1. Introduction

Nonlinear laser microscopy for nano- and microprocessing has long been plagued by dispersive effects that limit the shortest pulse duration achievable at the specimen plane. Negative pre-chirping with grating- and prism-pairs is a viable solution in microscopy systems seeded with pulses >30 fs [1-4]. Still, such compressors found little acceptance in the microscopy community because of their size, complexity and alignment sensitivity. Furthermore, grating- and prism-based compressors are not accurate enough to handle pulses in the 10 to 20-fs range that call for higher-order dispersion compensation [2]. Recent advances in the design and manufacture of dispersive mirrors [5] allowed increasing the group delay dispersion (GDD) introduced by these multilayer filters by almost one order of magnitude. This opens the way towards all-reflective dispersion management in complex optical systems [6,7]. Inherently compact, mirror-compressors allow for accurate compensation of higher-order dispersion (HOD) terms and are substantially insensitive to beam pointing drifts.

So far, femtosecond laser microscopes without any dispersion compensation at a typical pulse width of 200–400 fs have been employed to perform nano- and microprocessing of biological targets based on multiphoton ionisation and plasma formation at transient TW/cm² light intensities. The efficiency of the destructive multiphoton effects scales with the increase of the transient peak power and the reduction of the pulse width. Focussing optics with high numerical aperture and low pulse energies are required to induce multiphoton effects in a sub-femtoliter focal volume and to avoid collateral damage in the microenvironment due to photodisruptive effects based on the formation of bubbles and shock waves [8,9].

König and coworkers used nanojoule 270 fs laser pulses at 80 MHz to perform chromosome dissection, optical knock-out for contamination-free selection of living cells, intracellular and intratissue surgery, DNA nanoprocessing, and ocular refractive surgery [10-15]. Watanabe [16] used similar laser systems to optically knock-out intracellular mitochondria and Kohli et al. [17] as well as Pavone et al. [18] reported on membrane and microtubulus surgery, respectively, with femtosecond near infrared laser pulses. Yanik et al. [19] studied the functional regeneration after femtolaser axetomy.

Femtosecond laser pulses have been also used to demonstrate efficient targeted transfection of Chinese hamster ovary cells [20,21] by transient opening of the cellular membrane, the diffusion of the foreign DNA into the cytoplasm and the passive introduction into the nucleus when the nuclear envelope is disintegrated during cell division. Typically, mean powers of about 100 mW (~1 nJ) and high NA objectives were required to induce the transient holes which should be in the submicron range to ensure a successful self-repair of the membrane damage. It was shown that subwavelength, sub-100 nm cuts and holes can be realized in biological targets with near-infrared tightly-focussed laser beams due to the confined multiphoton processes in small central region of the illumination spot [10, 13-15]. When using low NA 0.85 objectives in an upright beam path where the laser beam has to be transmitted through the medium, the transfection efficiency drops down to about 50% likely due to relatively large laser-induced holes of 2.2 – 10 µm in the cell membrane [22].

However, significant problems were faced in the case of transfection of a variety of stem cells. Whereas the transfection efficiency by optoporation and by non-optical transfection methods may be up to 70% and higher [23] in the laboratory routine in certainly commonly used cell lines, efficient transfection failed to be effective in stem cells. Typical transfection rates are 0.5 – 12% for human smooth muscle cells with conventional methods such as electro-poration, viral transport, liposome-mediated transfer and calcium phosphate treatment and up to 29% when using a combination of electroporation with chemical methods (amexa chemical Nucleofector technology) [23]. After optimisation, even higher efficiencies were obtained for the case of human mesenchymal stem cells which survived the treatment. However, nearly 50% of the stem cell population died [24].

Stem cells are major research objects in medicine and biotechnology and may revolutionize current therapy. Genetically modified stem cells can be used for the production of stable and pure stem cell lines and to transfer genetic information into tissue where they can be employed to eliminate deceased genes, to add new functions to cells such as the production of immune system mediator proteins, etc.

Here we report on preliminary studies on the optical transfection of individual human stem cell of interests with femtosecond laser pulses at extremely short pulse duration of less than 20 fs and mean powers of even less than 7 mW.

2. Results

2.1 Femtosecond laser scanning microscope with high-order dispersion compensation

An ultracompact scanning nonlinear microscope with galvoscanners for beam scanning and piezodriven focussing optics (FemtOcut™, JenLab GmbH, Jena) equipped with highly-dispersive, large-NA objectives (Zeiss EC Plan-Neofluoar 40x/1.3, Plan-Neofluar 63x/1.25 oil) was directly seeded with a Kerr-lens passively mode-locked dispersive mirror based femtosecond Ti:Sapphire oscillator (Synergy Pro, Femtolasers Produktions GmbH, 0.67 MW peak power, 75 MHz, M²<1.3, 600 mW mean power output (8 nJ), cw pumped @ 5 W, 532 nm pump Spectra Physics Millenia V). Fig. 1 demonstrates the output laser parameters 12 fs, 80 nm FWHM and emission maximum at 792 nm by the measurement of the autocorrelation function and the spectrum. This corresponds to a time-bandwidth product (TBP) of 0.45 assuming a sech^2 shaped pulse.

 

Fig. 1. Spectrum and calculated bandwidth-limited second order interferometric autocorrelation of the pulses emitted from the Ti:Sapphire oscillator revealing values of 12 fs pulse width, 80 nm FWHM, and 792 nm emission maximum.

Download Full Size | PDF

In order to compensate the dispersion of all optical systems of the microscope including the beam expander, polarizer for beam attenuation, tube lens, the objective, filters etc., a pair of highly-dispersive mirrors (HDMs) was integrated in the laser housing. 20 bounces off these dispersive mirrors were enough to pre-compensate the dispersion of the full system including quadratic and cubic phase distortions. The insertion losses of the mirror compressor were <10%. The in situ pulse duration was measured by means of a second order interferometric scanning autocorrelator (FEMTOMETER, Femtolasers Produktions GmbH) at the focus of the two objectives. At the focus, a non-linear photodiode (NL-PD) was placed to record the interferometric signal while the Michelson interferometer part of the autocorrelator was inserted into the beam path between the laser and the microscope (Fig. 2(a)). Pulse durations <20fs were measured with both objectives (Fig. 2(b)). Anti-reflection-coated glass plates were employed for fine adjustment of the GDD. The microscope was used in the two-photon fluorescence excitation mode at mean powers in the microwatt range for nondestructive imaging of the stem cell of interest and to monitor the biosynthesis of fluorescent proteins after laser exposure as well as for nanoprocessing in two exposure modes: (i) scanning of a region of interest (ROI) and (ii) by single point illumination where the galvoscanners were fixed to a point of interest. A CCD camera attached to the side port enabled on-line imaging of the cells, the laser beam and the formation of plasma-filled cavitation bubbles. A photomultiplier with fast rise time attached to the front port was employed in combination with time-correlated single photon counting to image two-photon fluorescence.

 

Fig. 2. (a). Experimental setup and photograph. The beam emitted by the 12-fs oscillator passes a pair of HDM before coupling into the scanning microscope via a dispersion-balanced Michel-son scanning interferometer.

Download Full Size | PDF

 

Fig. 2. (b). Autocorrelation functions at the focal plane. In-focus second order interferometric autocorrelations recorded for the objectives Zeiss Plan-Neofluoar 40x/1.3 and Plan-Neofluar 63x/1.25 indicate pulse durations of 19.25 fs (TBP of 0.74) and 16.50 fs (TBP of 0.63), respectively. 20 bounces with HDM setup using the 40x Zeiss objective.

Download Full Size | PDF

2.2 Non-destructive two-photon autofluorescence imaging of stem cells

Human salivary gland stem cells (hSGSC) and human pancreatic stem cells (hPSC) were transferred into special 0.5 ml medium-filled miniaturized cell chambers (MiniCeM-grid, JenLab GmbH, Jena, Germany) consisting of two 170 µm thick glass windows and a silicon ring as spacer in a plastic housing. One of these windows has an etched 200-field-grid as marker. Dulbecco’s modified Eagle’s medium (DMEM, GIBCO, Germany) supplemented with 15 % fetal calf serum Gold (FCS-Gold, PAA Laboratories GmbH), 100 U/ml penicillin and 100 mg/ml streptomycin was used. Laser exposure was performed through the glass window on single stem cells attached as monolayers to the etched glass bottom of the microscope chamber.

Single stem cells were exposed to sub-20 fs laser pulses at different mean power levels and exposure times by x,y-scanning of a 250×250 µm² field (512×512 pixels). Interestingly, 400 µW mean power (5 pJ pulse energy) and 7-15 µs beam dwell time per pixel were sufficient to induce two-photon excited autofluorescence in the mitochondria of human salivary gland stem cells mainly based on the non-linear excitation of the reduced coenzyme nicotinamide adenine (phosphorylated) dinucleotide NAD(P)H. Figure 3(a) demonstrates one typical two-photon autofluorescence image with a high signal to noise ratio. No signs of photoinduced cell destruction were monitored 1 hour after exposure even after 10 scans. The membrane integrity and cell viability was probed with the fluorescent marker ethidium bromide (10 µM, Molecular Probes, Netherlands). This DNA–binding dye cannot penetrate across the intact cell membrane and therefore stains cells after membrane damage only.

Interestingly, higher mean powers up to 2 mW were required to obtain the same autofluorescence counting rates in the case of human pancreatic stem cells which means roughly a factor of 25 less fluorescent coenzyme molecules due to the squared dependence of fluorescence intensity on power. A further 5fold increase of the mean power resulted in intranuclear ethidium bromide fluorescence accompanied with cell destruction. Therefore, non-destructive high-resolution two-photon autofluorescence imaging can be performed at short beam dwell times within a certain power window which depends on the stem cell type.

 

Fig. 3. (a) Two-photon autofluorescence image of human salivary gland stem cells. The major endogenous fluorophore is the reduced coenzyme NADH. 250×250 µm² image. (b) Cell fluorescence after optoporation of two cells and the application of ethidium bromide. Nonexposed cells did not uptake the fluorescent probe. (c) Spectral imaging of optoporated cells reveals the blue/green NADH fluorescence of mitochondria and the red fluorescence of intranuclear ethidium bromide.

Download Full Size | PDF

2.3 Optoinjection of fluorescent molecules

In order to realize highly confined optoporation, at first the morphology of a particular stem cells was imaged by ROI scanning and the detection of transmitted light and two-photon autofluorescence. Afterwards, the laser beam was “parked” (single point illumination) at an area of the cellular membrane. Now the mean power and the beam dwell time were increased. A trace of 3.75 million pulses was applied to the target when opening the shutter for 50 ms. Ethidium bromide was given to the medium in different experiments immediately after laser exposure. Interestingly, a low mean power of 5 mW (66 pJ) was sufficient to provide a transient port into the laser-exposed membrane for the diffusion of ethidium bromide into the cell and further into the nucleus. Neighbouring cells remained unaffected by the laser exposure and did not uptake the fluorescent probe. Fig. 3(b) shows one example of a two-photon image of two optoporated cells after introduction of the membrane-impermeable fluorophore. Fig. 3(c) depicts the spectrum of the dye ethidium bromide and of the endogenous fluorophore NADH. When increasing the power level to values higher than 14±2 mW, typically the formation of microbubbles with a diameter of more than 5 µm was observed. In part, these cells changed morphology significantly and died.

2.4 Targeted transfection

Transfection was performed by using the same parameters as in the case of successful optical injection of ethidium bromide such as 5-7 mW mean power and 50-100 ms exposure time. The transfection was performed with the DNA plasmid vector pEGFP-N1 from Clontech (4.7 kb, molar weight 3 MDa). 0.2 µg of the plasmid were administered to the extracellular 0.5 ml medium. The diffusion of the non-fluorescent plasmid through the transient opening could not be detected. We did not observe any spontaneous, non-laser induced formation of green fluorescent proteins in our cell chambers with approximately 10,000 cells per chamber. After laser exposure the cell chambers were transferred to an incubator and maintained at 37°C in a 5% CO₂ humidified atmosphere. The laser-exposed stem cells and their neighbouring cells were tracked up to 8 days after laser treatment by phase contrast microscopy and fluorescence imaging using the same microscope in the non-destructive two-photon excitation mode. Tracking was possible due to the use of the grid pattern at the glass bottom. For additional proof of optically transfected stem cells, the ZEISS LSM 510-Meta microscope was employed for spectral analysis of the protein fluorescence (Fig. 4(b)). The formation of green fluorescent proteins (GFP) in the case of a successful transfection was monitored 24 hours after exposure for some laser-treated cells and 48 hours for all of them. All optoporated cells survived. Some transfected cells underwent cell division with a typical reproduction time of 48–72 hours. The daughter cells exhibited also green fluorescence with a maximum at 507 nm as probed by spectral imaging. A transfection efficiency (number of green fluorescent cells to number of exposed cells 48 hours after exposure) of 70-80% was obtained for both cell lines (n=30, first experiment hPSC: 21 green fluorescent cells, second experiment 3 days later: 24 green fluorescent cells, hSGSC: 24 green fluorescent cells). Fig. 4(a) shows green fluorescent human salivary gland cells 24 hours, 3, and 8 days after laser exposure. The protein fluorescence indicates the successful GFP biosynthesis after introduction of the plasmid into the cytoplasm through the nanopore and the uptake into the cellular DNA as well as the reproduction of laser-exposed cells. For comparison, CHO cells were laser-exposed. A 90% transfection rate was obtained (n=100).

 

Fig. 4. (a). Transfected human salivary gland stem cells 24 hours, 3 days and 8 days after laser optoporation. The two green fluorescent cells are daughter cells of one laser-exposed cell. The cell clone at day 8 was formed after transfection of 2 stem cells.

Download Full Size | PDF

 

Fig. 4. (b). Typical spectrum of intracellular GFP within transfected stem cells

Download Full Size | PDF

3. Discussion

State of the art two-photon fluorescence imaging is performed with pulses of 200-300 fs or even around 1 ps pulse length depending on the two-photon microscope manufacturer. The signal in two-photon microscopy depends on a P²/τ relation with P as laser power and τ as pulse width. Because of this linear pulse length relation, picosecond laser microscopes require just a relatively small increase in laser power to obtain similar images as for femtosecond laser microscopes, e.g. 1 ps systems require 3times higher power than 110 fs systems.

However, in the case of nanoprocessing based on multiphoton ionisation processes where several photons are involved (e.g. 5 NIR photons and 6.5 eV, respectively, are required to induce an optical breakdown in water [8,9]) the shortening of the laser pulse width would significantly improve the desired destructive effect. The non-distored delivery of sub-15 fs through a microscope objective and even sub-10 fs pulses by compensation of quadratic and cubic dispersion terms has been demonstrated and employed so far for nonlinear imaging [25,26]. No application of sub-20 fs laser pulses for nano- and microprocessing in living cells has been reported. As shown in this work sub-20 fs laser scanning microscopes can be realized and employed for nanoinjection and optical transfection of human stem cells. Interestingly, low mean powers of 5–7 mW (66-93 pJ @ 75 MHz) which is more than one order less power as in current femtosecond laser nanoprocessing were sufficient to realize transient high TW/cm² intensities and successful stem cell manipulation. The use of a sub-10 mW power guarantees also the absence of any destructive thermal effects and disturbing trapping effects. The potential use of low power systems may opens also the chance for the manufacturing of miniaturized sub-20 fs laser systems for nanoprocessing and imaging without bulky expensive pump lasers.

In addition, the novel dispersion technology overcomes the problems of beam positioning fluctuations observed in femtosecond laser systems based on prism technology. The absence of beam shifts may also contribute to the high transfection efficiency obtained in these sub-20fs studies. Two human pluripotent stem cell lines were chosen. The human stem cells obtained from a human pancreas possess a remarkable potential for self-renewal and multilineage differentiation including the development into specialized cells of all three germ layers [27, 28]. So far, no transfection studies have been conducted with these new type of stem cell sources.

Transfection of stem/primary cells is a major problem in up-to-date technology based on chemical, mechanical or electrical means. The non-invasive gentle creation of a transient nanohole in the cellular membrane by low mean power sub-20fs laser pulses without any collateral damage and disturbance of the self-repairing potency overcomes this problem. After optoporation, the extracellular foreign DNA diffuses into the cytoplasm before self-repair and enters the nucleus during cell division when the nuclear envelope is disintegrated. All laser-exposed cells in these preliminary studies survived and the majority of the treated human stem cells have been transfected. The cells exhibited normal reproduction behaviour (cell division after 2-3 days) and the daughter cells became also green fluorescent.

Femtosecond laser manipulation of stem cells opens new ways for stem cell based gene therapy, the generation of stable cell lines, and transplantation technology. In particular, the induction of specific genes may improve the efficiency of controlled differentiation. Cell types of interest may be collected using specific marker genes and antibiotic resistance genes. Tracking of stem cells derived after engraftment could be facilitated by tracer genes. Furthermore, cells can be reprogrammed by injection corrective genetic materials, so appropriate type of cell can be created to develop replacement tissues.

In conclusion, non-contact femtosecond laser transfection with sub-20 mW and sub-20 fs ultracompact nonlinear microscopes may become a novel promising technology for optical cell manipulation. In the future more studies on a large number of stem cells and different stem cell types have to be conducted including various DNA plasmids.