1. Introduction

In recent years, the advance in imaging methods has become apparent in biology, due to the tremendous progress in fluorescence tagging techniques and nano-metric probes. Nowadays, high-spatial and temporal resolution techniques as confocal [1], two-photon [2] total internal reflection microscopy (TIRF) [3] or fluorescence lifetime imaging microscopy [4, 5] permit to look at subcellular details and macromolecular organization with sub-micrometer resolution.

The possibility to optically explore the complex mechanisms which regulate cell function in living cells on a nanoscopic scale has generated, in parallel, the need for the development of manipulation techniques with a comparable resolution.

Optical tweezers (OT) enable an all-optical manipulation of matter with micrometer precision, piconewton control of the applied forces and, at present, is one of the most promising manipulation technique operating with minimal invasion in biology [6]. It permits manipulation of micrometric and submicrometric biological samples as viruses, bacteria, DNA and living cells [7, 8, 9, 10]. The movement and force of motor proteins can be measured by attaching micron-sized beads to single motor proteins [11, 12]. Cellular force transduction [13, 14, 15, 16, 17] or membrane elasticity [18] can be measured by inducing controlled localized forces or tensions via the attachment of beads on cell membranes.

The successive introduction of multi trap [19] optical tweezers has extended these manipulations to different objects simultaneously and recently it has been used to demonstrate multi force optical tweezers [20] where the force of each trap can be adjusted individually. Furthermore, the recent advent of spatial light modulators, has permitted to evolve from the ‘conventional’ optical tweezers systems towards the so called dynamical holographic optical tweezers. Here, thanks to the wave-front engineering of the trapping beam, it is possible to generate multiple traps in three dimensional (3D) geometries and to dynamically reconfigure them [21, 22, 23, 24]. Moreover the intensity profile of each spot can individually be modified either to change its mode profile or improve the trap efficiency. For example, the trapping beam can be shaped to generate optical vortices [25, 26, 27], with the advantage to trap reflecting-, absorbing- or low-index- particles. Or else it can be modified to compensate for spherical abberation for trapping objects deep in sample cells [28].

A further advance in the development of optical techniques for biological research is represented now by the combination of the progresses made in high temporal-and spatial-resolution imaging with those attained in optical manipulation.

Up to date, in most of the OT microscopes a wide field epi-fluorescence scheme is used. Very recently, to improve the optical resolution, new imaging solutions start to be exploited as presented in Ref. [29]. In this paper the authors measured the mechanical transitions corresponding to single DNA hybrid ruptures by combining TIRF and OT. They simultaneously achieved, in the near-field volume, pN force measurements and single molecule fluorescence detection.

To further exploit the potentiality of an OT system with a high resolution imaging set up, a possibility is to incorporate into the OT microscope a 3D scanning optical technique. However, this combination presents technical challenges when the trapping and imaging beams are combined by the same objective lens. One problem encountered in this configuration is that the axial movement of the microscope objective inevitably displaces the trapping plane of the same amount.

Alternatively, two objective lenses head-on with completely dissociated beam steering optics can be used. This configuration is however incompatible with the possibility to perform differential interference contrast (DIC). Moreover, to optimize both the quality of the optical traps and the optical resolution of the imaging system, oil or water immersion objectives with a high numerical aperture (≥1.2) and consequently a short working distance are largely preferred. Therefore, this configuration limits the sample thickness and the possibility to mount on the upper part of the microscope additional tools as micromanipulators or microinjectors, and definitively hamper an easy switch from fluorescence to transmission measurements.

Very recently, M. Goksör and colleagues [30] have combined a trapping- and a multiphoton-beam using a single objective. An external lens has been translated to control the divergence of the trapping laser and optical sectioning has been achieved by moving the trapped object through the image plane. However, this solution is inappropriate for imaging planes axially distinguished from the trapping plane.

One solution to this problem has been presented in Ref. [31] where a confocal and an OT microscope have been combined by coupling the trapping beam into an optical fiber mounted on a translation stage. In this way, the objective displacements have been compensated with a synchronized reverse motion of the trapping plane. However, the system is limited to a single optical trap.

To overcome these limits, we present an alternative method which consists in shaping the trapping beam via digital addressing a Liquid Crystal Spatial Light Modulator (LC-SLM). The wave front of the trapping beam is modified to compensate the movement of the objective by projecting on the LC-SLM fast moving correctional diffractive optical elements (DOEs). As a result, we can keep the optical traps at a fixed position whilst the objective is moved. The advantage of the proposed technique is that we can fully exploit the benefits of using a holographic optical tweezers system, such as the possibility to generate and move multiple traps, eventually organized in 3D geometries and/or in different modes, in combination with those of a 3D optical scanning technique.

The capability of our system is demonstrated in three different experiments. In a first one, we show that we can decouple the position of a trapping plane from the axial movements of the objective by performing optical sectioning of a circle of trapped beads. In a second experiment, we use a similar scheme to apply mechanical constraints on the dorsal surface of a cell whilst performing 3D optical sectioning of the cell. To this end, the nucleus of HeLa cells have been fluorescently labeled with H2B-GFP and a circular array of micro beads is attached on the dorsal surface of the cell and kept at a fixed position during the fluorescence and transmission sectioning. Finally, we extend our method to a 3D beads volume. Silica beads are organized in a 3D geometry and we show that we can simultaneously control the position of different planes while performing axial optical sectioning.

2. Microscope design

The microscope is based on a standard inverted microscope (Zeiss AxioVert 135) with differential interference contrast (DIC), and epi-fluorescence. The attenuated and expanded (3X) 1064-nm beam of a 10W single-mode CW fiber laser (IPG Photonics YLM-10) is directed onto the LC-SLM, reflected into the side port of the microscope and directed with an IR dichroic mirror to the focusing objective (100x NA-1.3 oil). To allow simultaneous recording of fluorescence and DIC while using the optical traps, the dichroic is positioned above the fluorescence filter block. In this way, the excitation light from a Mercury lamp mounted on the rear port of the microscope is focused into the sample by the same objective used for trapping. Fluorescence from the sample is sent to a high sensitive CCD camera (Cool Snap HQ, Roper) placed at the exit port of the microscope. To obtain DIC images, a polarizer is inserted below the fluorescence block filter and crossed with the DIC analyzer positioned above the microscope condenser.

In order to perform 3D optical sectioning, the objective is mounted on a nanofocusing positioner (PI Instruments, PIFOC 721.10). To improve the resolution, the collected images are deconvoluted by the programm Huygens (Scientific Volume Imaging), with the algorithm QMLE and using experimental Point Spread Functions (PSF) of 0.17 µm fluorescent beads (Molecular Probe).

 

Fig. 1. (a) Schematic of the trapping (red) and the imaging (green) beam paths. L - laser, SLM - spatial light modulator, DM - infrared dichroic mirror, MO - microscope objective, TL - tube lens, T - telescope; (b) Calibration curve which relates the SLM focal length, fSLM, to the axial position, z, of the trapping plane.

Download Full Size | PDF

The SLM (Hamamatsu PPM-X8267) is a programable phase modulator that uses an optical image transmitting element to couple an optically-addressed Parallel Aligned Nematic Liquid Crystal SLM (PAL-SLM) with an electrically-addressed intensity modulator.

A spherical wave approach [24, 32] is used to calculate phase DOEs which are displayed on the SLM to transform the incoming laser beam into a distribution of laser spots, each with individually specified characteristics and arranged in an arbitrary desired geometrical configurations.

2.1. Trapping beam path

A schematic of the optical set up is presented in Fig. 1(a). The trapping beam path (red lines) comprises the laser source (L), the SLM, the dichroic mirror (DM) and the microscope objective (MO).

The SLM is represented by a reflective thin lens and the microscope objective, MO, by its principal planes. The distance between the SLM and the MO first principal plane is d. The laser beam is reflected and modulated by the SLM in an array of spot focused on the plane F_SLM. This plane is imaged by the microscope objective at a distance z from its focal plane F_MO. f_SLM and f_MO are the focal lengths of the DOE implemented on the SLM and of the objective, respectively.

From the scheme in Fig. 1(a), by applying the conjugation equation written in focal coordinates for the MO, we have:

where:

Introducing Equation 1 in Eq. 2, we obtain a simple expression which relates the focal length f_SLM to the axial trapping position z:

Once the two fixed parameters d and f_MO are known, Equation 3 permits to derive for each position of the trapping plane, z, the focal length, f_SLM and viceversa.

 

Fig. 2. Transmission image of a 3D structure of beads trapped in three planes at an axial distance of 2 µm. The imaging plane is adjusted at z₁=9.6µm, so that the intermediate plane of the structure results in focus.

Download Full Size | PDF

The values for d and f_MO have been obtained by fitting the experimental curve shown in Fig. 1(b). To obtain the curve, we calculated and displayed on the SLM, which is placed at about 800 mm to the first surface of the MO, phase DOEs with different focal lengths, f_SLM . For each values, we optimized with an accuracy of 10nm the objective position, z_ob , which maximizes the intensity of the reflected signal produced by focusing the traps on the coverslip. In this condition, z ₀=0 and the values of z_ob correspond to the values of z. From the fitting of the curve, we derived the following values: d=833 mm and f_MO =1.5 mm.

2.2. Imaging path

In the presented microscope, it is crucial to have the possibility to focus objects located at planes different from the objective’s focal plane. To this end, the microscope needs an adaptation of the optical path from the sample to the imaging CCD, as it is schematized in Fig. 1(a). In a standard microscope, the CCD detection plane would be positioned at the exit port of the microscope, i.e., at the focal plane of the tube lens TL, F_TL, conjugated with the objective focal plane FMO. The image of a sample-object placed at a distance z ₁ from F_MO would form at the optical plane, F_T, shifted by D=M ² z ₁ (M being the magnification of the microscope objective) in respect to FTL and therefore inside the microscope. In our configuration, we transfer the image out of the microscope by using an external telescope T inserted between the exit port of the microscope and the CCD. The telescope is positioned with its first focal plane in coincidence with the position of F_T, and its second focal plane in coincidence with the CCD detection plane. The telescope and the CCD are mounted on a translation stage that allows us to adjust the position, z ₁, of the imaging plane by a rigid adjustment, D, of the telescope T and the CCD. For D=0, the position of F_T coincides with the focal plane of the tube lens, F_TL.

 

Fig. 3. Relative positions (not in scale) of the trapping, imaging and objective planes with respect to the coverslip and the objective focal plane during an axial scan.

Download Full Size | PDF

The efficacy of this imaging approach is demonstrated in Fig. 2. Here 7 silica beads are trapped in 3 different planes separated along the axial direction by a distance of 2µm. The intermediate plane is at 9.6 µm from the objective focal plane. The diameters of the bottom and intermediate planes are 14µm and 10.5µm, respectively. A single spot is centered in the top plane. The image shown in Fig. 2 has been acquired by adjusting the position of (T+CCD) in order to have the imaging plane coincident with the intermediate plane of the structure, i.e. at z ₁=9.6µm, the bottom and top plane are out of focus but still visible.

3. Results: 3D optical sectioning and multi trap optical tweezers

We first present an experiment where we decouple the position of a 2D trapping plane from the axial movement of the microscope objective. In a second experiment we apply a similar scheme to perform optical sectioning of a fluorescent labelled living cell while keeping a circular array of micro beads attached on its dorsal surface. Finally, we extend the method to perform an axial scanning of a 3D volume of beads.

The use of a DOE always gives rise to a non diffracted order which is focused on the focal plane of the microscope objective: the zero order spot. The latter can induce cell damaging and generate an extra signal that adds up to the fluorescence signal. In the following experiments, this two undesired effects are avoided by generating laser traps far enough from the objective focal plane and choosing the observation volumes out of this plane.

For the first experiment, we have calculated a temporal sequence of 46 phase DOEs that produce a circular array of 6 laser traps (diameter of 10µm) and keep it at a fixed position in respect to the coverslip while moving the objective over a distance of 9.2 µm by steps of 200 nm. Silica beads (2.34µm in diameter) diluted in water are trapped in the generated spots. DIC images are acquired at each step of the optical sectioning.

As schematized in Fig. 3, there are three optical planes to be considered:

(i) the objective focal plane (in gray) which coincides with the focusing plane for the zero order spot: this plane is at fixed distance, f_MO , from the principal plane of the objective, while it position, z_ob , with respect to the coverslip is varied during the axial scan.

 

Fig. 4. (a)–(f) Selected sequence from a 3D DIC axial scan with a step of 200nm of 6 beads trapped by the laser. The imaging plane position, z_IP , in respect to the coverslip and the trapping plane position, z, which compensate for the objective movements, are indicated in the figure. As a results of the compensation, the circle is kept at a fixed high, z ₀=4.6 µm, from the coverslip. For the whole optical sectioning see Multimedia file 1. (1622 KB)

Download Full Size | PDF

(ii) the imaging plane (in green): in a conventional microscope it would coincide with the objective focal plane, here it can be adjusted to an arbitrary axial plane by regulating the imaging path, (T+CCD). During the scan this plane lies at a fixed position, z ₁, from the objective focal plane and at a variable position, z_IP , from the coverslip. For the first experiment we set z ₁=9.6 µm.

(iii) the trapping plane (in red): its position, z, relative to the objective focal plane, depends on the focal length, f_SLM , that we set for the DOE (Equation 3). During the optical sectioning, the values of f_SLM are varied in such a way that the trapping plane is kept at a constant position, z₀, from the coverslip.

Figures. 4(a)–(f) show six cross sections from the optical sectioning of the circle of beads. In the same field, part of a pattern on the coverslip is visible. From the left to the right, the objective moves down to the coverslip, as demonstrated by the pattern which is gradually coming into focus. Correspondingly, the imaging plane moves from $z_{\mathit{\text{IP}}}^{\mathit{\text{in}}}$ =+9.2 µm to $z_{\mathit{\text{IP}}}^{\mathit{\text{fin}}}$ =0 µm. To compensate the objective movements, the distance z between the trapping and the objective focal plane is varied between zⁱⁿ =+5 µm to z^fin =+14.2 µm. As a results, the circle of beads is kept at a constant hight, z0=4.6 µm, during the whole scan. This is evident by looking at the sequence of images (Figs. 4(a)–(f)) which shows the circle of beads out of focus at the beginning and at the end of the axial scan and in focus at the half of the scan (Fig. 4(c)).

In a second experiment we show how this method can be extended to combine multi trap optical tweezers and fluorescence optical sectioning of living cells.

Similar to the previous experiment, a sequence of phase DOEs which generates a circular array of 6 beads disposed on a circle of 15µm diameter is addressed onto the SLM. Silica beads are trapped onto the circle and held on the dorsal surface of a living HeLa cell. The mechanical interaction between cell cytoskeleton and beads is induced by coating the beads with arginine-glycine-aspartic acid (RGD) according to the manufacturers procedure (Telios Pharmaceuticals Inc., CA USA). RGD is a small peptide contained in the tenth type III domain of the fibronectine which plays a central role in cell adhesion [33].

Once the circle of beads is positioned and attached on the cell cortex, we performed a transmission and fluorescence sectioning of the cell.

 

Fig. 5. Sequence of (a) transmission and (b) fluorescence images (step of 1µm) acquired during an axial scan where a circle of 6 beads is positioned and held on the dorsal cortex of a HeLa cell. The beads are kept at a fixed position while the objective scans a range of 8.2 µm, at step of 200 nm. Fluorescence images have been de-convoluted as explained in the text. For the whole fluorescence sectioning see Multimedia file 2 (3146 KB)

Download Full Size | PDF

To this end, cell nuclei have been marked by using histone H2B tagged with GFP (generous gift from T. Kanda; The Salk Institute, La Jolla, CA). HeLa cells have been cultured on glass coverslips in Dulbecco’s Modified Eagle Medium (Invitrogen, Cergy Pontoise, France), supplemented with fetal calf serum (10%), at 370C in a 5% CO₂ atmosphere. The cells have been transfected with the plasmid encoding H2B-GFP using FuGENE^TM 6 Transfection Reagent (Roche Molecular Biochemicals, Meylan, France). The coverslips have been mounted in a thermostated open chamber, allowing for long time observation at 370C in a 5% CO₂ atmosphere.

Wide field optical sectioning of the cell is performed by moving the objective and therefore the imaging plane from $z_{\mathit{\text{IP}}}^{\mathit{\text{in}}}$ =+12.6 µm to $z_{\mathit{\text{IP}}}^{\mathit{\text{fin}}}$ =4.4 µm. At the same time the trapping plane for the beads is kept at z ₀=8µm by running the sequence of DOEs that compensate for the objective movement (zⁱⁿ =+5 µm to z_fin =+13.2 µm). At each position of the objective, we acquired a transmission and a fluorescence image. To improve the resolution of the 3D scan, the fluorescence images have been afterward de-convoluted by the Huygens program. The experimental PSF for the de-convolution has been measured in the same imaging condition.

A sequence of transmission and fluorescence images acquired during the optical sectioning of an HeLa cell are shown in Fig. 5(a) and (b). For z_IP ⋍+8µm the imaging plane overlaps with the trapping plane as is reveled by the transmission sequence (Fig. 5(a)). The cell nucleus is crossed by the imaging plane towards the end of the scan, as it appear from the fluorescence sequence shown in Fig. 5(b).

Finally, we performed an optical sectioning of beads trapped in the 3D geometry of Fig. 2. In this case, we wanted to keep constant during the axial scan the position and the 3D shape of the structure. Therefore the movement of the objective has been compensated by calculating, for each step, the DOEs necessary to simultaneously move backwards the three planes of the 3D array of beads. Figures 6(a)–(f) show a selected sequence of images taken during an axial scan performed from $z_{\mathit{\text{IP}}}^{\mathit{\text{in}}}$ =+12.6 µm to $z_{\mathit{\text{IP}}}^{\mathit{\text{fin}}}$ =7.4 µm with a step of 200 nm. The three planes of the structure are coming in to focus each time that the imaging plane coincides with one of them (Figs. 6(a),(c),(f)).

 

Fig. 6. (a)–(f) Selection of 6 cross sections, at step of 1 µm, of an optical sectioning of the 3D structure shown in Fig. 2. The positions, z_IP , of the image plane in respect to the coverslip are indicated in the figure. In the original scan images are taken form z_IP =12.6 µm to z_IP =7.4 µm a step of 200 nm. For the whole optical sectioning see Multimedia file 3 (877 KB)

Download Full Size | PDF

4. Summary and conclusions

We have presented a new method for biological microscopy which enables to perform simultaneously 3D optical manipulation and optical sectioning. This technique enables to decouple the imaging and the trapping planes using the same objective. We change the imaging plane position via a mechanical movement of the objective while the position of the trapping beam is modified by engineering its wave front through a digitally addressed LC-SLM. The DOEs profiles that we used to address the LC-SLM are calculated by using a home made software based on a spherical wave approach. Moreover we have presented a modification of the microscope imaging path which allows us to image planes distinguished from the objective focal plane. Interestingly, our method can be easily extended to the case where optical manipulation is performed in combination with other optical scanning techniques as confocal or two photon microscopy.

To illustrate the potentiality of our method, we have performed three experiments. In the first, we have shown that we can move the imaging plane in respect to the trapping plane by performing a 3D optical sectioning of a circle of beads kept by the laser. This approach is very useful for high resolution optical sectioning of motile biological samples such as bacteria, non adherent cells, spermatozoa etc. In this case one can hold the sample with the trapping beam and simultaneously perform the 3D optical sectioning of the trapped object. Moreover, the possibility to generate multi traps allows one to manipulate many samples at the same time and to control their relative position. Consequently one can use this manipulation to study the mutual spatial influence between interacting cells during their optical sectioning. Finally, by modifying the mode of the optical traps one can imagine to extend this technique for optical sectioning of absorbing-, reflecting- or low index-samples.

In the last decade, several experiments have shown that cells can sense mechanical stimuli coming from their environment and traduce them in intra cellular signaling. Mechanotransduction occurs primarily at adhesive contacts where membrane receptors make a physical-chemical link between the extracellular matrix and cell cytoskeleton. To dissect these processes, a typical experiment consists in using OT to place micro-spheres coated with specific extracellular ligands on the dorsal surface of cells and probe the resistance of the receptor-cytoskeleton connection, e.g., by monitoring the recruitment or activation of specific proteins [13, 14, 15, 16, 17]. The second experiment of this paper has been configured in order to prove that we can indeed control the position of a circle of beads on cell cortexes whilst performing fluorescence optical sectioning through the cell. It will be straightforward now to extend this experimental scheme to monitor the activation or recruitment of proteins such as vinculin, talin, paxilin specifically involved in cell machanotransduction. This will allows one to reconstruct 3D map of cellular mechanotransduction and follow how the effects of mechanical constraints propagate in the cell.

Finally, in the last experiment we have shown that we can control and fix the position of optical traps located on different axial planes while simultaneously performing a 3D optical sectioning of the volume delimited by the planes. With this experimental scheme we can imagine to extend the previous experiment to the case where mechanical constraints are distributed in 3D patterns. As already shown, the rapid refreshing rate of the SLM enables to dynamically adapt the distribution of the optical traps to the cell shape and to hold a 3D volumes of beads on cell cortexes [34]. The combination of a holographic microscope and a 3D optical sectioning technique will allows us to complete this kind of manipulation with the possibility to monitor the effect of a 3D pattern of forces on cellular mechanotransduction.

As a final remark, we note that a similar approach is not limited to decouple an imaging and a trapping beam, but one can imagine to extend the same method to decouple an imaging path from any perturbation beam e.g a photoactivating-, uncaging- or scissor-laser.