to overcome scattering forces [1–4]. By design, high numerical aperture lenses also have high magnification and this imposes restrictions upon the field of view and focal depth [5,6]. Consequently, this limits the application of optical trapping to small volumes. Although inspection of biological cells using high N.A. lens undoubtedly has scientific merit, visualization of biological systems beyond single-cell interactions requires larger fields of view (e.g. millimetre scale). Furthermore, high N.A. lenses are unsuitable for trapping large (>10µm) diameter cells such as T cells [7] due to the small focal spot size.

There are a few optical trapping methods that make use of lower N.A. lenses to visualize a wider field of view. For example, fixed dual beam fibre trap technologies [8,9] can allow larger fields of view but such systems require sophisticated laser routing and fibre coupling, which adds to the complexity of the technique. Further alternative systems based on spatial light modulators such as that by Eriksen et al., [10] also offer some improvements in wide field trapping by employing multiple-spot trapping around large cells, but such approaches require pre-encoding of phase patterns and hence on-the-fly manipulation is impossible. We report a point-and-click optical manipulation and imaging method and system that combines weak optical forces and a small temperature gradient to enable simultaneous trapping, manipulating and imaging of live T cells with a single low magnification objective lens (10×/0.4 N.A.). Using this approach, individual T cells were transported at an average speed of 1.2 ± 0.2 µm/s within a static field of view exceeding 1mm², with minimum perturbation to the cells. This advantageous technique can enable the non-invasive manipulation of cells such as T cells in a large population, where the prevalence and behaviour of cells can provide answers to fundamental questions concerning the function of the immune system.

Such a technique could also enhance optically actuated cell sorting [11], where micro-grid array systems can be employed in conjunction with such optical trapping and sorting systems to confine highly motile cell types to study cell communication [12–14]. Due to the high magnification, high N.A. lenses normally required for such a technique, trapping large diameter cells is not possible and the field of view is severely restricted, thus precluding high-throughput studies. Overcoming these limitations and enabling the application of larger micro-grid systems would facilitate high-throughput manipulating, screening and sensing of live cells, their function and interactions using ‘lab-on-a-chip’ array devices. Consequently, such a method would prove extremely useful for studying cell-cell interactions with minimum perturbation to the sample.

2. Trapping mechanism with a low N.A. lens

As described by the ray optics model by Ashkin [1], one must consider the vectorial nature of the optical gradient force F _g and the scattering force F _s upon a trapped object As well as considering the refractive indices of the material to be manipulated and the surrounding suspension media, the angles of incidences and refraction are largely dictated by the N.A. of the objective lens used in the experiment and as the trapped object deviates from the point of equilibrium, the trapping efficiency decreases. This holds true for high [1] and low [4] N.A. lenses, but the trap efficiency is customarily higher with the larger range of convergence angles from the high N.A. lens. However, as described previously there are many biological studies where low magnification, low N.A. lenses could prove advantageous for optical trapping, particularly where large volumes are concerned. We can consider a system employing a 10×/0.4 N.A. air immersion lens (n ₁=1) and a P=6 mW average power laser incident upon T cells (Fresnel reflection coefficient of approximately R=5% [15], loaded with a low concentration of fluorescent indicator CFSE inducing <10% absorption) and prepared in a typical buffer medium (RPMI), From analysis of the vector-summing of the contributions of all rays with convergence angles ranging from zero through to the maximum possible using the 0.4 N.A. lens as per the well-known model of Ashkin [1], optical trapping cannot occur and hence additional support from another mechanism is required.

Thermal gradients induced by the laser have been demonstrated previously for temperature-dependent optical trapping, albeit of microspheres. As described by Mao et al [16], an infrared laser source can initiate a thermal gradient force in a sample through absorption. This thermal gradient force, which varies laterally with intensity across the laser beam profile at half-width-half maximum, is described by:

where P is average power, α is the absorption coefficient of the buffer medium and K is the thermal conductivity of the buffer medium. The beam radius is r while r’ is the outer beam radius, i.e. immediately adjacent to the beam where the thermal gradient commences. In a typical experiment as described above with P=6 mW to trap live cells in PBS buffer, α _w ~50 m^-1, K=0.6 W/m.K, r=1 µm and r’=10 µm [16]. This yields a weak thermal gradient of 0.5 ˚K, which scales well with thermal increases reported in multi-photon laser scanning microscopy applications [17], where increases in temperature of less than a few ˚K were reported for similar conditions. The buffer medium was chosen specifically as change in conductivity and absorption properties in RPMI do not change significantly with wavelength in the near-infrared spectral region. As a consequence the trapping beam wavelength was chosen only with reference to water absorption [18].

We can then consider the effect of this thermal gradient to calculate the thermal force F _T acting upon a live T cell in buffer as a consequence of laser interaction. As first described by Gabriel Stokes, the thermal drag coefficient of a small isolated sphere in a viscous medium is given by [19]:

In our study, the sphere is a T cell of diameter d=10 µm and the viscosity of the buffer medium η=65.2 mPa.s [20], hence γ=1.22×10^-5. The thermal force resulting from laser irradiation can be expressed as [21]:

where k _B is the Boltzmann constant and T is temperature. In our model, where T=310.65˚K, the thermal force acting upon the T cell is F _T=0.65 pN. If we then consider this small thermal force contribution in addition to the vectorial forces F _g and F _s, the total force available overcomes the scattering force to facilitate trapping with a low magnification, low N.A. lens.

3. Experiment

 

Fig. 1. Cartoon representing the experimental configuration.

Download Full Size | PDF

Our experiment was designed to perform and validate temperature-assisted optical trapping of live T cells based on a single low magnification, low N.A. lens. The system (setup as shown in Fig. 1) was based on an inverted microscope (Olympus I×81) coupled to an Olympus FV1000 scanhead. This scanhead was modified to contain two independently controlled scanning mirror galvanometer systems.

Two laser sources were used in this study. The first, a continuous wave, fibre-coupled Ar+ laser with a pre-attenuation average power of <5mW was used for confocal imaging. This was directly coupled to the primary scanhead. The second laser, a 100 fs pulsed Ti:Sapphire laser (Mai-Tai, Spectra-Physics) propagating in free-space, served as the trapping beam with a maximum average power of 1.8 W measured at a wavelength of 780 nm. The source was heavily attenuated using variable neutral density filters control the average power at the sample and a laser wavelength of 780 nm was chosen to minimize reflections from the cells and hence increase F _g. The Ti:Sapphire laser was coupled into the secondary scanning system via a second port. The average power at the back aperture of the lens was <1 mW for the Ar+ laser and 8 mW for the Ti:Sapphire source respectively for the first part of the experiment. The variable neutral density filter immediately following the Ti:Sapphire laser was also used in the latter part of the study to determine the effect of power scaling upon the trapping capability of the system and to evaluate cell viability upon trapping.

A single 10×/0.4 N.A. objective lens air-immersion lens (Olympus, UPlanSApo) with a working distance of 3.1 mm was used in the experiment to simultaneously trap and confocally image cells, through which an average trapping power of 6 mW from the Ti:Sapphire source was provided at the sample. As calculated previously, this corresponded to a thermal gradient of 0.5 ˚K, which was deemed sufficient to overcome the scattering forces to assist the trapping process, but which would importantly not cause unnecessary thermal overloading to the cell which could potentially cause damage.

The sample was placed on a heated stage providing an ambient temperature of 310.15 ˚K, while the trapped cell was therefore held at a calculated temperature of 310.65 ˚K.

The first scanning system, for confocal imaging of the sample, operated independently of the ‘trapping’ scanning mirrors. The second scanning system comprised a simple computer-based point-and-click interface to enable simple movement of the trapping spot independently of the confocal scanner in the xy plane. This method of repositioning the location of the trapping spot required no movement of the microscope stage and therefore minimised mechanical perturbation to the sample.

This approach provided movement of the trapped object while confocal imaging within a field of view extending over 1.273 mm×1.273 mm using the 10×/0.4 N.A. lens at a frame rate of 2 Hz. This easily exceeded the field of view of conventional optical trapping systems: for example, a 60×/1.3 N.A. lens would provide only a 0.054 mm² field of view with a much shorter working distance of 0.15 mm. In addition to the extremely large volume over which trapping could occur, an optional digital optical zoom function was available to enable full flexible control over the magnification of the resultant image.

To demonstrate the performance of the system, we performed concurrent trapping, manipulation and confocal imaging of live T cells which were obtained by homogenising mouse lymph nodes. The resulting cell suspensions were washed twice and resuspended in RPMI, prior to labelling with 5,6-carboxyl-succinimidyl-fluorescein-ester (CFSE; Invitrogen) fluorescent dye immediately before the experiment [22]. The CFSE dye has a peak excitation wavelength of 492 nm, and was therefore compatible with excitation using the described 488 nm Ar+ laser. The fluorescent signal emitted from the CFSE labelled T cells had a peak wavelength at 517 nm. The back-scattered fluorescent signal was collected by a variable spectral filter which was set to only collect light between λ=500-550 nm and a photomultiplier tube (PMT) before being converted into a contrast image.

4. Results

Figure 2 presents still images from a representative movie which indicates the movement of a single T cell within a large population. The images presented here were cropped for clarity but the movie featured in Fig. 3(a) shows a wider field of view. The cell in Fig. 2 was manipulated and repositioned at right angled paths as a clear demonstration: such movement is atypical of Brownian motion normally observed in a live cell suspension. By manipulating a sample of n=50 cells and analysing the resultant images (ImageJ, NIH), an average cell transport rate of 1.2 +/- 0.2 µm/s was measured.

 

Fig. 2. Snapshots demonstrating trapping and controlled movement of a single T cell within a large population. Scale bar=300µm.

Download Full Size | PDF

The movie featured in Fig. 3(b). shows movement of cells obtained using the same 10×/0.4 N.A. lens, but employing a 4x digital zoom, thus providing 40x magnification but without compromising the depth of focus within the sample. Again, the cells were manipulated at a similar rate and with the exception of the tweezed cell no movement beyond the normal Brownian motion was observed.

 

Fig. 3. (a). A movie demonstrating optical trapping a T cell using a 10×/0.4N.A. lens. This is a typical movie which shows the movement of a T cell within a large population (Media 1). Figure 3(b). features a close-up movie of an optically trapped T cell using a 10×/0.4N.A. lens. A digital zoom of 4x was employed to visualize the cell trapping at higher magnification (Media 2).

Download Full Size | PDF

By titrating the average power of the trapping laser source, we studied the immediate and short-term invasiveness of the trapping process upon the cell viability. Propidium Iodide (PI) (Sigma) is a prolific fluorescent marker in cell viability studies [23] which, when applied at low concentration (µg/ml) to the medium, can only penetrate viable cells weakly and slowly before subsequently binding to the nucleic acids, resulting in a very weak fluorescence signal at a peak wavelength of 617 nm. However, PI can penetrate compromised cells rapidly, enabling a time window of distinct contrast between healthy and damaged cells that is immediately distinguishable using fluorescence imaging. It is therefore possible to discriminate healthy cells from those with compromised viability by capturing fluorescence resulting from PI loaded cells.

Supplementary to the experimental protocol above, a He-Ne laser with a wavelength of 543 nm and average power <1 mW at the sample was used to excite fluorescence from the PI and a second photomultiplier channel with spectral detection confining the collection across the wavelength range 600–650 nm was used to collect fluorescence from the PI as a consequence of compromised cells.

 

Fig. 4. (a). Results of short-term cell viability studies using Propidium Iodide as a cell damage indicator. (a), features a grid of images displaying T cells labelled with CFSE (green) under varying average laser power and exposure times. Cell damage is indicated by propidium iodide (PI, red). Scale bar, 50µm. Figure 4(b). displays a graphic representation of the result data.

Download Full Size | PDF

A sample of n=10 cells were trapped and repositioned using the previously described method over a one minute period. Over the following 25 minutes, the cells were trapped and imaged at 5 second intervals in both the CFSE (green) and PI (red) detection channels. Figure 4(a) shows the merged images of both channels at 5 minute intervals, using average trapping laser powers of 3, 6 and 9 mW. When using 3mW average power to trap the cells, no PI uptake was detected throughout the entire 25 minute experiment. This indicates that the trapping process did not immediately compromise the cell viability over this time. By increasing the average laser power to 6 mW, fluorescence from PI was observed in the final stage (20–25 minutes of trapping), indicating increased cell damage resulting from increased laser power. This trend continued with a further increase in average laser power to 9 mW, where fluorescence from intracellular PI was detected earlier in the experiment (after 15 minutes of trapping) and 6 of the 10 cells expressed PI at the end of the experiment, thus indicating significant cell damage. Furthermore, at this higher average power, a gradual decrease in the CFSE emission intensity due to photobleaching was noticeable after 15 minutes, reducing the contrast of the confocal images obtained. Practical application of this trapping protocol would not involve such long imaging periods and extremely high average powers would not be necessary to manipulate the cells, hence this would not limit the efficacy of this powerful method. Given the high speed with which cells can be manipulated, even transporting the cells over >1 mm distances requires appreciably less than 20 minutes of optical trapping at modest average powers.

5. Conclusion

We reported a computer-controlled point-and-click system that used both weak optical and small temperature gradients to simultaneously enable trapping, manipulating and imaging of live T cells with a low magnification, low N.A. objective lens (10×/0.4 N.A.). This innovative method overcomes the requirement for a high N.A. lens used in traditional optical trapping and hence broadens the scope of trapping in biological studies in larger volumes.

Concurrent confocal imaging and trapping was performed using a dual scanning system and two independent hands-free laser systems. This approach provided independent control of the trapping spot without pre-programming and while simultaneously acquiring confocal images in close to real-time. Trapping and movement of live T cells was performed over a field of view exceeding 1 mm², with cell movement at an average rate of 1.2 µm/s. By using PI staining, it was clear that the cells remained viable for periods of up to 25 minutes during constant trapping and manipulation at modest average powers. In practice, even to manipulate the cells over long (>1 mm) distances, the trapping time would be less than 20 minutes and hence the cells would remain viable beyond the manipulation period to enable extended time studies. Furthermore, the adaptability of the magnification by firstly employing a low magnification lens and then subsequently employing digital zoom magnification offers a flexible method to studying both single-cell interactions and cell-cell interactions and behaviour in a large population in vitro and within micro-grid arrays. This enabling approach allows, for the first time, fully flexible control and concurrent visualization of cell-cell interactions within large dimensions which is undeniably key to understanding the behaviour of the immune system and beyond.