1238
OPTICS LETTERS / Vol. 23, No. 15 / August 1, 1998
Two-photon near- and far-field fluorescence microscopy with
continuous-wave excitation
Stefan W. Hell, Martin Booth, and Stefan Wilms
High Resolution Optical Microscopy Group, Max Planck Institute for Biophysical Chemistry, D-37070 Göttingen, Germany
Christoph M. Schnetter, Achim K. Kirsch, Donna J. Arndt-Jovin, and Thomas M. Jovin
Department of Molecular Biology, Max Planck Institute for Biophysical Chemistry, D-37070 Göttingen, Germany
Received April 8, 1998
We report on scanning far- and near-field two-photon microscopy of cell nuclei stained with DAPI
and bisbenzimidazole Hoechst 33342 (BBI-342) with the 647-nm laser line of a cw ArKr mixed-gas
laser. Two-photon-excited f luorescence images are obtained for 50 – 200 mW of average power at the
sample. A nearly quadratic dependence of f luorescence intensity on laser power confirmed the two-photon
effect. The nonlinearity was further supported by evidence of three-dimensional sectioning in a scanning farfield microscope. We find that the cw two-photon irradiation sufficient for imaging within typically 5 s does
not significantly impair cell cycling of BBI-342-labeled live cells. Finally, high-resolution imaging in scanning
near-field microscopy with good contrast is demonstrated.  1998 Optical Society of America
OCIS codes: 170.0180, 140.3460, 170.2520, 170.5280.
Excitation by cooperative absorption of two longwavelength photons provides unique advantages for
f luorescence microscopy, such as inherent optical sectioning without the need for a pinhole as in scanning
confocal microscopes, improved signals from efficient
large-area detectors that are insensitive to scattering
of the emitted light, greater specimen penetration, suppression of photobleaching and background originating
outside the focal plane, and excitation of near-UVabsorbing dyes with less-cytotoxic visible or infrared
light.1 The last advantage virtually eliminates the
requirement for UV lasers and optics for f luorescence
imaging in the UV. Unfortunately, the prevailing
impression that two-photon-excitation microscopy
(TPM) is feasible only in conjunction with expensive
pulsed lasers such as the mode-locked Ti:sapphire
laser has hindered its routine implementation.
This impression stems from the following relation1
describing the molecular two-photon-absorption rate at
a given average power Pave :
ffl ø 0.56s1ytf ds2Pave 2 ys h̄vAd2 ,
(1)
where t and f denote the pulse width in the focus and the pulse repetition rate, respectively, s2
represents the two-photon cross section, h̄v is the
photon energy, and A is the focal area. The factor of 0.56 holds for hyperbolic-secant-squared shaped
pulses. Relation (1) indicates that pulsed excitation
boosts the signal by a factor 0.56s1ytf d ­ 35, 000 for
a pulse train of frequency f ­ 80 MHz and a focal
pulse width t ­ 200 fs as given with Ti:sapphire laser
illumination. In a high-N.A. (1.2–1.4; water- or oilimmersion) microscope, an average power of Pave ­ 10
mW is associated with a transient peak intensity of
,250 GWycm2 owing to strong focusing. Therefore,
in applications requiring high spatial resolution, saturation, bleaching, or other nonlinear optical effects of0146-9592/98/151238-03$15.00/0
ten restrict the peak intensity 2 – 4 and therefore the
average power Pave to 3 –10 mW in practice. In the
case of cw illumination, however, the sample should
be compatible with a higher average power. Following relation (1), the same two-photon molecular
absorption rate and the same
p image brightness
p are anticipated for Pcw ­ P 3 0.56ytf ­ P 3 35, 000 ­
P 3 187 ø 50021800 mW of cw excitation.
In fact, it was shown that for the UV-absorbing nuclear stain DAPI (excitation peak at 360 nm) ,1 mW of
femtosecond pulsed light (800-nm wavelength) suff ices
for good two-photon imaging.5 The equivalent cw
excitation power of Pcw ø 187 mW is available, for example, from a water-cooled ArKr mixed-gas laser operating at the 647-nm emission line. The value of s2 at
647 nm has not been determined, but the single-photon
excitation spectra of DAPI and Hoechst 33342 (BBI342) indicate that s2 should be higher at 647 than at
800 nm. This view is supported by the finding that
two-photon excitation spectra of organic molecules are
usually slightly blueshifted.6 The 187-fold reduction
of peak power leads to a shift in the potential limitations of this alternative illumination strategy (cw TPM)
from nonlinear to linear phenomena, such as heating
and electromagnetic trapping of the specimen. Initial
studies2,7 with cw TPM were carried out by scanning of
the sample with a piezoelectric stage relative to a f ixed
beam, which is a slow process that is limited to a power
of a few milliwatts owing to optical trapping.2,8
Here we report fast-mode cw TPM of fixed and live
cell nuclei labeled with DAPI and BBI-342, achieved
by coupling of the 647-nm ArKr laser line into a
laser scanning miscroscope (Leica TCS NT) adapted
for 400– 550-nm detection. We used 1003, 1.4-N.A.
oil-immersion and 633, 1.2-N.A. water-immersion objectives for specimens mounted in glycerol and water,
respectively. We typically obtained Pcw ­ 802200 mW
at the sample. Discrimination for the f luorescence
 1998 Optical Society of America
August 1, 1998 / Vol. 23, No. 15 / OPTICS LETTERS
signal was achieved by a combination of suitable
dichroic mirrors and emission f ilters. The f luorescence light was collected by a large-area bialkali
photomultiplier operating in the current mode as
a detector.9 Large-area detection excluded any confocal discimination by a hidden spatial filter in the optical train.
DAPI-stained Drosophila melanogaster salivary
gland chromosomes imaged with 100-mW illumination
are displayed in Fig. 1. Living fibroblasts grown on
coverslips, labeled10 with BBI-342, and imaged with
,160-mW power are displayed in Fig. 2. Detection
of distinct, optically sectioned images and the nearly
quadratic s1.8 6 0.1d dependence of f luorescence on
laser power conf irmed the two-photon effect. Contrast was remarkably high even with the short 1-s
recording time that was used.
Potential damage to living cells induced by cw laser
exposure was investigated in Ref. 4. Detailed calculations of the heating of water showed that heating by
linear or nonlinear absorption of water was insignificant11 s,,1 Kd. Optical trapping was observed for
dust particles in the buffer close to the coverslip but not
for biological specimens mounted in water or glycerol.
Exposure to cw two-photon excitation resulted in morphological changes such as rounding up and shrinkage,
effects that were more pronounced for the stained cells.
Similar phenomena also occur in conventional singlephoton UV microscopy.
Approximately 97% of the BBI-342-stained and 90%
of the unstained cells remained adherent after light
exposure. The capacity of scanned cells to synthesize
DNA and divide was assessed. Most s.90%d of both
the stained and the unstained cells incorporated thymidine analog 5-bromo-2′-deoxyuridine (Ref. 12) and
multiplied ,1.7-fold within 17 h. We conclude that
irradiation under the experimental conditions did not
significantly impair cell viability and cycling.
As a next step we combined cw two-photon excitation
with scanning near-f ield optical microscopy (SNOM)
operated in the shared aperture mode,13 in which an
uncoated optical fiber tip drawn to an ,50-nm radius
serves as an eff icient probe for both excitation of the
sample and collection of the f luorescence. The strong
autof luorescence of f ibers excited directly by shortwavelength light generally precludes application of
SNOM to UV- and blue-excitable dyes. With cw twophoton excitation, autof luorescence no longer overlaps
the spectral detection window. In addition, nonlinear
processes in the fiber are weak. Both effects extend
the spectral range of accessible dyes to the near UV
and increase the image contrast. The greater axial
and radial conf inement of the functional excitation
can lead to a significant improvement in resolution
compared with that of single-photon excitation SNOM
with the same laser line. A SNOM image of BBI-342stained polytene chromosomes excited with 55 mW of
power from a 647-nm ArKr laser (4-ms pixel dwell
time) displaying high resolution and contrast is shown
in Fig. 3.
Figure 4 shows the measured f luorescence as a
function of excitation power in far- and near-f ield
microscopes. The SNOM measurements were carried
1239
out on a f ixed spot of the chromosome. The quadratic
s1.9 6 0.1d dependence that was found confirmed
the two-photon nature of the excitation. The powerdependence measurements in the far-f ield microscope
are based on the mean of eight averaged images
(frames) of DAPI-stained calf thymus DNA.
Cw TPM at 100 –200 mW corresponds to ,1 mW
of pulsed excitation and thus generates images ,10
times fainter than those generated in the pulsed mode
with 3 –4 mW of power. In low-resolution applications
involving strong scattering, such as the deep imaging
of brain slices, fast far-f ield cw TPM will almost
certainly not be possible.14 Similarly, for dyes with
significantly lower s2 , immunof luorescent labels, and
Ca11 indicators present at lower concentration in
the cell, cw TPM will be less efficient. Multifocal
multiphoton microscopy15 operating at video rate will
also require a pulsed light source.
Fig. 1. Cw two-photon far-f ield microscopy of a DAPIstained polytene chromosome recorded with cw excitation
at 647 nm by an ArKr laser (total exposure, ,4 s).
Fig. 2. Living BBI-342-labeled (15-mM BBI-342 in phosphate-buffered saline) 3T3 Balb/c fibroblast nuclei imaged
in a closed perfusion chamber and averaged over two
consecutive 1-s frames (total exposure, ,0.5 s).
1240
OPTICS LETTERS / Vol. 23, No. 15 / August 1, 1998
equipment. In the future it will be interesting to increase the power further, for example, by a factor of 2,
which could potentially lead to an approximately fourfold increase in f luorescence signal or imaging speed.
Cw two-photon SNOM adds a new dimension to
near-field optics by extending the applicability of
f luorescence SNOM to the imaging of UV dyes and
confining the near-f ield interaction area. As opposed
to single-photon UV confocal microscopy of DAPI- or
BBI-342-stained samples, cw TPM requires only conventional optics and dispenses with the UV laser. In
far-f ield f luorescence microscopy, high-resolution
three-dimensional imaging of the important UVabsorbing nuclear stains DAPI and BBI-342 can be
achieved without recourse to pulsed or UV lasers.
References
Fig. 3. SNOM f luorescence image of a BBI-342-stained
and dried polytene chromosome.
Fig. 4. Double-logarithmic plots of f luorescence as a function of laser power for a BBI-342-stained polytene chromosome (observed with SNOM) and calf thymus DNA 1 DAPI
(far-f ield laser scanning microscope).
However, our experiments indicate that, in highaperture, high-resolution imaging of DAPI and BBI342, cw TPM is an effective and useful nonlinear
imaging option. Cw TPM is superior to pulsed TPM or
UV excitation with regard to system simplicity, maintenance, and cost. Furthermore, cw TPM is backed
by a growing palette of available cw semiconductor
lasers.2,16 The power of Pcw ­ 200 mW was the highest power that we could technically achieve with our
1. W. Denk, J. H. Strickler, and W. W. Webb, Science 248,
73 (1990).
2. P. E. Hänninen, M. Schrader, E. Soini, and S. W. Hell,
Bioimaging 3, 70 (1995).
3. K. König, L. Hong, M. W. Berns, and B. J. Tromberg,
Nature (London) 377, 20 (1995).
4. K. König, H. Liang, S. Kimel, L. O. Svaasand, B. J.
Tromberg, T. Krasieva, M. W. Berns, K. J. Halbhuber,
P. T. C. So, W. W. Mantulin, and E. Gratton, Proc. SPIE
2983, 37 (1997).
5. R. Wolleschensky, T. Feurer, R. Sauerbrey, and U.
Simon, ‘‘Characterization and optimization of a laserscanning microscope in the femtosecond regime,’’ Appl.
Phys. B (to be published).
6. C. Xu and W. W. Webb, J. Opt. Soc. Am. B 13, 481
(1996).
7. P. E. Hänninen, E. Soini, and S. W. Hell, J. Microsc.
176, 222 (1994).
8. Y. Liu, G. J. Sonek, M. W. Berns, K. König, and B. J.
Tromberg, Opt. Lett. 20, 2246 (1995).
9. M. Booth and S. W. Hell, J. Microsc. 190, 298 (1998).
10. D. J. Arndt-Jovin and T. M. Jovin, J. Histochem.
Cytochem. 25, 585 (1977).
11. A. Schönle and S. W. Hell, Opt. Lett. 23, 325 (1998).
12. M. H. Fox, D. J. Arndt-Jovin, T. M. Jovin, P. H.
Baumann, and M. Robert-Nicoud, J. Cell Sci. 99, 247
(1991).
13. A. K. Kirsch, C. K. Meyer, and T. M. Jovin, in Integration of Optical Techniques in Scanning Microscopy, E.
Kohen and J. G. Hirschberg, eds. (Plenum, New York,
1996), p. 317.
14. W. Denk and K. Svoboda, Neuron 18, 351 (1997).
15. J. Bewersdorf, R. Pick, and S. W. Hell, Opt. Lett. 23,
655 (1998).
16. Z. Zhang, G. J. Sonek, X. Wei, M. W. Berns, and B. J.
Tromberg, Jpn. J. Appl. Phys. 36, 1598 (1997).