2.1.

Experimental Procedure

The setup is sketched in Fig. 1. The light source, a mode-locked $\mathrm{Ti}$ :sapphire laser (Mira 900, Coherent, Santa Clara, CA), was pumped with a frequency-doubled solid state laser (Verdi V, Coherent) and tuned to emit about 200-fs pulses with a repetition rate of $76\mathrm{MHz}$ at a center frequency of $803\mathrm{nm}$ . The laser emission was then coupled into the tapered fiber with a $10\times 0.3\mathrm{NA}$ objective after passing through a Faraday isolator that prevented back-reflections into the $\mathrm{Ti}$ :sapphire cavity. Coupling $650\mathrm{mW}$ into the tapered fiber resulted in a perfect ${\mathrm{TEM}}_{00}$ spatial shape, with a supercontinuum output spectrum according to Fig. 1 (inset 1) and a total power of $280\mathrm{mW}$ . Although the coupling objective and the approximately $10\mathrm{cm}$ of SMF 28 fiber lead to a significant temporal pulse broadening, the peak energy is sufficient for the nonlinear optical effects that result in the reported generation of supercontiuum light.¹⁸ The resulting laser beam was spectrally decomposed with a standard dichroic mirror (Chroma, Rockingham, VT, USA) into a visible part from about $430$ to $700\mathrm{nm}$ and a near-infrared (NIR) part ranging from about $700$ to $1300\mathrm{nm}$ . The visible part of the light was coupled into the CLSM through a fiber with a $4\text{-}\mathrm{\mu }\mathrm{m}$ core diameter and single-mode properties in the visible spectrum. This fiber was coupled to the AOBS, where an arbitrary wavelength could be selected between $450$ and $633\mathrm{nm}$ . Selecting the wavelengths 458, 476, 488, 514, 543, and $633\mathrm{nm}$ , which were the wavelengths used for confocal imaging, we measured a total intensity of $62\mathrm{\mu }\mathrm{W}$ before the objective. For all images, a $40\times 1.25\mathrm{NA}$ (Leica Microsystems, Bensheim, Germany) oil immersion objective was used.

The NIR part of the light was coupled into the optical path of the microscope through the standard NIR port. Since this light was not spectrally filtered by the AOBS, the total power of the supercontinuum pulsed NIR light was $11.6\mathrm{mW}$ at the sample, which was sufficient to perform multiphoton microscopy, even with the picosecond pulses. We measured the pulse length to be between $1$ to $5\mathrm{ps}$ , depending on the spectral range.¹⁸ As reported previously,^{22, 23, 24} a pulse duration of picoseconds is indeed sufficient to reach the peak intensities required for multiphoton processes if the average power is increased accordingly. The relation between the peak power and the time average power simply reads $P_{\mathrm{peak}}=0.56*P_{\mathrm{ave}}∕(\tau *f)$ ,²³ where $\tau$ is the pulse duration and $f$ is the pulse frequency. For the 5-ps pulse and an average power of $11.6\mathrm{mW}$ this calculates to a peak power of about $17\mathrm{W}$ . Assuming a spot size of $<1\mathrm{\mu }{\mathrm{m}}^2$ , this gives a peak intensity at the sample of at least $1.7\mathrm{GW}∕{\mathrm{cm}}^2$ , which is sufficient for multiphoton excitation.²⁴

2.2.

Confocal Imaging with Arbitrary Excitation Wavelengths

To demonstrate the compatibility of our novel method, we used standard dyes. Standard excitation wavelengths were applied to record confocal multicolor images. However, within the accessible spectral area there are no physical limitations on the wavelengths chosen. With the described setup, up to eight different wavelengths can be used simultaneously with the ability to switch between wavelengths within milliseconds. Figure 2 shows a neuron-like cell (NG108-15, ATCC, Manassas, VA)²⁵ fixed and stained for actin (red) and microtubules (green). In both channels the image shows details with diffraction-limited resolution. The images were recorded using the 488-nm and 543-nm lines from the supercontinuum white-light source. To gather information about the out-of-focus planes, we performed a $z$ scan of different sample cells. Figure 3 presents the maximum-projection rendering of these cuts, which compares the values of a certain pixel coordinate along all recorded $z$ -scan images and displays the maximum value. The recorded image displays the same resolution and details as obtained with a conventional excitation source.²⁶

Fig. 2

Images of (a) the actin (red, stained with TRITC-tagged phalloidin) and (b) the microtubule (green, immunolabeled with Alexa 488 coupled secondary antibodies) cytoskeleton of an NG108-15 cell recorded with the supercontinuum white-light source. For better evaluation of the different channels, the two fluorescent markers are presented as single images and as an overlay (c). Images were recorded by averaging over eight scans of each image line (line average), and two scans of total frames (frame average). Scale bar: $10\mathrm{\mu }\mathrm{m}$ .

Fig. 3

Maximum projection of a $z$ series recorded with the supercontinuum white-light source. The actin (red) and the microtubule (green) cytoskeleton of an NG108-15 cell are fluorescently labeled. The projected $z$ series consists of 15 spatial cuts in the $xy$ plane, separated by $1\mathrm{\mu }\mathrm{m}$ from each other in the $z$ direction. Images were recorded with a line average of eight and a frame average of two. Scale bar: $25\mathrm{\mu }\mathrm{m}$ .

To reveal the CLSM’s unaltered 3-D capabilities while using the supercontinuum white-light laser source, a $30\text{-}\mathrm{\mu }\mathrm{m}\text{-}\mathrm{thick}$ coronal section from a paraformaldehyde-fixed rat brain, stained with histochemical methods for vessels (green) and astrocytes (red), was imaged. Figure 4(a) shows the maximum projection of a $z$ scan. With these data, the usual 3-D operations of the microscope’s software can be exploited, yielding 3-D visualization of the sample, as the color-coded 3-D illustration [Fig. 4(b)] demonstrates.

Fig. 4

Imaging of histochemically labeled rat cortical vessels and astrocytes. (a) Maximum projection of a $z$ series of a rat brain section displaying vessels ( $\mathrm{Cy}2$ , green) and astrocytes ( $\mathrm{Cy}3$ , red). The series represents 10 recordings of the $xy$ plane, separated by $1\mathrm{\mu }\mathrm{m}$ along the $z$ direction. Images were recorded with a line average of eight and a frame average of two. Scale bar: $50\mathrm{\mu }\mathrm{m}$ . (b) Color-coded 3-D projection of vessels and astrocytes presented on the left. Due to the high image quality, usual 3-D operations can be performed. The color table translates the colors into the relative $z$ depth of the picture. The image was calculated from the merged red and green channel data of Fig. 5(a), with a perspective view of a 40-deg rotation around the horizontal axis. Scale bar: $50\mathrm{\mu }\mathrm{m}$ .

2.3.

Multiphoton Microscopy

To illustrate the multiphoton capabilities of the supercontinuum white light, the picosecond-pulsed NIR radiation was successfully used for multiphoton excitation. To simultaneously excite both the tetramethylrhodamine isothiocyanate (TRITC) labeled actin and the Alexa 488 stained microtubule cytoskeleton, no further filtering of the two-photon excitation light for specific wavelengths was necessary. Rather, the entire NIR spectrum from $700$ to $1300\mathrm{nm}$ was used. For the multiphoton imaging, the imaging pinhole was opened completely in order to maximize the detection efficiency and because the described nonlinear effect excites samples only at the focus. Hence, with multiphoton microscopy, the pinhole is not needed to discard out-of-focus and scattered emission light. Figure 5 shows the maximum projection of NG108-15 cells, prepared as described. Although a novel technique was used as pulsed radiation source, the image quality was comparable to conventional two-photon microscopy or CLSM.

Fig. 5

Fluorescence image of an $xy$ section of two NG108-15 cells recorded with two-photon microscopy. Images were recorded with a line average of eight and a frame average rate of three. With the multiphoton technique, the same details are observable as images recorded with the visible part of the supercontinuum white light in Figs. 2 and 3, or with a conventional excitation laser source. Scale bar: $20\mathrm{\mu }\mathrm{m}$ .

4.1.

Manufacturing of Tapered Fibers

The tapered fibers were pulled using a home-built drawing rig. The SMF 28 fiber (Thorlabs, Newton, NJ) was heated over a propane-butane-oxygen flame and drawn out to generate a homogeneously thin waist with a diameter of about $2.1\mathrm{\mu }\mathrm{m}$ and a length of approximately $90\mathrm{mm}$ . The tapered regions where the fiber diameter changes from the original diameter of $125\mathrm{\mu }\mathrm{m}$ to the above mentioned thin waist diameter of about $2.1\mathrm{\mu }\mathrm{m}$ were approximately $15\mathrm{mm}$ long on either side. Thickness control was achieved by varying the drawing velocity. The fiber was profiled using a 532-nm frequency-doubled Nd:YAG laser (Adlas, Lübeck, Germany) by recording the diffracted image on a diffuse screen using a standard CCD camera (DSC 717, Sony, Cologne, Germany). The diffracted image was evaluated using a specially developed algorithm.³⁰ The homogeneity over the waist length was better than 10%.

4.2.

Coupling of Femtosecond Pulsed Light into Tapered Fiber

The pulsed light was directed through a Faraday isolator (FR 820 BB, Linos Photonics, Göttingen, Germany) using two silver mirrors (New Focus 5103, New Focus, San Jose, CA). The laser beam was then focused into the fixed end of the tapered fiber with a $10\times 0.3\mathrm{NA}$ objective (Linos Photonics). A three-axis kinematic mount (MDT-612, Thorlabs) for the objective is necessary to optimize coupling efficiency. The light emerging from the fiber was collected and collimated by a $4\times 0.1\mathrm{NA}$ objective (Linos Photonics).

4.3.

Coupling of the Visible White Light of the Supercontinuum into the Microscope

In order to couple the visible supercontinuum white light into the Leica TCS SP2 AOBS (Leica Microsystems, Bensheim, Germany), we removed the KineFlex fiber coupler (Point Source, Southampton, England) from the microscope controller table. This fiber coupler was mounted onto the optical table and used to couple the light into the fiber, which delivered the light directly into the microscope. Since there was no direct access to the output end of this fiber to detect how successful we were in coupling the light, we applied the following procedure for fast coupling. Using the microscope control, all six selected laser lines were deflected into the optical beam path with maximum intensity. Instead of a microscope objective, a reflective object was set onto the objective revolver in the optical path. Next, the AOBS controller was set to detect maximum reflection on all the transmitted laser lines. To obtain feedback of the coupling efficiency, one of the built-in photomultiplier tubes (PMT) was used to detect the reflection over the entire accessible spectrum. This method exploited the sensitivity of the PMT to detect any light that was reflected back from the reflective object, allowing for a fast optimization of the coupling. The fiber coupler was adjusted until the coupling efficiency was sufficient to replace the reflective object in the beam path with a sensitive power meter (Fieldmaster, Coherent), and the fiber adjustment was then continued to maximize power at the objective revolver. After achieving a sufficient coupling efficiency, no further adjustment in the optical path of the microscope was necessary.

4.4.

Coupling of the NIR Part of the Supercontinuum into the Microscope

To couple the supercontinuum NIR white light into the optical beam path of the microscope, the visible component of the spectrum was filtered with a dichroic mirror (Chroma), and the remainder was directed into the NIR port of the microscope with a pair of broadband silver mirrors (New Focus 5103, New Focus). The standard procedures for the adjustment of the NIR port were then performed, resulting in a power of 20 mW before the objective.

4.5.

Histochemical Staining

The histochemical staining of cytoskeletal components was performed with NG108-15, a neuron-like mouse neuroblastoma/rat glioma hybrid cell line (ATCC, Manassas, VA). These cells were cultured in medium (90% Dulbecco’s Modified Eagle Medium, 10% fetal bovine serum, $10\mathrm{mM}$ Hepes) and plated on glass cover slips $24\mathrm{h}$ before fixation with 0.3% glutaraldehyde in Brinkley Buffer (BRB80) for $10\min$ . Next, cell membranes were permeabilized with 0.1% Triton X-100 in phosphate buffered saline (PBS) for $10\min$ and unspecific binding sites were blocked with 1% normal goat serum in PBS for $30\min$ . Microtubules were revealed by incubation with a monoclonal mouse antibody (E7, 1:10 in the blocking solution; provided by the Hybridoma Bank, University of Iowa, Iowa City, IA) for $3\mathrm{h}$ and Alexa 488-goat anti-mouse IgG (Molecular Probes, Eugene, OR; $10\mathrm{\mu }\mathrm{g}∕\mathrm{ml}$ in PBS) overnight. Subsequently, cells were washed with 0.05% Tween 20 in PBS to further reduce unspecific binding of the applied immunoreagents. Filamentous actin was then stained with TRITC-coupled phalloidin( $2.5\mathrm{\mu }\mathrm{g}∕\mathrm{ml}$ in PBS) for $30\min$ . Cells were washed and kept in PBS during the microscopic analysis.

For the histochemical staining of rat brain sections the tissue preparation started with the transcardial perfusion of young adult Wistar rats according to the guidelines of the Laboratory Animal Care and Use Committee of the Leipzig University. The animals were perfused with 4% paraformaldehyde and 0.1% glutaraldehyde in $0.1\mathrm{M}$ phosphate buffer, pH 7.4 (PB). Brains were then postfixed in 4% paraformaldehyde in PB overnight and cryoprotected by equilibration with 30% sucrose in PB. $30\text{-}\mathrm{\mu }\mathrm{m}$ -thick coronal forebrain sections were cut on a freezing microtome and collected in $0.1\mathrm{M}$ Tris-buffered saline (TBS, pH 7.4). To stain for vessels and the astroglial marker, glial fibrillary acidic protein (GFAP) sections were washed in TBS and preincubated with 5% normal donkey serum (Dianova, Hamburg, Germany) and 0.3% Triton X-100 in TBS (NDS-TBS-T) for $1\mathrm{h}$ . Sections were then incubated in a cocktail of rabbit-anti-GFAP (1:1000, Dako, Hamburg, Germany) and biotinylated Solanum tuberosum agglutinin ( $20\mathrm{\mu }\mathrm{g}∕\mathrm{ml}$ , Vector, Burlingame, CA) in NDS-TBS-T overnight. Subsequently, sections were extensively rinsed in TBS and incubated with a mixture of carbocyanine $\left(\mathrm{Cy}\right)3$ -conjugated donkey anti-rabbit IgG and $\mathrm{Cy}2$ -tagged streptavidin (both from Dianova at $20\mathrm{\mu }\mathrm{g}∕\mathrm{ml}$ TBS containing 2% bovine serum albumin) for $1\mathrm{h}$ . Finally, sections were rinsed in TBS and in distilled water, mounted onto a slide, dried, and sealed with a coverslip in Entellan (Merck, Darmstadt, Germany). All chemicals were purchased from Sigma (Taufkirchen, Germany) unless stated otherwise.