Hawkes P.W., Spence J.C.H. (Eds.) Science of Microscopy. V.1 and 2

Подождите немного. Документ загружается.

768 A. Diaspro et al.

(Born and Wolf, 1980). Using excitation light with wavelength λ, the

intensity distribution at the focal region of an objective with numerical

aperture NA = sin(α) is described [see also Eq. (2)] in the paraxial regime

(Born and Wolf, 1980; Sheppard and Gu, 1990) by

Iuv J v e d

(,) ( )

(/ )

−

∫

ρρρ

(16)

where rho is a dimensionless radial, J

is the zeroth-order Bessel func-

tion, ρ is a radial coordinate in the pupil plane, and u = 8π sin

(α/2)z/λ

and v = 2π sin(α)r/λ are dimensionless axial and radial coordinates,

respectively, normalized to the wavelength (Wilson and Sheppard,

1984). Now, the intensity of ﬂ uorescence distribution within the focal

region has an I(u, v) behavior for the one-photon case and I

(u/2, v/2)

for the TPE case as demonstrated above. The arguments of I

(u/2, v/2)

take into proper account the fact that in the latter case wavelengths are

utilized that are approximatively twice those used for one-photon exci-

tation. As compared with the one-photon case, the TPE intensity dis-

tribution is axially conﬁ ned (Nakamura, 1993; Gu and Sheppard, 1995;

Jonkman and Stelzer, 2002). In fact, considering the integral over v,

keeping u constant, its behavior is constant along z for one-photon and

has a half-bell shape for TPE. This behavior, better discussed in Wilson

(2002), Torok and Sheppard (2002), and Jonkman and Stelzer (2002),

explains the three-dimensional discrimination property in TPE.

Now, the most interesting aspect is that the excitation power falls off

as the square of the distance from the lens focal point, within the

approximation of a conical illumination geometry. In practice this

means that the quadratic relationship between the excitation power

and the ﬂ uorescence intensity results in the fact that TPE falls off as

the fourth power of distance from the focal point of the objective. This

fact implies that those regions away from the focal volume of the objec-

tive lens, directly related to the numerical aperture of the objective

itself, therefore do not suffer photobleaching or phototoxicity effects

and do not contribute to the signal detected when a TPE scheme is

used. Because they are simply not involved in the excitation process, a

confocal-like effect is obtained without the necessity of a confocal

pinhole. It is also immediately evident that in this case an optical sec-

tioning effect is obtained. In fact, the observed image o(x,y,z) at a plane

j, produced by the true ﬂ uorescence distribution i(x,y,z) at plane j, dis-

torted by the microscope through s, plus noise n, again corresponds to

the confocal ideal situation where contributions from adjacent k planes

can be set to zero as in the confocal situation: o

= i

∗ s

+ n.

This means that TPE microscopy is intrinsically three dimensional.

It is worth noting that the optical sectioning effect is obtained in a very

different way with respect to the confocal solution. No ﬂ uorescence has

to be removed from the detection pathway. In this case it should be

possible to collect as much ﬂ uorescence is possible. In fact ﬂ uorescence

can come only and exclusively from the small focal volume traced in

Figure 11–7, which also shows a comparison with the confocal mode,

that is of the order of a fraction of a femtoliter.

In TPE over 80% of the total intensity of ﬂ uorescence comes from a

700- to 1000-nm-thick region about the focal point for objectives with

numerical apertures in the range of 1.2–1.4 (Brakenhoff et al., 1979;

Chapter 11 Two-Photon Excitation Fluorescence Microscopy 769

Wilson and Sheppard, 1984; Wilson, 2002; Jonkman and Stelzer, 2002;

Torok and Sheppard, 2002). This fact implies a reduction in back-

ground that allows compensation of the poorer spatial resolution com-

pared to the single-photon confocal mode due to the longer wavelength

utilized. However, the utilization of an infrared wavelength instead of

UV-visible ones also allows deeper penetration than in the conven-

tional case (So et al., 2001; Periasamy et al., 2002; König and Tirlapur,

2002). The long wavelengths used in TPE, or in general in multiphoton

excitation, will be scattered less than the ultraviolet–visible wave-

lengths used for conventional excitation (de Grauw and Gerritsen,

2001). Hence deeper targets within a thick sample can be reached. Of

course, for ﬂ uorescence light, scattering on the way back can be over-

come by acquiring the emitted ﬂ uorescence using a large area detector

and collecting not only ballistic photons (Soeller and Cannel, 1999;

Bauhler et al., 1999; Girkin and Wokosin, 2002).

7 The Optical Setup

A TPE architecture including confocal modality includes the following:

a high peak-power laser delivering moderate average power (femto second

or picosecond pulsed at a relatively high repetition rate) emitting infrared

or near-infrared wavelengths (650–1100 nm), CW laser sources for confo-

cal modes, a laser beam scanning system or a confocal laser scanning

head, high numerical aperture objectives (>1), a high-throughput

Figure 11–7. Illustration of the two different modalities for selecting 3D infor-

mation under a confocal (left) and TPE regime (right). In the confocal case the

selection is realized during the emission process. The different case of two-

photon excitation shows how the 3D selection can be realized during the ﬂ uo-

rescence excitation process.

770 A. Diaspro et al.

microscope pathway, and a high-sensitivity detection system (Denk et al.,

1995; So et al., 1996; Soeller and Cannell, 1996; Wokosin and White, 1997;

Centonze and White, 1998; Potter et al., 1996; Wolleschensky et al., 1998;

Diaspro et al., 1999a,b; Wier et al., 2000; Soeller and Cannell, 1999; Tan et

al., 1999; Mainen et al., 1999; Majewska et al., 2000; Diaspro, 2002; Girkin

and Wokosin, 2002; Iyer et al., 2002). Figure 11–8 shows a general scheme

for a TPE microscope incorporating a confocal mode.

In typical TPE or confocal microscopes, images are built by raster

scanning the x–y mirrors of a galvanometrically driven mechanical

scanner (Webb, 1996). This fact implies that image formation speed is

mainly determined by the mechanical properties of the scanner, i.e., for

single line scanning it is of the order of milli seconds. Faster beam-

scanning schemes can be realized, even if the “eternal triangle of com-

promise” should be considered for sensitivity, spatial resolution, and

temporal resolution. While the x–y scanners provide lateral focal-point

scanning, axial scanning can be achieved by means of different posi-

tioning devices, the most popular being a belt-driven system using a DC

motor and a single objective piezo nano positioner, such as the PIFOC

(Physik Instrumente, Germany). Usually, it is possible to switc h bet ween

confocal and TPE modes retaining x–y–z positioning on the sample

being imaged (Diaspro, 2001; Diaspro and Chirico, 2003). Acquisition

and visualization are generally completely computer controlled by ded-

icated software. Figure 11–9 shows a TPE microscope.

Let us now consider two popular approaches that can be used to

perform TPE microscopy, namely, the descanned and nondescanned

Figure 11–8. Optical conﬁ guration for a TPE microscope operating in a descanned (upper inset box)

and nondescanned (lower inset box) mode; see text. (Courtesy of M. Cannel and C. Soeller.)

Chapter 11 Two-Photon Excitation Fluorescence Microscopy 771

modes. They are skectched in Figure 11–8. The former uses the very

same optical pathway and mechanism employed in confocal laser

scanning microscopy. The latter mainly optimizes the optical pathway

by minimizing the number of optical elements encountered on the

way from the sample to detectors, and increases the detector area. The

TPE nondescanned mode provides very good performances giving a

superior signal-to-noise ratio inside strongly scattering samples

(Masters et al., 1997; Daria et al., 1998; Centonze and White, 1998;

So et al., 2000).

In the descanned approach pinholes are removed or set to their

maximum aperture and the emission signal is captured using an exci-

tation scanning device on the back pathway. For this reason it is called

the descanned mode. In the latter, the confocal architecture has to be

modiﬁ ed in order to increase the collection efﬁ ciency: pinholes are

removed and the emitted radiation is collected using dichroic mirrors

on the emission path or external detectors without passing through the

galvanometric scanning mirrors. A high-sensitivity detection system

is another critical issue (Wokosin et al., 1998; So et al., 2000; Girkin and

Wokosin, 2002).

The ﬂ uorescence emitted is collected by the objective and transferred

to the detection system through a dichroic mirror along the emission

path (Figure 11–8). Due to the high excitation intensity, an additional

barrier ﬁ lter is needed to avoid mixing the excitation and emission light

at the detection system that is differently placed depending on the

acquisition scheme being used. Photodetectors that can be used include

photomultiplier tubes, avalanche photodiodes, and CCD cameras

(Denk et al., 1995; Murphy, 2001). Photomultiplier tubes are the most

commonly used. This is due to their low cost, good sensitivity in the

blue-green spectral region, high dynamic range, large size of the sensi-

tive area, and single-photon counting mode availability (Hamamatsu

Photonics, 1999). They have a quantum efﬁ ciency around 20–40% in

the blue-green spectral region that drops down to <1% moving to the

red region. This is a good condition, especially in MPE mode, because

it is desirable to reject as much as possible wavelengths above 680 nm

Figure 11–9. The TPE setup at LAMBS, Micro ScoBio Research Center of the

University of Genoa (from left to right: Ilaria Testa, Paolo Bianchini, and

Davide Mazza).

772 A. Diaspro et al.

that are mainly used for excitation. Another advantage is that the large

size of the sensitive area of photomultiplier tubes allows efﬁ cient col-

lection of signal in the nondescanned mode within a dynamic range

of the order of 10

. Avalanche photodiodes are excellent in terms of

sensitivity exhibiting quantum efﬁ ciency close to 70–80% in the visible

spectral range. Unfortunately they are high in cost and the small active

photosensitive area, <1 mm size, could introduce drawbacks in the

detection scheme and requires special descanning optics (Farrer et al.,

1999). CCD cameras are used in video rate multifocal imaging (Fuijta

and Takamatsu, 2002; Girkin and Wokosin, 2002). However, once the

best quality image possible has been obtained then image restoration

algorithms can be applied to enhance the features of interest to the

biological researcher and to improve the quality of data to be used

for three-dimensional modeling, such as those used for single-

photon optical sectioning microscopy, available at http://www.

powermicroscope.com (van der Voort et al., 1995; Shotton, 1995; Diaspro

et al., 1990, 2000; Boccacci and Bertero, 2002; Carrington, 2002; Difato

et al., 2004; Bonetto et al., 2004).

Laser sources, as often happened in optical microscopy, represent an

important resource, especially in ﬂ uorescence microscopy (Gratton

and van de Ven, 1995; Svelto, 1998). For nonresonant TPE, owing to the

comparatively low TPE cross-sections of ﬂ uorophores, high photon ﬂ ux

densities are required, >10

photons cm

−2

−1

(König, 2000). Using radia-

tion in the spectral range of 600–1100 nm for TPE, excitation intensities

in the MW–GW cm

−2

range are required. This high energy can be

obtained by the combined use of focusing lens objectives and CW

(Hanninen and Hell, 1994; König et al., 1995) or pulsed (Denk et al.,

1990) laser radiation of 50 mW mean power or less (Girkin and Wokosin,

2002; Diaspro and Sheppard, 2002). TPE microscopes have been real-

ized using CW, femtosecond, and picosecond laser sources (Periasamy,

2001; Diaspro, 2001, 2002; Masters, 2002). Since the original successful

experiments in TPE microscopy, advances have been made in the tech-

nological ﬁ eld of ultrashort pulsed lasers. Today laser sources suitable

for TPE can be described as “turnkey” compact systems (Fisher et al.,

1997; Wokosin et al., 1996; Diaspro, 2001).

Figure 11–10 shows a new generation ultrafast Ti:sapphire laser

source. The emission range between 700 and 1050 nm of the Ti:sapphire

laser allows a large number of commonly used ﬂ uorescent molecules

to be excited. Other laser sources used for TPE are Cr-LiSAF, pulse-

compressed Nd-YLF in the femtosecond regime, and mode-locked

Nd-YAG and picosecond Ti-sapphire lasers in the picosecond regime

(Gratton and Van de Ven, 1995; Wokosin et al., 1996). Most of the laser

sources used for TPE operate in a mode-locking mode. This endows

the laser with the ability to generate a train of very short pulses by

modulating the gain or excitation of a laser at a frequency with a period

equal to the roundtrip time of a photon within the laser cavity (Fisher

et al., 1997; Svelto, 1998) (Figure 11–11). The resulting pulsewidth is in

the 50–150 fs regime. The parameters that are more relevant in the

selection of the laser source are average power, pulsewidth and re-

petition rate, and wavelength also according to Eq. (15). The most

Chapter 11 Two-Photon Excitation Fluorescence Microscopy 773

popular features for an infrared pulsed laser are 700 mW–1 W average

power, 80–100 MHz repetition rate, and 100–150 fs pulse width.

At present, the use of short pulses and small duty cycles are manda-

tory to allow image acquisition in a reasonable time while using power

levels that are biologically tolerable (Denk et al., 1994; Denk, 1996;

Koester et al., 1999; König et al., 1996, 1998; König, 2000; König and

Tirlapur, 2002). To minimize pulse width dis persion problems König

(2000) suggested work ing with pulses around 150–200 nm, and this

constitutes a very good compromise both for pulse stretching and

sample viability. It should always be remembered that a shorter pulse

broadens more than a longer one. Pulse width measurement is a very

delicate issue. In fact, because it is not very easy to measure it at the

focal volume within the sample, little can be deﬁ nitely said about it

(Hanninen and Hell, 1994; Guild et al., 1997; Wolleschensky et al., 2002).

Although users do not perform measurement of the pulse width at the

Figure 11–10. Typical laser sources in use for TPE microscopy.

Figure 11–11. Laser emission time scale for TPE excitation: a short pulse at

high photon density is released for approximately 100 fs; this laser shot is able

to prime ﬂ uorescence without damaging the sample so ﬂ uorescence occurs in

the next few nanoseconds. The laser is silent for 10 ns and then delivers a new

high-density photon pulse. This modality allows TPE to be experienced at

tolerable time-averaged power (see text).

774 A. Diaspro et al.

sample when they use two-photon microscopy, which would require

a speciﬁ c procedure that, even if not too complex for a researcher in

the ﬁ eld, could be irksome for the majority of users, it is a reasonable

approximation to assume that at the focal volume, 1.5–2. times tempo-

ral pulse broadening occurs using high-quality optics (Wolleschensky,

2002; Girkin and Wokosin, 2002). As an example, for a measured laser

pulse width of about 100 fs, an estimate at the sample is about 150–180 fs

under favorable experimental conditions, sample characteristics

included. Sample properties are mentioned because for thick samples

the role played by thickness, also in terms of pulse width broadening,

is not so obvious (de Grauw and Gerritsen, 2002; So et al., 2001; Gu

et al., 2000; Saloma et al., 1998).

8 Conclusion

Confocal microscopy, in the authors’ opinion, constitutes one of the

most signiﬁ cant advances in optical microscopy within the past

decades, and has become a powerful investigative tool for the molecu-

lar, cellular, and developmental biologist, the materials scientist, the

biophysicist, and the electronic engineer. It is entirely compatible with

the range of “classical” light microscopic techniques, and, at least in

scanned beam instruments, can be applied to the same specimens on

the same optical microscope stage. Its peculiar advantages result in its

ability to generate multidimensional (x–y–z–t) images by noninvasive

optical sectioning with a virtual absence of out-of-focus blur, its capac-

ity for multiparametric imaging of multiply labeled samples, and its

property of investigating at microscopic resolution large objects as a

result of the rejection of scattered light. So far, the advent of confocal

microscopy in the mid-1980s favored the rapid spreading of two- and

multiphoton excitation microscopy, since Denk’s report at the begin-

ning of the 1990s, bringing dramatic changes in designing experiments

that utilize ﬂ uorescent molecules and, more speciﬁ cally, in ﬂ uorescence

3D optical microscopy.

While confocal microscopy is moving to spectral and fast-scanning

architectures in terms of acquisition, it is mainly two-photon micro-

scopy that occupies the scene of advances in ﬂ uorescence optical

microscopy. TPE micro scopy, with its intrinsic three-dimensional reso-

lution, the absence of background ﬂ uorescence, and the attractive pos-

sibility of exciting UV excitable ﬂ uorescent molecules, thus increasing

sample penetration, constitutes signiﬁ cant progress in science. In fact,

in a TPE scheme two 720-nm photons combine to produce the very

same ﬂ uorescence conventionally primed at ∼360 nm, and to be utilized

in a classical confocal microscope using conventional excitation of ﬂ uo-

rescent molecules. The excitation of the ﬂ uorescent molecules bound

to the speciﬁ c components of the biological systems being studied

mainly takes place (80%) in an excitation volume of the order of mag-

nitude of 0.1 ﬂ . This results in an intrinsic 3D optical sectioning effect.

What is invaluable for cell imaging and, in particular, for live-cell

imaging is the fact that weak endogenous one-photon absorption and

Chapter 11 Two-Photon Excitation Fluorescence Microscopy 775

highly localized spatial conﬁ nement of the TPE process dramatically

reduce phototoxicity stress. To summarize the unique characteristics

and advantages of TPE we recall the following properties:

1. Spatially conﬁ ned ﬂ uorescence excitation in the focal plane of the

specimen can be con sidered the key feature of TPE microscopy. It is one

of the advantages over confocal micro scopy, where ﬂ uorescence emis-

sion occurs across the entire thickness of the sample being excited by the

scanning laser beam. A strong implication is that there is no photon

signal from sources out of the geometric position of the optical focus

within the sample. Therefore, the signal-to-noise ratio increases, photo-

Figure 11–12. Multiple excitation of three ﬂ uorescent dyes using 740 nm u nder a TPE regime. The

conventional excitation would have required the utilization of 360 or 405 nm, 488 nm, and 543 nm laser

lines. The ﬁ nal image (lower right quadrant) is realized by merging the three subsets. (This image has

been acquired by students of the Biotechnology School during the course of Advanced Microscopy

Techniques activated at the University of Genoa, academic year 2005. Advisors: Grazia Tagliaﬁ erro

and Alberto Diaspro.) (See color plate.)

776 A. Diaspro et al.

degradation effects decrease, and optical sectioning is immediately

available without the need for pinhole or deconvolution algorithms. In

addition, very efﬁ cient acquisition schemes can be implemented such as

the nondescanned one operating at an excellent signal-to-noise ratio.

2. The use of near-infrared (IR)/IR wavelengths permits examina-

tion of thick specimens in depth. This is due to the fact that, apart from

some cases such as pigmented samples and portions of the absorption

spectral window of water, cells and tissues absorb poorly in the near-

IR/IR region. Cellular damage is globally minimized, thus allowing

cell viability to be prolonged with long-term 3D sessions. Moreover,

scattering is reduced and deeper targets can be reached with fewer

problems than in one-photon excitation. The depth of penetration can

be up to 0.5 mm. In addition, whereas in one-photon excitation, the

emission wavelength is comparatively close to the excitation one (about

50–200 nm longer), in TPE the ﬂ uorescence emission occurs at a wave-

length substantially shorter and at a larger spectral distance than in

one-photon excitation. Thus separation of the excitation light and the

emitted light can be easily performed.

Continuing research in this ﬁ eld is focused on very intriguing prob-

lems (www.focusonmicroscopy.org offers a complete scenario of the

evolution of three-dimensional microscopy in the past 5 years) such as

local heating from absorption of IR light by water at high laser power

(Schonle and Hell, 1998) and photothermal effects on ﬂ uorescent mol-

ecules (Chirico et al., 2003a), phototoxicity from long wavelength IR

excitation and short wavelength ﬂ uorescence emission (König et al.,

1996c; Tyrrel and Keyse, 1990; König, 2000; Hopt and Neher, 2001;

König and Tirlapur, 2002), photoactivation and photocycling of visible

ﬂ uorescent proteins (Post et al., 2004; Chirico et al., 2004; Schnedier

et al., 2005), development of new ﬂ uorochromes better suited for TPE

and multiphoton excitation (Albota et al., 1998a; Abbotto et al., 2005),

and the investigation of the cross-sections of uncharacterized mole-

cules (Gostkowski et al., 2004; Wokosin et al., 2004).

One of the major beneﬁ ts in setting up an MPE microscope is the

ﬂ exibility in choosing the measurement modality favored by the

simpliﬁ cation of the optical design. In fact, a TPE microscope offers a

greater variety of measurement options without changing any optics

or hardware. This means that during the very same experiments real

multimodal information can be obtained from the specimen being

studied (Zoumi et al., 2002; Wang et al., 2004). Moreover, the usefulness

of the TPE scheme for spectroscopic and lifetime studies (So et al.,

1996; Sytsma et al., 1998; Schwille et al., 2000; Diaspro et al., 2001;

Wiseman et al., 2002), for optical data storage and microfabrication

(Cumpston et al., 1999; Kawata et al., 2001), and for single molecule

detection (Mertz et al., 1995; Farrer et al., 1999; So et al., 2000; Chirico

et al., 2001; Cannone et al., 2003b) has been well documented. Other

very interesting applications involve the study of impurities affecting

the growth of protein crystals (Caylor et al., 1999), TPE imaging in the

ﬁ eld of plant biology (Tirlapur and König, 2002), and measurements in

living systems (Squirrel et al., 1999; Yoder and Kleinfeld, 2002; Diaspro

et al., 2002b,d; Post et al., 2004). Here the combination of MPE and

Chapter 11 Two-Photon Excitation Fluorescence Microscopy 777

second-harmonic generation offers the opportunity to investigate the

morphometric properties on the basis of the microstructure of blood

cells (Zoumi et al., 2004). Another promising ﬁ eld is the investigation

of complex formation where the TPE properties will improve the infor-

mation accessible (Heinze et al., 2004). Another, more indirect usage

that provides a look at the sample with nanometer resolution is the

excitation of an evanescent wave at a metal surface (Novotny et al.,

1998). For microscopic purposes the evanescent wave needs to be local-

ized at a nanoparticle or a ﬁ ne metal tip (Sánchez et al., 1999; Gerton

et al., 2004). The MPE microscope can also be used as an active device,

with increasing applications related to nanosurgery (König, 2000),

selective uncaging of caged compounds (Diaspro et al., 2003), and

photodynamic therapy (Bhalwalkar et al., 1997; So et al., 2000). Recently

TPE microscopy, even if in an evanescent-ﬁ eld-induced conﬁ guration,

has been extended to large area structures of the order of square cen-

timeters (Duveneck et al., 2001). This has application in the realization

of biosensing platforms such as genomic and proteomic microarrays

based upon large planar waveguides. It is easy to perceive that the

range of applicability of MPE microscopes is rapidly increasing in the

biomedical, biotechnological, and biophysical sciences and is expand-

ing to clinical applications (Diaspro, 2002; Masters, 2002; Periasamy

and Diaspro, 2003).

Acknowledgments. The ﬁ rst Italian TPE architecture realized at LAMBS

has been supported by INFM grants. LAMBS-MicroScoBio is currently

funded by IFOM (Istituto FIRC di Oncologia Molecolare, FIRC Institute

of Molecular Oncology, Milano). This chapter is dedicated to the

memory of Osamu Nakamura, who passed away January 23, 2005 at

Handai Hospital.

References

Abbe, E. (1910). In: Die Lehre von der Bildentstehung in Mikroskop (O.

Lummer, Ed.). F. Reiche, Braunschweig.

Abbotto, A., Baldini, G., Beverina, L., Chirico, G., Collini, M., D’alfonso, L.,

Diaspro, A., Magrassi, R., Nardo, L. and Pagani, G.A. (2005). Dimethyl-

Pepep: A Dna probe in two-photon excitation cellular imaging. Biophys.

Chemi. 114(1), 35–41.

Agard, D.A. (1984). Optical sectioning microscopy: Cellular architecture in

three dimensions. Annu. Rev. Biophys. 13, 191–219.

Agard, D.A., Hiraoka, Y., Shaw, P.J. and Sedat, J.W. (1989). Fluorescence

microscopy in three-dimensions. Methods Cell. Biol. 30, 353–378.

Albota, M., Beljonne, D., Bredas, J.L., Ehrlich, J.E., Fu, J.Y., Heikal, A.A., Hess,

S.E., Kogej, T., Levin, M.D., Marder, S.R. and others. (1998a). Design of

organic molecules with large two-photon absorption cross sections. Science

281(5383), 1653–1656.

Albota, M.A., Xu, C. and Webb, W.W. (1998b). Two-photon ﬂ uorescence excita-

tion cross sections of biomolecular probes from 690 to 960 nm. Appl. Opt. 37,

7352–7356.

Amos, B. (2000). Lessons from the history of light microscopy. Nature Cell Biol.

2, E151–E152.

Andrews, D.L. (1985). A simple statistical treatment of multiphoton absorption.

Am. J. Phys. 53, 1001–1002.