Microbeam-integrated multiphoton imaging system
Alan W. Bigelow, Charles R. Geard, Gerhard Randers-Pehrson, and David J. Brenner
Center for Radiological Research, Columbia University, New York, New York 10032, USA
?Received 22 September 2008; accepted 18 November 2008; published online 12 December 2008?
Multiphoton microscopy has been added to the array of imaging techniques at the endstation for the
Microbeam II cell irradiator at Columbia University’s Radiological Research Accelerator Facility
?RARAF?. This three-dimensional ?3D?, laser-scanning microscope functions through multiphoton
excitation, providing an enhanced imaging routine during radiation experiments with tissuelike
samples, such as small living animals and organisms. Studies at RARAF focus on radiation effects;
hence, this multiphoton microscope was designed to observe postirradiation cellular dynamics. This
multiphoton microscope was custom designed into an existing Nikon Eclipse E600-FN research
fluorescence microscope on the irradiation platform. Design details and biology applications using
this enhanced 3D-imaging technique at RARAF are reviewed. © 2008 American Institute of
Physics. ?DOI: 10.1063/1.3043439?
Multiphoton excitation can occur within a high photon
density, where, through energy superposition, a combination
of photons can act like one to induce an electronic transition
in a sample. In the instance of two photon excitation, two
photons that are spatially and temporally coincident can act
as one photon with twice the energy ?half the wavelength? to
excite a fluorescent molecule to emit fluorescent signal
which has a quadratic dependence on the incident light
intensity.1First theorized by Göppert-Mayer,2multiphoton
excitation is now achievable using focused photon densities
from modern, high-powered, pulsed lasers. A multiphoton
microscope is a laser-based, three-dimensional ?3D?, mini-
mally damaging imaging tool with optical-sectioning capa-
bilities. When compared to conventional confocal micros-
copy, multiphoton microscopy allows for greater penetration
depth with reduced phototoxicity and photobleaching in the
At the Radiological Research Accelerator Facility
?RARAF?,3Columbia University, our facility users who
study the ionizing radiation effects on mammalian cells are
increasingly utilizing tissue samples and small organisms.4
To image their 3D samples, our users requested the multi-
photon microscope that we developed and which is now fully
integrated into the Microbeam II endstation5at RARAF, see
Fig. 1. A custom design for the multiphoton microscope was
necessary, given the geometrical constraints of the pre-
existing microscope fitted at the terminus of the vertical ion
beamline. Intended for detecting and observing short-term
molecular kinetics of radiation response in living tissue and
in cell-culture samples, the multiphoton microscope at
RARAF is the first of its kind to be assembled and imple-
mented onto a microbeam cell-irradiation platform.
The multiphoton microscope at RARAF was custom de-
signed around a Nikon Eclipse E600-FN research fluores-
cence microscope equipped with a high-precision stage at the
microbeam cell-imaging platform. This Nikon microscope
was previously modified to function over a vertical ion beam
by removing the base and mounting the top of the micro-
scope to the end of a pivot arm, allowing the microscope to
be placed either in an on-line position over the vertical ion
beam or in an off-line position.
Along the incident laser beam path depicted in Fig. 2, a
half-wave plate and polarized beam splitter act as an attenu-
ator and are placed in the beam path to control the laser beam
power. After passing a fast laser shutter, mirrors direct the
laser beam up through the microscope pivot shaft, allowing
the multiphoton microscope to operate in either on-line or
off-line modes. Along the pivot-shaft axis, a Galilean beam
expander increases the beam to a size that will fill the back
aperture of the objective lens. The expanded beam is directed
through a scan head and a scan lens, which are fixed to a
platform above the microscope mount. The scan lens focuses
the scanned laser beam to a conjugate image plane ?a charge-
coupled device ?CCD? camera is placed at a corresponding
image plane for use with standard fluorescent microscopy?.
The incident light path continues through the side of a
custom-built trinocular tube where the position of a retract-
able mirror selects between fluorescence and multiphoton
microscopy uses. This mirror guides the laser vertically
down through both the microscope tube lens and an objective
lens to a focal point within a specimen, where multiphoton
absorption preferentially occurs. The scanned laser’s focal
plane establishes an optical section within the specimen. Re-
turning along the collection pathway, light emitted from the
specimen is selectively deflected by a series of dichroic mir-
rors to an array of photomultiplier tubes ?PMTs?. Images are
constructed through correlating PMT signals with the scan
head position. The optics path was designed to optimize laser
delivery to the sample. The optical parameters are discussed
below and were specified through optical considerations
from the laser output to the throughput of the microscope
REVIEW OF SCIENTIFIC INSTRUMENTS 79, 123707 ?2008?
0034-6748/2008/79?12?/123707/6/$23.00 © 2008 American Institute of Physics
The excitation light source for the multiphoton micro-
scope at RARAF is a Chameleon ?Coherent Inc., Santa
Clara, CA? tunable titanium:sapphire laser that provides 140
fs pulses at a 90 MHz repetition rate with a wavelength tun-
ing range of 705–950 nm. Specifications for this laser men-
tion the 1/e2beam diameter ?1.2?0.2 mm? at the exit port
and at the peak of the tuning curve but omit mentioning the
beam divergence, an important quantity for designing an op-
tics system. In order to have a better grasp of the laser’s
divergence, beam spot size measurements were made at
about 25 nm increments of wavelength along the 705–950
nm tuning curve. These measurements were made by mea-
suring the 1/e2beam diameter at 12.5 cm from the laser and
at 3 m from the laser. In both cases the 1/e2beam diameter
was measured using a power meter and an adjustable iris.
The power-tuning curve ?Fig. 3? looks similar to the diver-
gence curve ?Fig. 4?, suggesting that the divergence variation
is related to laser power.
Since the Chameleon Ti:sapphire laser operates exclu-
sively at full power, an external attenuator was installed to
regulate the laser power. The attenuator section, consisting of
an achromatic half-wave plate ?Thor Laboratories, Newton,
NJ? with a broadband polarizing beamsplitter cube and a
beam dump ?CVI Laser, Albuquerque, NM?, is located im-
mediately after the laser exit window. The half-wave plate is
mounted in a rotating holder, so the laser power is adjusted
by rotating the half-wave plate. Downstream from the attenu-
ator, an automated fast laser shutter ?nmLaser, San Jose, CA?
blanks the beam between image acquisitions. Since the op-
tics table has a solid top, an added half-inch thick, 24
?12 in.2bench plate ?Edmund Industrial Optics, Bar-
rington, NJ? provides attachment holes for the attenuator op-
tics and primary broadband Ti:sapphire laser turning mirrors.
The microscope pivot-post is also mounted on top of this
bench plate, with bolts passing through holes in the plate to
anchor the microscope pivot-post to the optics top.
C. Beam expander
The geometry of the infinity optics in the microscope
provides guidance to establishing the diameter of the inci-
dent laser beam and its scanning angle range, which will be
addressed later in Sec. II D. Following the specifications for
a particular objective ?Nikon CFI LU Epi Plan Fluor 50?,
0.80 NA, 1.0 mm WD ESD?, the diameter of the back aper-
ture of the lens is found to be 6.4 mm. Additionally, a
diffraction-limited spot is achievable for a microscope objec-
tive that has its back aperture filled with a planar wave; with
FIG. 1. ?Color online? Multiphoton microscope at RARAF for observing
responses in cell cultures and in tissues after particle irradiation.
FIG. 2. ?Color online? Diagram of the multiphoton microscope and optics
path. See the text for more details.
FIG. 3. Power measurements for the Chameleon ?Coherent, Inc.? Ti:sap-
FIG. 4. Divergence measurements for the Chameleon ?Coherent, Inc.?
Ti:sapphire laser. The divergence measurements were made by comparing
1/e2beam diameters at 12.5 cm and at 3 m from the laser exit window. At
12.5 cm from the exit port, 1/e2beam diameters measured from 1.1 to 1.4
mm and increased linearly as a function of wavelength. At 3 m from the exit
port, the 1/e2beam diameter was as large as 5.6 mm at the peak of the
123707-2Bigelow et al.Rev. Sci. Instrum. 79, 123707 ?2008?
Gaussian beams, to approximate this condition, it is common
that the 1/e2beam diameter overfills the back aperture by a
factor of 2.6With this scenario, the 1/e2laser beam diameter
at the objective back aperture should be 12.8 mm.
Considering the variation in initial laser beam spot size
and divergence, there is a corresponding distribution in the
laser beam spot size at the objective back aperture. Optics
transport calculations show that the minimum beam spot size
at the objective back aperture is at a wavelength of 705 nm,
and the expansion requirement for that 1/e2beam diameter
is from 1.5 to 12.8 mm, or 8.4?. Exceeding the plane wave
approximation mentioned above, a commercial 10? Galilean
beam expander ?Thor Laboratories, Newton, NJ? is used in
conjunction with an adjustable iris to enlarge the beam to
overfill the objective back aperture. A beam expander de-
signed for the smallest beam spot size will function for the
rest of the laser output spectrum. Wavelengths with greater
divergence ?see Fig. 4? are truncated, making the divergence
distribution in the continuing beam more uniform.
D. Scanning head
The scanning head design for this multiphoton micro-
scope incorporates two scanning mirrors and a scan lens,
whose focal length ?200 mm? matches the focal length of the
microscope tube lens and establishes a one-to-one optical
relationship. These components provide an array of laser fo-
cal points at a point along the optics path that is conjugate
with the position of the CCD chip used for standard fluores-
cent microscopy. To optimize laser scanning, the scan lens is
positioned with one focal point coinciding with the image
plane of the microscope and the other focal point ?or conju-
gate telecentric plane? matching the geometric midpoint be-
tween the scan mirror surfaces.6
With the goal of generating images that are similar in
size as the CCD chip, the scanning angles can be found by
taking the trigonometric relationships between the CCD chip
dimensions to the focal length of the scan lens. The CCD
dimensions can be found from the CCD specification 756
?581 pixels. With a 40? objective lens and a microscope
graticule, images were experimentally determined to view an
areaof 332.6?255.64 ?m2
0.44 pixels/?m. Accounting for the 40? objective, the
CCD dimensions are 13.3?10.23 mm2. From this it is
found that the X-scanning angle=?arctan?13.33/2/200?
=?1.9°, and the Y-scanning angle=?arctan?10.23/2/200?
Scanning mirrors were selected based on the back aper-
ture diameter of the objective lens ?6.4 mm? and the desired
scan rate’s fast trace ??500 Hz?. Galvanometer speed limits
correlate with the mirror paddle size, and for the desired scan
rate, industry specifications indicated that the mirror aperture
be less than 8 mm in order to operate smoothly without
overheating. A model 6215H series galvanometer scanner
?Cambridge Technology, Inc., Cambridge, MA? with 6 mm
mirror aperture, a step down from the 8 mm mirror aperture,
was chosen as an ideal compromise between size and speed
requirements. Even though the 6 mm mirror aperture size
specification is less than the 6.4 mm diameter requirement,
mechanical drawings for this mirror set allow for a maxi-
mum 7.0 mm aperture over the above-mentioned scanning
angles. This model is rated for speeds at 1000 Hz over ?2°,
with the understanding that occasional clearing moves
should be incorporated to run the galvanometers through
their range of motion to maintain proper lubrication. This
choice of scan lens and the associated one-to-one optical
relationship with the tube lens has an effect, in that, to main-
tain a diffraction-limited laser focal spot, the objective back
aperture should not exceed the scanning mirror aperture.
E. Image acquisition
Image acquisition in laser-scanning microscopy relies on
correlating the laser scan position with the photon yield from
the sample. Scan positions translate to pixel position and
detected light converts to pixel brightness. Light emitted
from the specimen enters the microscope objective vertically
and gets reflected laterally by a dichroic element mounted in
an especially designed filter cube. This deflected signal then
enters a custom-built, light-tight PMT housing shown in Fig.
5, where an additional dichroic mirror assigns signal toward
one of the two H5783 PMTs, each followed by a C7319
preamplifier unit ?Hamamatsu, Japan?. An emission filter
placed before each PMT designates the wavelength range of
detected signal. For samples with multiple fluorophores,
color images can be produced using appropriate dichroic
mirrors and emission filters. Two potentiometers mounted to
the side of the PMT housing provide independent signal gain
control. With these two simultaneous imaging channels, ex-
periments using Förster resonance energy transfer ?FRET? to
monitor postcell-irradiation dynamics are possible. FRET op-
tics sets are on hand for the following transitions: BFP
?GFP, CFP?YFP, and Alexa 488?Alexa 594.
ScanImage, a multiphoton microscope control software
package developed by Karel Svoboda’s group at Cold Spring
Harbor,7is installed on the RARAF multiphoton microscope.
This software automates the correlation between the laser
scan position and the signal at the PMT to produce the im-
ages. Written in MATLAB®, ScanImage communicates with
the multiphoton microscope hardware via an NI-6110 multi-
function DAQ device ?National Instruments, Austin, TX?.
FIG. 5. ?Color online? Light-tight attachment for fluorescent-signal detec-
tion using a dichroic mirror and PMTs.
123707-3Multiphoton imaging system Rev. Sci. Instrum. 79, 123707 ?2008?
This interface supplies the scanning-voltage waveforms and
receives signal pulses from the PMT/preamplifier units. Sc-
anImage has an array of imaging modes, including averag-
ing, looping, and acquiring z-stacks for 3D imaging. At
RARAF, an LP-200 piezoactuated nanopositioner ?Mad City
Laboratories Inc., Madison, WI? works in conjunction with
ScanImage to move vertically during z-stack image acquisi-
tion. After acquisition, images are typically postprocessed to
add false coloring and to compile movies. Time-lapse movies
can document dynamic change and movies of a z-sequence
offer the observer a unique, informative excursion through a
F. Point spread function
The point spread function ?PSF? for a multiphoton mi-
croscope defines the volume of multiphoton excitation. This
volume is an ellipsoid that represents the resolution of the
microscope. The PSF can be measured by imaging a subreso-
lution, pointlike fluorescent object.8For measuring the PSF
in our system, z-stacks were taken of individual 0.2 ?m di-
ameter blue fluorescent microspheres ?B200, Microgenics
Corporation? using a Nikon 60X NA1 water-dipping objec-
tive. Considering the maximum of the fluorescent micro-
sphere absorption curve ??388 nm?, the incident wave-
length required for two-photon excitation is double that
value, or 780 nm. Axial and radial PSF data were obtained
by extracting maximal pixel values and fitting them to the
peak fluorescence intensity, zois the centroid, ?zis the inten-
sity distribution width, and C is the background. The associ-
ated Gaussian fits are shown in Figs. 6 and 7. For these PSF
measurements, the full widths at half maximum for the axial
and radial Gaussian fits are 2.8 and 0.65 ?m, respectively.
2?+C, where Iozis the
The image of bovine pulmonary artery endothelial cells
in Fig. 8 is an example of a color image produced by
wavelength-specific detection with the RARAF multiphoton
microscope. The cells in this image were stained with BO-
DIPY FL phallacidin to label the filamentous actin and were
counterstained with DAPI to label the nuclei ?shown in red?.
The objective lens used for imaging these cells was a Nikon
CFI LU Epi Plan Fluor 50? objective, with 0.80 numerical
aperture and 1.0 mm working distance. To develop this im-
age, 20–30 single frames were acquired per color at a rate of
2 s/frame. A routine written in MATLAB® subsequently
summed and combined the images into a matrix for color
To validate our multiphoton microscope design inte-
grated into our microbeam endstation, a microbeam irradia-
tion experiment with a multiphoton observation was re-
quired. To meet this goal, the microbeam endstation was
used to examine particle-induced single strand breaks in
DNA within single cells. Aroumougame Asaithamby from
David Chen’s group ?University of Texas Southwestern, Dal-
las, TX? used Human HT1080 fibrosarcoma cells that had
been transfected with a GFP-tagged XRCC1 DNA single-
strand break repair protein. Figure 9 shows multiphoton im-
ages of one cell ?a? before irradiation and ?b? 4 min after
exposure to 400 3Me V alpha particles at a predetermined
position within a cell nucleus. 4 min after irradiation, the
region of enhanced GFP signal corresponds to the XRCC1
focus and the irradiation position. In this example, ion-beam
targeting was deliberate to avoid areas deficient in GFP
FIG. 6. ?Color online? Axial and PSF measurements. Gaussian fits to bright-
est pixel values from z-stack images of a 0.2 ?m blue fluorescent bead
taken using a 60? NA1 water-dipping objective.
FIG. 7. ?Color online? Radial PSF measurements. Gaussian fits to brightest
pixel values from z-stack images of a 0.2 ?m blue fluorescent bead taken
using a 60? NA1 water-dipping objective.
FIG. 8. ?Color online? Multiphoton microscope image of bovine pulmonary
artery endothelial cells. Cell nuclei diameter ?10 ?m.
123707-4 Bigelow et al. Rev. Sci. Instrum. 79, 123707 ?2008?
?likely nucleoli?. This clear response demonstrates that the
multiphoton microscope was successful at recording high-
resolution, time-lapse images of particle-induced focus for-
mation within a single cell in real time. As an additional
capability, projections of 3D z-stacks can quantify focus
track volume to correlate with the amount of radiation ?num-
ber of particles? delivered to the cell nucleus.
As radiation experiments move more toward tissue
samples and small organisms, the ability to image optical
sections within a 3D object was a driving factor in develop-
ing a multiphoton microscope into the endstation of the
RARAF microbeam. Our user, Peter Grabham, brought hu-
man umbilical vein endothelial cells ?HUVECs? to RARAF
for optical sectioning. Derived from walls of the umbilical
vein, HUVEC cells were seeded onto a collagen matrix, a
disk approximately 200 ?m thick by 4 mm in diameter.
While these cells “tunneled” through the matrix, forming a
3D structure similar to the capillaries of the blood circulatory
system, collagen was displaced by cell growth and was de-
graded by cell enzyme action. Series of optical sections
through these tissue samples ?z-stacks? were compiled as
movies to effectively transport the observer through the tis-
sue slices. Figures 10 and 11 show views from these z-stack
movies. In addition to using the nuclear stain YOYO-1, au-
tofluorescence ?AF? and second harmonic generation ?SHG?
are two nonstain imaging modes that were used to capture
the cytoplasm and the collagen matrix. These two nonstain
imaging modes require less sample preparation and allow
structural imaging in 3D. With the capacity for optical sec-
tioning, the multiphoton microscope at RARAF offers a
mode for observing radiation-induced effects, such as foci
formation, within tissue.
Researchers at RARAF are now irradiating small, living
organisms, such as the C. elegans nematode. With its ge-
nome fully characterized in the literature, this model organ-
ism offers an ideal model for particle-irradiation effects. The
worms undergo anesthesia prior to microbeam irradiation.
Immobilization of the worms is important for multiphoton
microscopy, where image exposure can range from 0.5 to 2 s.
Figure 12 shows the cross-sectional multiphoton images of
GFP expression in live C. elegans under anesthesia. Follow-
ing the anesthesia period and the multiphoton imaging pro-
cess, the C. elegans specimen regained normal motion and
health. This is a testament to the nondamaging quality of
multiphoton microscopy. By successfully imaging optical
sections within a live, small organism, the microbeam-
integrated multiphoton imaging system at RARAF offers us-
ers an array of imaging modes for 3D samples.
RARAF facility users often apply red fluorescent protein
?RFP? to their biological samples for imaging purposes.
However, the original Ti:sapphire laser system on the multi-
photon microscope cannot produce wavelengths needed for
RFP excitation. Following up on user requests, RARAF re-
cently purchased a laser upgrade, a Chameleon Ultra II ?Co-
herent, Inc., Santa Clara, CA?, which covers a wavelength
FIG. 9. Multiphoton microscopy images of an HT1080 cell nucleus with
GFP-tagged XRCC1: ?left? prior to irradiation and ?right? 4 min after mi-
crobeam exposure of 400 3 MeV alpha particles.
FIG. 10. ?Color online? Multiphoton microscopy images of HUVEC tissue
samples. The cell nuclei are stained with YOYO-1. Cell nuclei imaged by
YOYO-1 ?green? and cell cytoplasm imaged with AF ?blue?.
FIG. 11. ?Color online? Multiphoton microscopy images of HUVEC tissue
samples. The cell nuclei, are stained with YOYO-1. Cell nuclei imaged by
YOYO-1 ?green?, cell cytoplasm imaged with AF ?blue?, and collagen im-
aged by SHG ?red?.
123707-5 Multiphoton imaging systemRev. Sci. Instrum. 79, 123707 ?2008?
range of 680–1080 nm. At the focal point of this new laser
beam, the two photon process effectively provides ?340–540
nm?, which enables RFP excitation.
Because our multiphoton microbeam is the first of its
kind to be located on the endstation of a microbeam cell
irradiator, RARAF is unique with its capacity to offer high-
resolution, 3D imaging of bulk samples for radiobiological
studies. Our multiphoton microscope is fully functional and
has been successfully used for a variety of experiments, in-
cluding those by our outside users.
This work has been sponsored by NASA Grant No.
NNJ05HI37G and a P41 grant supported by the National
Institute of Biomedical Imaging and Bioengineering: NIBIB
5 P41 EB002033-13. We thank Aroumougame Asaithamby
from David Chen’s group, University of Texas Southwestern,
Dallas, TX, for providing HT1080 cells with GFP-tagged
1W. Denk, J. H. Strickler, and W. W. Webb, Science 248, 73 ?1990?.
2M. Göppert-Mayer, Ann. Phys. ?Leipzig? 401, 273 ?1931?.
3A. W. Bigelow, D. J. Brenner, G. Garty, and G. Randers-Pehrson, IEEE
Trans. Plasma Sci. 36, 1424 ?2008?.
4O. V. Belyakov, S. A. Mitchell, D. Parikh, G. Randers-Pehrson, S. A.
Marino, S. A. Amundson, C. R. Geard, and D. J. Brenner, Proc. Natl.
Acad. Sci. U.S.A. 102, 14203 ?2005?.
5A. W. Bigelow, G. J. Ross, G. Randers-Pehrson, and D. J. Brenner, Nucl.
Instrum. Methods Phys. Res. B 231, 202 ?2005?.
6E. K. Stelzer, in Handbook of Biological Confocal Microscopy, edited by
J. B. Pawley ?Plenum, New York, 1995?, pp. 141–147.
7T. A. Pologruto, B. L. Sabatini, and K. Svoboda, Biomed. Eng. Online 2,
8C. Y. Dong, K. Koenig, and P. So, J. Biomed. Opt. 8, 450 ?2003?.
FIG. 12. ?Color online? Multiphoton microscope images of GFP expression
in C. elegans: ?left? pharynx and ?right? tail.
123707-6 Bigelow et al. Rev. Sci. Instrum. 79, 123707 ?2008?