1. Introduction

The use of lasers in microscopy has provided researchers with a means to remotely modify, dissect, and optically trap biological tissue, whole cells, or subcellular regions of interest [1]. In particular, for near-infrared wavelengths and sub-picosecond pulsed lasers, multiphoton absorption and related ionization physics provides an optical method of nanodissection with inherent sterility, localized interactions, and deep penetration in biological samples [2,3]. By tightly focusing the laser beam using high NA optics, the photon density localization in space and time can yield multiphoton and avalanche ionization that can generate a transient plasma within the focal zone. This effectively ionizes a sub-femtoliter volume of the sample without damaging adjacent areas in a cell or biological sample [3–5]. As a tool for remotely and selectively modifying local regions in a single cell, femtosecond laser irradiation has been used for cellular surgery [6,7], gene inactivation [8,9], gene transfection [10,11], and cell stimulation [12–14]. In these studies, it can be particularly difficult to determine the lower threshold for these ablation-type effects in cells. Since two-photon microscopy is in widespread use, and laser powers for two-photon imaging are often within an order of magnitude of those used to perform nanodissection in cells [15], it is important to determine the sample modifications that occur at the onset of ablation-type effects. Ideally, we should understand and characterize the effects of femtosecond laser irradiation in cells, and also determine a “safe” window of power levels (if any such window exists) so that experiments using two-photon microscopy do not inadvertently modify the cell during imaging. The use of tightly focused femtosecond pulses can also be exploited as a useful biophysics tool to modify cellular targets [16]. Whether subcellular effects of femtosecond laser irradiation are intended or whether avoiding such effects is preferable varies depending on the application. For either case, the experimental conditions that bring about the onset of such effects is an issue of critical interest.

Such subcellular effects can, however, be extremely difficult to observe directly. Cells are by nature, water-based, transparent objects, with inherently low contrast. Furthermore, the liquid nature of the cell means that intracellular organelles may move within the cell, cytoplasm can flow, bubbles can collapse and any ablation-effects induced by laser can rapidly diffuse through the cell, leaving little or no evidence of laser irradiation. [17,18]

If the sample absorption properties are known, or assumed, then some progress in understanding the laser effects can be made through simulations based on accepted photoionization physics [16]. This approach is complicated, however, in laser irradiation where the average power is at or near the threshold for inducing observable effects in inhomogeneous cells. At or below the threshold for observable structural changes in a cell, fluorescence photobleaching [19–21] and photoinduced generation of singlet oxygen and Reactive Oxygen Species (ROS) will occur [22,23] and can be detected by optical or chemical methods [14,22,23]. As the laser power approaches the ablation threshold, physical and structural changes in the sample occur, and the ability to image such effects becomes of paramount importance. Researchers have therefore used a variety of microscopy methods in order to observe the effects of intracellular laser irradiation. Such microscopy methods include: confocal fluorescence [24], brightfield [25], differential interference contrast (DIC) [26–28], multiphoton [20], time-resolved probe beam scattering analysis [29,30], scanning electron microscopy (SEM) [31], and transmission electron microscopy (TEM) [26–28,32]. Using light microscopy in vitro allows for immediate observation of any laser-induced effects that have sufficient size and contrast to be resolved. For laser ablation of a power level sufficient to cause generation of bubbles or other high contrast effects in the cell, light microscopy is ideal. Fluorescent staining can be also used to highlight organelles or molecules of interest and can be used to track laser-induced changes in a cell, provided that the photobleaching effects [26–28,32] and the interaction of the dye with the light absorption properties [5,33] are taken into consideration.

If, however, we require observation of ablation-type effects at or near the onset power threshold where laser-induced changes are of sub-micrometer size, then electron microscopy offers unmatched resolution, albeit with the disadvantage that the sample requires some form of treatment prior to imaging. SEM has been used to provide clear images of membrane dissection by femtosecond laser irradiation [6], but is generally limited to observation of the surface of a cell that has been chemically fixed and coated with conductive metal. TEM has been used to observe the onset of laser ablation by scanning the laser focus at varying pulse energies and imaging changes in the total sample opacity [32]. However, to obtain the clearest images of the onset of laser ablation, it is necessary to mount the sample, and obtain physical slices by ultramicrotome with a thickness of 100 nm or less. These slices can then be viewed sequentially under TEM, and offer the greatest potential for imaging nanoscale ablation-type effects in a cell. This technique has rarely been used due to the time-consuming nature of the sample preparation, the expense of the equipment, and the difficulties in fixing the sample immediately after the laser irradiation. The potential for high-contrast, high-resolution imaging has, however, been successfully demonstrated by TEM observation of mitochondrial ablation by 532 nm 7 ns pulsed laser irradiation [28].

Here, we use transmission electron microscopy with fixed, mounted and sliced cells in order to investigate the visibility of physical changes in a cell induced by femtosecond laser irradiation. By using glutaraldehyde fixation before laser irradiation, we are able to clearly observe laser ablation of cytoplasmic, nuclear regions in the cell. While the effect of pre-fixation on the laser-cell interaction is not yet fully clear and is difficult to analyze (due to bubble collapse and diffusion of laser-induced changes in the cell when fixation is not used), pre-fixation allows repeatable observation of intracellular laser ablation at power levels and allows an investigation of the threshold for physical modification in the cell. If, on the other hand, the fixation is performed after the irradiation, the visibility of the laser effects are significantly reduced, but may still be observed. In the nucleus, laser irradiation was seen to cause aggregation of electron dense material at the irradiation site, while irradiation of the cytoplasm produced swollen mitochondria (which may indicate the onset of apoptosis) but no observable physical changes at the irradiation site.

2. Experimental methods

2.1 Optical setup

As an irradiation light source, a mode-locked Ti:sapphire (Spectra physics, Tsunami) laser was used (wavelength: 780 nm, repetition rate: 82 MHz, emission pulsewidth: 80 fs). The laser beam was expanded and then directed into an inverted microscope (Olympus IX 70). Inside the microscope, the beam was reflected by a dichroic mirror and focused in the HeLa cell by a water immersion objective lens (Olympus, PlanApo WLSM, N.A.1.0, x60). The pulse length at the sample will broaden to approximately 190 fs due to GVD based pulse broadening from the optics [34]. The intensity of the laser irradiation was controlled using two polarizers set to maintain a constant polarization at the sample. Power was measured with a power meter (Spectra Physics, 407A) at the sample stage. The irradiation time was 8 ms (creating a pulse train of ~650000 pulses at 82 MHz) and was controlled by a mechanical shutter. The lateral location of laser irradiation inside the cell was precisely controlled by a stepping motor controlled microscope stage (Sigma, BIOS-202T), and in the optical axis the laser was positioned as close as possible to halfway between the lower and upper cell plasma membranes. Unless otherwise noted, the output polarization of the laser was not changed. However, for experiments testing polarization-dependence, the polarization was changed by first inserting a half-wave plate in the beam path so that the polarization could be rotated by 90° by a 45° rotation of the wave plate. A quarter wave plate was also used to create circular polarization for comparison experiments.

2.2 Sample preparation and laser irradiation

HeLa (cancerous epithelial) cells were cultured on 35 mm dish with a 0.17 mm glass bottom (Matsunami glass), using Dulbecco’s Modified Eagle’s Medium (DMEM) which contained 10% fetal calf serum, 2 mM glutamine, 100 unit/ml penicillin and 100 µg/ml streptomycin, in a humidified atmosphere (5% CO₂) at 37 °C. Details can be found in a previous report [35]

2.3 Laser irradiation before fixation

Cultures were washed with tyrode solution (NaCl 8.5 g/L, Glucose 1.8 g/L, HEPES 10 mM, KCl 4 mM, MgCl₂ 1 mM, CaCl₂ 1 mM, NaOH 4 mM) 3 times. Then, the cells in tyrode solution were irradiated. After that the cells were fixed immediately (i.e. within 3~5 seconds) for the best chance of observing the effects of laser irradiation in experiments not using a fixative before the laser exposure. As a fixative, 2.5% glutaraldehyde in 0.1 M Millonig’s phosphate buffer (NaH₂PO₄, NaOH, PH 7.3) was used.

2.4 Laser irradiation after fixation

After washing cells with tyrode solution three times, the cells were bathed in the 2.5% glutaraldehyde in 0.1 M Millonig’s buffer for 30 minutes and then washed with tyrode solution a further three times. After that, the cells were bathed in fresh tyrode solution and were then subjected to laser exposure.

2.5 Transmission electron microscopy (TEM)

Whether cells are fixed before or after laser irradiation, TEM observation requires further processing to allow the observation of the intracellular structures. Cells need to be stained with an electron density contrast agent suitable for TEM, and also to be processed to allow microtomed sections after fixation. We used a conventional method [27] which is outlined below. Cells were washed with 0.1 M Millonig’s buffer twice and then treated with 1% OsO4 in 0.1 M Millonig’s buffer for one hour (4 °C). After one hour, cells were washed in the same buffer and then treated with 0.15% tannic acid in buffer for one min. Next, cells were washed in the buffer once, then twice in distilled water, stained with 1% uranyl acetate (4 °C) for one hour, and washed in distilled water. This was followed by dehydration in a graded series of ethanol and embedded in Epoxy resin. Ultrathin sections were cut at a thickness of 104 nm and successive sections were laid on slotted copper grids with a support membrane coated by carbon. After staining with uranyl acetate and lead citrate, sections were viewed with TEM (Hitachi, H-7100). Sequential features from sections were constructed in 3D using the pc software (ImageJ, ver. 1.39j, NIH) using the “volume viewer” plugin (ver. 1.31).

2.6 Distortion of morphological features in the samples

Although TEM has an impressive spatial resolution, the sample preparation procedures often introduce physical distortion of the sample that detracts from the nanoscale resolution of the microscope itself. We investigated two types of distortion that may occur in TEM sample preparation procedures, namely the shrinkage of the sample due to the ethanol dehydration, and the error occurring in axial measurements due to the inaccuracies of the microtome slice thickness. The slicing thickness error was estimated by observing sequential slices from the microtome with atomic force microscopy (AFM, Seiko Instruments Inc., SPM400). Over a series of 17 slices where the microtome was set to 100 nm slice thickness, the average thickness as measured by AFM was 104 nm, with a standard deviation of 16 nm. For an axial measurement over all 17 slices (i.e. for a measurement of the axial length of an object which appears to be 1.7 µm in axial length based on the ultrathin sections) the measured error is less than 50 nm. The resolution limit on axial measurements is then the 100 nm slice thickness, which itself introduces an error of +/- 50 nm for measurements of axial dimensions. To determine distortion of the sample caused by shrinkage, we compared distances between laser focal spots measured by TEM with the distances predicted by the motorized scanning stage. The distances between focal spots were set to between 2 and 10 micrometers (with an error of approximately 2.5% due to the scanning stage itself). A total of 27 measurements between focal spots from 6 cells in 4 dishes were compared with reference distances read from the scanning stage. The result was that distances measured in the cells by TEM were marginally shorter than the distances specified by the motorized stage. Measurements were normalized to the distances set by the scanning stage and the average normalized distance was 0.955 in x and 0.974 in y, with standard deviations of 0.08 and 0.054, respectively, showing shrinkage of several percent. As evident from the standard deviation, some samples actually showed slight expansion. This is known to be an effect of the osmium fixation protocol [36]. Overall, the ethanol stage of the fixing and mounting stage generally results in an overall shrinkage of the final specimen, as shown in the TEM-based measurements we report above. We assume isotropic shrinkage and estimate the lateral and axial shrinkage to be 5% on average. Measurements referred to throughout the paper are not calibrated or corrected for the measured shrinkage.

3. Results and discussion

3.1 TEM analysis of ultrafast laser ablation in fixed cell cytoplasmic regions

We observed changes in fixed cells resulting from a focused femtosecond near-infrared laser pulse train. The glutaraldehyde fixing technique allows a nanoscale investigation of laser-induced effects in cells. The laser power dependency of these modifications was analyzed. Although laser power is the typically measured parameter in microscopy, laser ablation analysis with pulsed excitation sources is more commonly performed in terms of laser pulse energy and radiant exposure [16], and we use pulse energy as the main parameter for comparison throughout this paper since it is not affected by aberrations in the focus spot size. The laser pulse energies of 0.48, 0.6, 0.72 nJ correspond to 40 mW, 50 mW and 60 mW of average metered laser power from a Ti:Sapphire laser operating at an 82MHz repetition rate. The equivalent radiant exposures are 0.068, 0.084, 0.101 J/cm², respectively, for the spot size calculated for a 1.0 NA lens and 780 nm wavelength. For these pulse energies, laser-induced changes were clearly observed in the cytoplasm (as shown in Fig. 1). When the pulse energy was reduced to 0.36 nJ, no clear physical evidence of laser ablation could be observed. The threshold of pulse energy to induce cellular structural changes in fixed cytoplasm was therefore observed to be between 0.36 nJ and 0.48 nJ for these conditions. In Fig. 1 for pulse energies of 0.6 nJ and 0.72 nJ (b and c), 10 independent locations in the cell were exposed to irradiation, and in the resulting images, all 10 sites show physical evidence of the laser irradiation. However in Fig. 1(a) where the pulse energy was reduced to 0.48 nJ, only 5 out of 15 sites in the cell showed observable effects of the irradiation. The diameter of the observed laser affected regions varied from 500 nm to 1 µm, with no clear size dependence on pulse energy. Ultrashort pulsed near-infrared lasers are generally considered to be advantageous over other lasers for cell surgery due to the localized energy deposition and multiphoton absorption producing a small effective focal region. Since cellular material and water does not significantly absorb NIR light, the multiphoton effect should allow ablation with dimensions significantly below the optical diffraction limit. The results in Fig. 1, however, show subcellular ablation with dimensions of approximately diffraction limited size, a point which is analyzed in the next section. It is interesting to note that ablation was clearly visible for some sites but not others in Fig. 1(a). The binary nature of the results seems to indicate that a certain amount of seed absorption is necessary and indicates that the ablation was not significantly influenced by gradual thermal or chemical effects but may instead result from a low-density plasma that is seeded by a high order photon absorption [3,16] and may depend on the specific position within the cell. The results in Fig. 1 evidently show that fixation before laser irradiation provides very clear images of laser effects in cells since fixation immobilizes cellular activity and restricts the dispersion of any laser-induced changes.

 

Fig. 1. TEM images of laser irradiation sites in the cytoplasm of HeLa cells. The black arrows indicate the rows of laser-irradiated sites. The pulse energies were: (a) 0.48 nJ, (b) 0.6 nJ, and (c) 0.72 nJ. The distances between adjacent irradiated sites are 2 µm, and exposure conditions were 8 ms for each irradiation site, with a 780 nm wavelength. The nuclear region is marked with an ‘N’. For the irradiation location marked with a white arrow, detailed sectional images are shown in Fig. 2.

Download Full Size | PDF

3.2 Ablation power dependence and morphology

The lowest pulse energy that was observed to cause bubble-type physical modification of the cell was 0.48 nJ. Previous work by other researchers has indicated dissection of cell organelles but not necessarily laser-induced bubble formation at sub-nJ pulse energies [6, 16]. Another report has indicated that the lower threshold for subcellular ablation was 0.39 nJ [37], where the laser parameters were similar (800 nm, 76 MHz, 145 fs at the specimen through the microscope), but cells were stained by fluorescent dyes. Absorption in the sample is increased by the staining and can reduce the threshold energy because more electrons are easily excited via multiphoton absorption [5]. The experiment reporting a 0.39 nJ threshold corroborates the disruptive effect via ablation (as opposed to mere photobleaching of the dye molecules) by using a restaining method [7]. Heisterkamp et al. reported a threshold pulse energy of 1 nJ [32]. They used chemical fixation without fluorescence staining. The pulsewidth and wavelength were similar, but the repetition rate was much lower (at 1 kHz). Our lower limit of pulse energy was found to be 0.48 nJ, using a much higher (82 MHz) repetition rate. Previous calculations for our system indicate that although heating effects are relatively weak [35], there is a significant dependence on repetition rate [23], and for laser irradiation below the threshold for laser ablation, the repetition rate affects the diffusibility of both ionized molecules and free radical scavengers. It is therefore not surprising that our minimum pulse energy for observing laser ablation is lower than that found by Heisterkamp et al. [16,32]. It should be noted that a direct comparison between different experiments is difficult due to the single exposure vs scanning type irradiation, and that the absorption by the fixative in our experiments as well as those of Heisterkamp et al. [32] may not be negligible.

 

Fig. 2. Images from subsequent microtome slices of the laser-irradiated region marked by the white arrow shown in Fig. 1. The direction of laser light propagation was from the lower numbered sections to the higher numbered sections. The thickness of one sliced section was ostensibly 100 nm (and measured by AFM to be 104 +/-16 nm) and the irradiation pulse energy was 0.6 nJ. In the lower numbered sections, images indicate that the laser ablation mechanism may be dependent on the laser polarization. The arrow indicates the direction of polarization.

Download Full Size | PDF

 

Fig. 3. Outline of laser-ablated region in 3D. This image is constructed with the edge of the laser-induced changes shown in Fig. 2. Image (b) is a rotation of image (a) around the z axis by 90 degrees. White arrows indicate the direction of polarization, and the distance between each sliced section is approximately 104 nm.

Download Full Size | PDF

Figure 2 shows sectional images of the laser-induced changes following irradiation by 0.6 nJ pulses. The evidence of laser ablation is visible in sections 2 to 19. The thickness of each cell section was 104 +/-16 nm, hence the axial length of the laser-ablated volume is 1.8 +/- 0.05 µm (see section 2.6 for estimates of error). Inside ablated regions, the apparent emptiness indicates that either cellular materials were pushed out of the vicinity of the focal spot by formation of a bubble or that denatured materials were present inside the hole, and were subsequently washed out in the procedure for TEM observation or both. Interestingly, irregular ellipsoidal shaped ablation perimeters were observable in some sections. The long axes of ellipsoids seen in some sections were aligned with the polarization of the laser, indicating a polarization effect. Such effects are known in laser processing of semiconductor and metallic samples [38,39], but to our knowledge have not been reported in cellular ablation by femtosecond near-infrared laser pulses. We attempted to verify or rule-out the indications of a polarization effect by measuring the degree of asymmetry (i.e. the ratio of the long and short widths of the focal spots parallel, and perpendicular to, the linear polarization) using both linear and circular polarizations. From 15 different ablation zones for linear polarizations, the average degree of asymmetry was 1.32 with a standard deviation of 0.26. When using circular polarization, where the polarization effect should cancel out, the average degree of asymmetry was 1.20 with a standard deviation of 0.29 from 13 different ablation zones. Further work would be required to make any conclusions as to the existence of such a polarization effect.

Using the outlines of sequential sections, the laser ablation region was reconstructed in 3D, and is shown in Fig. 3. The dashed line shows the full-width half-maximum (FWHM) edge of the calculated diffraction limited laser spot size for the irradiation wavelength (780 nm) focused by a 1.0 N.A. objective lens which has a lateral width of 596 nm and an axial length of 2043 nm. If the focal spot boundaries are alternatively defined as the locations where the intensity drops by e² times the peak intensity, then the spot dimensions are larger with a lateral width of 1013 nm and an axial length of 3470 nm. As mentioned in section 2.6, the dehydration and embedding process (which were performed after laser irradiation in all experiments) can cause shrinkage of the sample and therefore the ablated region diameter by up to 10%, or higher [40]. However, our measurements indicate that shrinkage effects are approximately 5% for our experimental protocols. In terms of the dimensions of the ablation zones, multiphoton absorption effects at threshold pulse energies, if considered in isolation, would indicate that the laser-affected region should be smaller than the diffraction limited laser spot size. Indeed, calculations for the evolution of free-electron distributions resulting from focused femtosecond laser pulses predict electron densities that exist in regions smaller than the spot size, primarily due to the nonlinearities involved in the ionization processes [16,41]. However, bubbles that result from the free-electron generation can grow to a size larger than the electron distribution itself. For the size of the predicted electron distribution, it has been reported that below optical breakdown thresholds, the quasi-free electron density distribution scales in proportion to the k ^th power of the intensity, where k is the multiphoton order of the ionization and avalanche ionization is neglected and a well defined dielectric band-gap is assumed to be present [16,41,42]. Using the 1/e² definition of the focal spot size, the dimensions of the electron density distribution may similarly be calculated for our experimental setup and predict a 1/e² lateral electron density width of 426 nm and an axial electron density length of 1552 nm.

It is, however, the effects of the electrons (such as bubble formation or chemical bond disintegration [3]) together with any other interactions that occur, which combine to produce the observable effects in the cell. The electron distribution itself is highly reactive, but is transient and not visible by TEM. As can be seen in Fig. 2, the effects of the laser interaction are bubble-type objects that remain in the cell, and the observed diameters of the laser affected regions (500 nm to 1 µm) are close to the theoretical FWHM diffraction-limited spot. In combination with the fact that independent ablation sites were either clearly visible or completely invisible, these results indicate that the laser affected zone is expanding as a result of bubble formation, which corresponds with recent studies of femtosecond laser-induced nanocavitation in water and cell targets. These studies showed clearly that, at threshold, the cavitation bubble size is similar to the diffraction-limited focal spot size [29], and may easily exceed the spot size [30]. The dimensions of the effective ablation region are often related to the bubble size which is in turn regulated by the laser-sample interaction as well as the mechanical properties of the cell matrix surrounding the bubble [43,44].

Although effects aside from bubble formation, such as chemical decomposition via ROS diffusion from the laser zone, are also likely to occur, the clear boundary between the laser affected zone and the bulk cell conflicts with the view that chemical modifications alone are responsible for the visible laser effects observed in Fig. 2, and indicates that bubble formation is a dominant mechanism of the laser effect on the cell.

 

Fig. 4. Laser ablation lengths in cytosol were measured by pre-fixation with glutaraldehyde followed by irradiation for 8 ms. No substantial pulse energy dependence was observed, even at threshold. Instead, the probability of ablation was dominated by the pulse energy. Error bars were calculated using the standard deviation of all observable irradiation sites. The laser pulse energies of 0.36, 0.48, 0.6, 0.72 nJ correspond to 30, 40, 50 and 60 mW of average metered laser power, and the ‘X’ stands for no observable ablation at 0.36 nJ pulse energy. The number of ablation zone lengths combined in the graph is: 13 sites from 4 cells in 3 dishes (0.36 nJ); 9 sites from 2 cells in 2 dishes (0.48 nJ); 44 sites from 5 cells in 5 dishes (0.6 nJ); and 12 sites in 3 cells in 3 dishes (0.72 nJ).

Download Full Size | PDF

The range of axial lengths for visible ablated sites was compiled for all observable ablated regions and is shown in Fig. 4. The results show that changing the pulse energy does more to affect the probability of observing laser ablation in the cell than it affects the size of the ablated region. Even at threshold levels, the size of the ablated region showed only weak dependence on the pulse energy.

3.3 Ultrafast laser ablation in fixed cell nuclear regions

 

Fig. 5. TEM images showing 200 nm lateral diameter laser-induced changes in the nucleus. The black arrows indicate the rows of laser-irradiated sites. (a) 12 out of 12 irradiation sites were visible following irradiation at a pulse energy of 0.6 nJ. (b) for a pulse energy of 0.72 nJ, 9 out of 9 irradiation sites were visible. At a pulse energy of 0.48 nJ (not shown), 9 out of 9 irradiation sites were not visible. The distance between adjacent irradiated sites was 2 µm, and the exposure time was 8 ms for each site. Images from subsequent microtome slices are shown in Fig. 6, taken from slices of the region marked by a white arrow.

Download Full Size | PDF

 

Fig. 6. Images from subsequent microtome slices of the laser-irradiated region marked by the white arrow shown in Fig. 5. The direction of laser light propagation was from the lower numbered sections to the higher numbered sections. The thickness of one sliced section was approximately 104 nm, and the irradiation pulse energy was 0.72 nJ.

Download Full Size | PDF

When the laser was irradiated in the nucleus of the cell following fixation, 0.6 nJ and 0.72 nJ pulses also produced observable laser-induced changes (Fig. 5). In nuclear irradiation, in contrast to irradiation of the cytoplasm, pulse trains of 0.48 nJ pulse energy did not create any observable changes in the nucleus. An additional dissimilarity from the case of cytoplasm irradiation was the observation of significantly smaller ablation regions in the nucleus. The laser affected regions had maximum diameters of 200 nm, as shown in the sections in Fig. 6. These are smaller than half of the diffraction limited focal spot for these irradiation conditions (Fig. 6). The shape is also different; sectional images of one of the ablation sites are shown in Fig. 6, exhibiting symmetrical circularly shaped laser-induced changes in all sections. Polarization effects were not observed and the 3 dimensional ablated region is cylindrical in shape. The primary difference in ablation size between the results in the nucleus and those in the cytoplasm can be attributed to the protein concentration difference between the nucleus and cytoplasm. While the ablated sizes were significantly different, the ablation thresholds were only different by 20% between the nucleus and cytoplasm. This indicates that, rather than significantly affecting ablation threshold via the absorption properties of the laser-cell interaction, the increased protein concentration in the nucleus seems to constrain the expansion of the laser affected zone. Similar conclusions have been reached by other groups studying the cavitation effects of pulsed laser irradiation in dense media such as water, cells, or tissue [18,43,44]. It is worth mentioning that the smaller observed ablation diameter in the nucleus is contrary to a simplistic view of the ablation process where a higher density at the sample would lead to more absorption of the laser light, more ionization and consequently a larger total physical effect. The results are, however, in agreement with analysis of pulsed laser-induced bubble formation and confinement [18,43,44].

3.4 TEM analysis of ultrafast laser ablation cytoplasmic regions without pre-fixation

When glutaraldehyde fixation was used immediately after laser irradiation of the cytoplasm, no locally observable laser-induced changes were seen during TEM (Fig. 7(a)). All other sliced sections were similar in that no laser effects could be observed near to the focal spot, even at pulse energies of up to 0.72 nJ. This can be attributed to the elastic properties and diffusing nature of the cytoplasm, where even if laser irradiation creates a bubble or modifies the local structure at the focal spot, the effects diffuse or collapse and disappear before fixation can occur. Laser-induced bubbles are inherently unstable and in other reports on laser nanosurgery of cells [16,43], bubbles have been observed to exist for time frames of less than 1 µs to several seconds. Even relatively long-lasting bubbles would still be unlikely to be observable by the post-fixation technique used in Fig. 7, where the results do not show local effects of the laser irradiation.

Some non-local effects were, however, observed in Fig. 7. The cell was irradiated at 10 independent locations with 0.72 nJ pulses (corresponding to 60 mW of irradiation power), and consequently in Fig. 7(b) a swollen mitochondrion and disordered cristae are visible. For comparison, a mitochondrion from a non-irradiated cell is shown in Fig. 7(c). The mitochondrial swelling was a non-local effect observed only above threshold pulse energies. At 0.36 nJ, 0 out of 4 cells showed mitochondrial swelling. When the pulse energy was raised, swollen mitochondria were observed in 7/7 cells at 0.48 nJ (40 mW), and at 0.72 nJ (60 mW) 4/4 cells showed mitochondrial swelling. Similar changes in mitochondrial ultrastructure have been observed by other groups after NIR femtosecond laser irradiation of a cell, and were attributed to be the result of ROS generation at the focal spot [45]. Previous results from a separate study also showed the onset of cell death begins to occur at 0.72 nJ for the same irradiation conditions [35], and the mitochondrial images in Fig. 7 show that cell lysis may be imminent. The observation of bubbles in the cytoplasm in section 3.1 shows that bubbles may also have been formed the Fig. 7(a), but were not observable by post-fixation. The attribution of 0.72 nJ pulse train-induced cell death to ROS effects is not mutually exclusive with the existence of laser-induced bubbles. Vogel et al. conclude that for bubble radii of a few hundred nm, the resulting pressure wave and effect on the cell will not be immediately lethal based on physical considerations [16]. Previous experiments from our group have shown that the cell viability decreases compared to control cells, but remains at approximately 80% or higher for similar irradiation conditions with a 0.72 nJ pulse energy [35], although the threshold for an effect on cell viability is known to be drastically reduced if the exposure is continuous [22]. The cause of mitochondrial swelling, which was observed at an onset threshold pulse energy of 0.48 nJ is therefore likely to be due to the fact that multiple sites were irradiated within the same cell within a short time frame (10 sites were irradiated in approximately 50 seconds).

 

Fig. 7. TEM image of laser-irradiation sites where the cytoplasm of a HeLa cell was irradiated without pre-fixation. Fixation was performed immediately after 10 sites were irradiated in a pattern similar to that shown in Fig. 1. The pulse energy was 0.72 nJ and other exposure conditions were identical to Fig. 1. (a) No visible local modifications by laser could be observed, but mitochondria in the cell were observed to exhibit swelling. The white dashed rectangle outlines the laser-irradiated regions. (b) Magnified image of swollen mitochondrion. (c) For comparison, the mitochondria of HeLa cell, without irradiation. The nuclear region is marked with an ‘N’.

Download Full Size | PDF

3.5 Ultrafast laser ablation in nuclear regions without pre-fixation

By contrast, when the laser was irradiated in the nucleus, even without pre-fixation, the TEM images of irradiation sites revealed electron-opaque insoluble materials at the laser focus sites that remained after irradiation (Fig. 8). Since there is no fixative, dye or any other modification of the sample, the black materials must result from biological material that has been denatured or modified by the laser. The usage of lead citrate in the TEM observation protocol highlights chromatins, glycogen, and mitochondrial membranes. It has also been reported that ROS generated by laser irradiation can produce thymine-tyrosine cross-links [46], and the black material remaining at the center of the focal spot after irradiation may consist of aggregated chromatins and DNA. From subsequent sectional images through the cell, we can determine that the axial size of the residual black region is 2 µm, with a maximum lateral size of about 300 nm. The viscosity of the nucleus is higher than the cytoplasm, and this can account for the residual observable laser-induced changes. Using optical microscopy, previous researchers have also shown that post-irradiation diffusion distances of multiphoton induced DNA damage in cell nuclei were small compared to the dimensions of the nucleus [47].

 

Fig. 8. TEM image of laser-irradiation in the nucleus without pre-fixation. 4 sites were irradiated in a rectangular pattern. The larger image (a) shows 4 irradiated sites from one microtome slice of the cell. All sections are presented magnified in the sequential images shown in (b). The four point diagram shows the irradiation pattern at the same scale as the sectioned images. Sections numbered 9 to 14 were not recoverable from this sample. The laser pulse energy was 0.48 nJ, and the exposure time was 8 ms.

Download Full Size | PDF

The results in this paper show that the post-irradiation movement of laser-induced nuclear damage may be as small as a few tens of nanometers between adjacent irradiation points. Although such observable residual effects must be partly caused by the higher viscosity of the nucleus, the viscosity alone cannot account for the qualitative difference in the type of laser-induced change, and we must consider the nature of the laser induced change. With fixation, ablated holes were observed. Without fixation, the only evidence of laser irradiation was the black material in the center of the focal spot. We then hypothesize that the black residual material visible in Fig. 8 results from the generation and collapse of a micro-bubble at the focal spot. As mentioned in the previous section, laser-induced bubbles are common in nanosurgery and may easily collapse within one second [16,43]. The bubble may also retain non-condensable gas which is derived from the fracture of biomolecules in the focal zone, prolonging its existence [16]. However, in our experiments, no bubbles were observed by fixation immediately after irradiation. If a short-lived bubble collapses before fixation could occur, some physical evidence of this may remain since the material is not completely elastic [18], and the transient bubble temperature can be significant [16]. In considering the nature of the dark material observed in Fig. 8, we should compare the results to those observed when fixation was used (in section 3.3, see Fig. 6). Not only is a bubble effect visible, but dark shaded electron-opaque areas are also visible in Fig. 6, surrounding the white ablated hole. In particular, Fig. 6 section number 11 clearly shows that a concentrated region of dark material exists at the boundary of the ablated region. The dark material observed in Fig. 8 may consist of biomolecules that were modified, either directly by the laser action, or as a result of the expanding bubble which then collapsed. The black material was stationary in the nucleus long enough that the post-irradiation fixation protocol could be used to image the laser effect. The dark material in the center of the focal spots in Fig. 8 and the dark matter surrounding the ablation sites shown in Fig. 6 (where fixation was used) may consist of the same material. Further study of these effects would be required to conclude the nature of these intracellular changes, but the current results show that even without pre-fixation, TEM has the potential to investigate nanoscale laser-induced changes in cells.

3.6 Fixation effect on laser cell interaction

Finally, in considering the necessity of fixation, we must consider the possible effects of chemical fixation on the laser-cell interaction. The modification of the photophysics of the interaction between the laser irradiation and cell sample may occur via the increase in linear or multiphoton absorption due to the presence of glutaraldehyde molecules in the laser focus. Glutaraldehyde molecules also cross-link proteins in the cell, and may modify the cell’s absorption properties. However, due to the low concentration of glutaraldehyde molecules, and the fact that cell absorption is low at near-infrared wavelengths, we can predict that the main effect of the fixation on the laser-cell interaction is physical; the cross-linked proteins constrain the formation, expansion, or collapse of laser-induced bubbles and restrict the diffusion of any modified cell components. Indeed, it has been reported that the maximum bubble size is often restricted by the elasticity of surrounding materials at the focal spot [18,44]. We observed ablation sizes in the cytoplasm that were larger than the observed ablation sizes in the nucleus, which may be due to the fact that the density of the nucleus is higher than that of the cytoplasm. This last property of restricting diffusion and movement in the cell provides the useful benefits of fixation for investigating laser-cell interactions. Other researchers have also reported physical evidence of subcellular laser ablation in living cells without using pre-fixation [26]. Their experiment used 532 nm scanning irradiation, with a 7 ns pulsewidth and a pulse energy of 300–500 nJ. The high pulse energy and the increased absorption in the sample at the 532 nm wavelength combines to produce physical modifications of the sample that does remain visible at the irradiation site. By contrast, our results show that the laser-induced modification at the irradiation site by multiphoton absorption at the threshold of ablation was not observed without fixation.

3.7 Implications for intracellular nanosurgery by femtosecond laser irradiation

Nanosurgery of living cells or dissection of cell components is often performed with femtosecond lasers similar to the laser source used in this manuscript, which typically have advantages over other laser sources including localized energy deposition and deep penetration though samples [6,10,16,37]. For these laser sources, the desired effects on the sample are generally achieved via amplified pulses at kHz order repetition rates, or by long trains of femtosecond pulses of nJ order pulse energy and high repetition rate. The results presented here use trains of high repetition rate femtosecond pulses. The pulse width is not especially critical for pulse widths of one to several hundred femtoseconds but does influence the relative amounts of multiphoton ionization versus other types of ionization [48,49]. The pulse width also affects some of the optical parameters by which different experiments are compared (e.g. the irradiance). Here, the pulse width used in experiments (estimated at 190 fs) results from an 80 fs pulse at the laser output traveling through the optical components described in section 2. This pulse width will then be comparable to other experiments using similar equipment. The exposure time used here of 8 ms is chosen since it requires only the laser source and a mechanical shutter to produce truncated trains of pulses of approximately 650000 in number. The effect of a large number of pulses complicates the analysis of the laser-cell interaction, but is necessary to induce observable effects at low pulse energies. The repetition rate here (82 MHz) is also typical for high repetition rate operation of femtosecond laser sources, and the pulse energies used are easily within reach of commonly available Ti:Sapphire sources.

A notable difference between the experiments reported here and those in other investigations of cell nanosurgy (several of which are summarized by Vogel et al. [16]) is that the laser beam in our experiments was not scanned. This means that when comparing laser power, pulse energy and the consequent effects in the cell, it should be noted that the thresholds will be higher for the non-scanned case since the total exposure time for the experiment in a given cell is short (in our case ~8 ms multiplied by the number of irradiation sites). If there is a statistical probability of an effect occurring in the cell at a specified pulse energy, then it is clear that such effects will be observed more often in the case when the beam is scanned once or continuously through the specimen [6]. The review of nanosurgery effects by Vogel et al. [16] contains a number of useful parameters by which to compare femtosecond laser nanosurgery and we therefore provide similar parameters for the experiments reported in this manuscript to aid comparisons of the results. The single point exposure of 8 ms is somewhat equivalent to a dwell time of 8 ms, albeit with the consideration that the threshold for a single point exposure will be less than the threshold for scanning the beam. The laser parameters here (as mentioned in section 2) are: a pulse duration of approximately 190 fs, repetition rate of 82 MHz, and an NA of 1.0 in water. The number of pulses per focal spot is then 650000. The threshold pulse energy was found to be 0.48 nJ with a radiant exposure of 0.068 J/cm², for the 952 nm diameter airy spot size (not full-width half maximum) calculated by considering the 1.0 NA and 780 nm wavelength. The irradiance is another parameter commonly used to compare high intensity laser effects, although it does have the disadvantage of being dependent on the pulsewidth and therefore subject to discrepancy by uncompensated dispersion of the pulse. For our estimated pulsewidth of 190 fs, the irradiance is 0.36×10¹² W/cm², when calculated using the average irradiance within the spot diameter. For ease of comparison with similar experiments [16], the peak irradiance should be approximated by multiplying the averaged irradiance by 2. Doing so gives a (pulsewidth dependent) irradiance value for comparison with table 2 in reference [16] of 0.71×10¹² W/cm². The normalized irradiance (i.e. the irradiance divided by the irradiance threshold for optical breakdown) is therefore roughly 0.11. Although there are other complicating factors such as aberration of the spot size and inaccuracies in the estimation of the pulsewidth, the normalized irradiance indicates that the visible ablation effects reported in this manuscript are occurring at approximately 10 times lower irradiance than the reported optical breakdown threshold for a single pulse. This is because the large number of pulses can sequentially modify the sample. The observations show similar thresholds to those measured by other groups in similar experiments [6,10]. As a comparison, the normalized irradiance threshold [16] for intracellular chromosome dissection by an 80 MHz 170 fs pulse train focused by a 1.3 NA lens (with a focal spot that is correspondingly smaller by a factor of approximately 1.8) was 0.15 [6]. Using the same system, the normalized irradiance for gene transfection resulting from laser induced membrane perforation was 0.25 [10].

The high resolution of the TEM imaging experiments can potentially improve the existing understanding of the onset of laser ablation and physical modification of the sample. It is interesting that although these experiments were carried out using long trains of high repetition rate femtosecond pulses with pulse energies below the optical breakdown threshold, where bubble formation is not predicted to occur [16,50], bubble formation is evident in the TEM images presented here. The current results indicate that the physical effects of laser dissection at pulse energies of less than 1 nJ may include bubble effects that are more disruptive than previously indicated. However, the role of glutaraldehyde in the physics of the laser-cell interaction and also in the mechanics of bubble expansion will require further study. The clear presence of bubbles is of itself noteworthy, since bubble formation must by definition be associated with a significant change in the temperature and/or pressure of the region in or around the focal spot [16], with possible peak temperatures transiently reaching as high as several hundred °C for bubble formation of 400 nm diameter. Most calculations of the temperature rise for conditions similar to the experimental parameters used here report low or negligible temperature rise resulting from high NA focusing of femtosecond pulse trains with pulse energies on the order of 1 nJ or lower [16,35,51]. It will be interesting to further investigate the conditions that lead to the bubble effects at low pulse energies.

As reported by other groups, [18,43,44], bubbles formed within a dense media can be constrained in growth and/or movement capability. Our experiments showed a significant difference in the ablation size between irradiation of the cytosol and irradiation of the nucleus. Moreover, the ablation morphology was less predictable in the case of cytosol irradiation. While there is a possibility of a polarization effect, the main mechanism behind the anisotropic appearance of the cytosol ablation zones is more likely to be due to the anisotropic distribution of material surrounding the ablation zone. A clear example of this occurred where the ablation zone was situated in the cytosol, in close proximity to the nuclear membrane, where the local density is higher. For experiments using glutaraldehyde where the laser ablation zone occurred on or immediately adjacent to the nuclear membrane, we saw ablation sites where the morphology of the bubble appeared to be constrained by the nuclear membrane. For these sites, we subjectively divided them into one of two groups: the first group contained ablation zones where the bubble confinement either did not occur or was not clear, the second group contained ablation zones where the bubble appears to be restricted in growth and shape by the nuclear membrane. The results are shown in Fig. 9. For pulse energies of 0.60 nJ, 5 out of 9 sites in close proximity to the nuclear membrane showed confinement, and for pulse energies of 0.72 nJ, 3 out of 5 sites showed confinement. Similar effects were indicated in the cytosol, where mitochondria, or microtubules appeared to constrain the bubble, but due to the difficulty and inaccuracy in subjective classification of this data as being constrained or not constrained, the data is not shown. Although ablation of higher density cell components is clearly possible (see for example Fig. 9(a) where the nuclear membrane appears to be dissected), the constrained bubbles (Fig. 9(e-i), for example) indicate that for bubble-related laser ablation effects inside a cell, the mechanical properties of adjacent material in the cell may strongly affect the ablation itself, as reported by other groups [18,43,44] and that the confinement may act in an anisotropic manner.

 

Fig. 9. For ablation sites adjacent to the nuclear membrane, the bubble often appeared constrained by the nucleus. For pulse energies of 0.60 nJ (50 mW average laser power), 9 sites were observed in close proximity to the nucleus. 4 sites not clearly showing constrained bubbles are labeled (a-d), and 5 sites which show constrained bubbles are labeled (e-i). In all images, the nucleus side is labeled by ‘N’. For pulse energies of 0.72 nJ (60 mW average laser power), 5 sites were observed in close proximity to the nucleus. 2 sites not clearly showing constrained bubbles are labeled (j) and (k), and 3 sites which show constrained bubbles are labeled (l-m).

Download Full Size | PDF

4. Conclusion

We have observed the morphology and power dependence of laser ablation in the cytoplasm and nuclei of cultured cells. By fixation before irradiation, laser-induced changes were confined to the region of the laser focus, allowing precise 3d reconstruction of the laser affected zone with nanoscale resolution. The threshold for the observable effects was 0.48 nJ for the cytosol regions and 0.6 nJ for the nuclear regions, corresponding to tightly focused irradiation from a Ti:Sapphire laser of 40 and 50 mW, respectively. The measured threshold energy for observable physical effects in the cell with glutaraldehyde fixation was close to the energy required for photoinduced Ca²⁺ wave propagation (without fixation) for the same cell type, which was measured to be ~0.36 nJ [12]. This implies that photoinduced Ca²⁺ wave generation occurs concurrently with nanoscale subcellular ablation that occurs inside the laser focal spot, but is not easily observable by conventional microscopy. With glutaraldehyde fixation, we observed ablation zones which exist as bubbles, and which had a pulse energy threshold for formation. Under threshold, however, we could not find any direct evidence of the laser effect. This means that at least in terms of TEM observations, the threshold behavior of the physical modification of cells by femtosecond pulse trains of sub nJ pulse energy appears to be an all-or-nothing effect. In terms of comparison with other research in laser ablation of biological samples, numerical calculations estimate a threshold energy density of 0.33 J/cm² with a single pulse of 100 fs pulse width and 800 nm wavelength [16]. In our case, a pulse train of 0.48 nJ pulses and a 1.0 NA lens corresponds to 0.068 J/cm², which is about five times smaller than the calculated ablation threshold for a single pulse. The difference can be largely attributed to the accumulation of sequential laser-induced effects by the pulse train. The absorption in the sample may change over subsequent pulses by carbonization and/or chemical modification of the sample by the pulse train. These effects are intriguing, and combined with the fact that bubbles were clearly observed at sub-nJ pulse energies, indicate that further work should be done to understand the intricacies of long pulse train irradiation effects in cells as well as to elucidate any effect glutaraldehyde has on the absorption properties of cells.

The size of the ablated volume was observed to vary significantly depending on the location of the laser focal spot in the cell. At threshold, the nuclear irradiation sites showed lateral dimensions of 200 nm, and an axial length of 1 µm, whereas the cytoplasm irradiations sites had dimensions between 500 nm and 1 µm laterally and show hints of polarization dependence. The axial length of the cytoplasm irradiation sites was as large as 1.8 µm. The size of the laser affected zone in cytoplasm was larger than considerations based on multiphoton physics alone would indicate. Although at pulse energies of 0.3 to 0.7 nJ, multiphoton absorption by the cell is considerable and would point toward increased localization of the absorbed power, the subsequent bubble effects can increase the size of the laser affected zone. The observed diameters of laser ablation sites and the clear boundary between the laser affected zone and the bulk cell region indicate that a micro-bubble mediates the laser effect in the cell. The fact that no effects were seen at pulse energies below bubble formation thresholds indicates that, at least for fixed cell samples, bubble formation occurs before any other physical modification of the cell target. Although the effect of glutaraldehyde on the laser-cell interaction requires further study, we hypothesize that the main effect, if any, of the presence of glutaraldehyde will act as mechanical modification of the structural properties of the intracellular components. Additionally, the results showing that the nuclear membrane appears to constrain the bubble indicates that the mechanical constraints may be largely due to the cell itself and not the glutaraldehyde

When fixation was performed post-irradiation, no local physical evidence of laser irradiation in the cell cytosol could be observed by TEM imaging. We can conclude that, for these exposure conditions, the laser-induced modification of the cell cytosol is insufficiently strong to be visible by TEM, or more likely, that any bubbles induced by laser irradiation are able to collapse and any other potentially visible effects diffuse away from the focal zone before chemical fixation can immobilize them. When the laser was moved to the nucleus and fixation was again performed after irradiation, residual electron-dense material was observed in the centre of the laser irradiation site resulting from denatured proteins or other nuclear structures. Due to the abundance of proteins in the nucleus, we attribute this to cross-linking of nuclear proteins around the irradiation site. We expect that this study will be useful in the future to investigate the physical changes which precede cell responses to laser irradiation, and may be applied to quantifying the nanoscale changes in cell ultrastructure that result from laser interactions and the onset of bubble formation in femtosecond laser nanosurgery.