nonlinear optical imaging of extracellular matrix proteinsidgateway.wustl.edu › fitzpatrick ›...
TRANSCRIPT
CHAPTER
Nonlinear optical imaging ofextracellular matrix proteins 3
Chien-Cheng Shih, Dennis M. Oakley, Matthew S. Joens, Robyn A. RothJames A.J. Fitzpatrick1
Washington University Center for Cellular Imaging,
Washington University School of Medicine, St. Louis, MO, United States1Corresponding author: e-mail address: [email protected]
CHAPTER OUTLINE
1 Introduction..........................................................................................................58
2 Methods ...............................................................................................................61
2.1 Theoretical Principles of MHG ................................................................61
2.2 Implementing MHIM .............................................................................63
2.2.1 Instrumentation..................................................................................63
2.2.2 Image Acquisition...............................................................................65
2.2.3 Image and Data Processing ................................................................66
2.3 Sample Preparation ...............................................................................66
2.3.1 Materials ............................................................................................66
2.3.2 Sample Collection and Fixation ...........................................................66
3 Imaging the Microstructure of Mice Aorta Using MHIM ...........................................66
3.1 Detecting Mirror-Augmented MHG Signals From the Mouse Aorta ..............67
3.2 Spectral Comparison of MHG and Two-Photon Excited Autofluorescence
Highlights Differences in Tissue Microstructure .......................................69
3.3 Visualizing the Microstructures of Mouse Aorta Using Nonlinear Optical
Microscopy ...........................................................................................71
4 Conclusions..........................................................................................................74
Acknowledgments ..................................................................................................... 74
References ............................................................................................................... 75
AbstractOver the last 2 decades, nonlinear imaging methods such as multiharmonic imaging micros-
copy (MHIM) have become powerful approaches for the label-free visualization of biological
structures. Multiharmonic signals are generated when an intense electromagnetic field
Methods in Cell Biology, Volume 143, ISSN 0091-679X, https://doi.org/10.1016/bs.mcb.2017.08.004
© 2018 Elsevier Inc. All rights reserved.57
propagates through a sample that either has a specific molecular orientation or exhibits
certain physical properties. It can provide complementary morphological information when
integrated with other nonlinear optical imaging techniques such as two-photon excitation
(TPE). Here, we present the necessary methodology to implement an integrated approach
for multiharmonic and TPE imaging of the mouse aorta using a commercial two-photon mi-
croscope. This approach illustrates how to differentiate the microstructure of the mouse aorta
that are due to collagen fibrils and elastic laminae under 820 and 1230nm excitation. Our
method also demonstrates how to perform multiharmonic generation by reflectance of the for-
wardly propagating emission signal. The ability to visualize biological samples without addi-
tional genetically targeted or chemical stains makes MHIM well suited for studying the
morphology of the mouse aorta and has the potential to be applied to other collagen and
elastin-rich tissues.
1 INTRODUCTIONThe utilization of multiharmonic imaging microscopy (MHIM) to visualize biolog-
ical structures has become more prevalent over the past 2 decades. This approach,
based on a nonlinear optical (NLO) effect known as multiharmonic generation
(MHG) has been established as a viable microscope imaging contrast mechanism
for the visualization of in situ cellular and tissue function without the need for chem-
ical or biochemical labels (Campagnola & Loew, 2003; Freund, Deutsch, &
Sprecher, 1986; Tai, Kung, Yu, Huang, & Chan, 2007; Weigelin, Bakker, &
Friedl, 2016). Owing to the necessity of using lower energy infrared wavelengths
for harmonic generation, the process facilitates both optical sectioning and deeper
penetration into turbid milieu. This provides inherent three-dimensional (3D) struc-
tural information in both in vitro and ex vivo sample types (Bayan, Levitt, Miller,
Kaplan, & Georgakoudi, 2009; Osman et al., 2013; Rehberg, Krombach, Pohl, &
Dietzel, 2011). Typically, second-harmonic generation (SHG) requires a material
to have a specific molecular orientation in order for the incident light to be frequency
doubled. However, some biological materials can be highly polarizable, and assem-
ble into fairly ordered, large noncentrosymmetric structures. Biological materials
such as collagen (Freund et al., 1986; Friedl, Wolf, von Andrian, & Harms, 2007)
and microtubules (Kasischke & Hyman, 2003) can produce SHG signals. Further-
more, the presence of water–lipid or water–protein interfaces (Chen & Sun, 2009;
Jay et al., 2015; Tai et al., 2007) can elicit third-harmonic generation (THG), or
frequency-tripled signals. Because these processes are nonlinear in nature and do
not involve molecular excitation (as in the case of fluorescence microscopy) the ef-
fects of phototoxicity and photobleaching are not a concern (Brakenhoff, Squier,
Wilson, & M€uller, 1998; D�ebarre, Olivier, Supatto, & Beaurepaire, 2014). With
these obvious advantages, MHIM microscopy has been widely employed in studies
of the extracellular matrix (Weigelin, Bakker, & Friedl, 2012; Wu, Hsieh, Tsai, &
Liu, 2015), inorganic structures (Chu et al., 2005; Gannaway & Sheppard, 1978;
Hellwarth & Christensen, 1974; Marangoni, Lobino, & Ramponi, 2006), and cellular
58 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
membranes/organelles in both fixed and living specimens (Chen & Sun, 2009;
Rehberg et al., 2011; Weigelin et al., 2012).
SHG, also known as frequency doubling, is a NLO effect that occurs when an in-
tense electromagnetic field passes through amaterial with a noncentrosymmetricmo-
lecular structure (Campagnola & Loew, 2003; Campagnola, Wei, Lewis, & Loew,
1999). Second-harmonic light emerging from a SHG material is exactly half the
wavelength (i.e., frequency doubled) of the light entering the material. Two incident
photons are combined into one photon (second harmonic) with a higher energy, which
propagates in the same direction as the incident photons. Typically, an inorganic crys-
tal such as lithium niobate (LiNbO3) potassium titanyl phosphate (KTP¼KTiOPO4)
or lithium triborate (LBO¼LiB3O5) are used to produce visible light from near-
infrared lasers or UV light from visible lasers in the optics industry. Early research
by Hellwarth and Christensen (1974) and Gannaway and Sheppard (1978) suggests
that the observation of SHG induced by biological specimens under microscope il-
lumination is theoretically possible, assuming illumination with a pulsed, high
peak-power laser source. However, second-harmonic imaging was not attempted
on a biological sample until themid-80swhen the dipolar formation of collagen fibers
in the rat-tail tendonwas elucidated (Freund et al., 1986). This study demonstrated the
potential of second-harmonic imaging for the visualization of biological ultrastruc-
ture based on the molecular organization of noncentrosymmetric macromolecules
such as collagen. Types I and II collagen, which form aligned fibers, are primarily
found in skin, tendon, vascular ligature, bone, and cartilage (Chen, Nadiarynkh,
Plotnikov, & Campagnola, 2012). Being able to image fibrillar collagen without
the addition of an exogenous fluorescent label broadly expands the application of this
technique. In addition, the other protein-based macromolecules such as the acto–myosin complex and tubulin are also SHG active (Campagnola et al., 2002;
Mohler, Millard, & Campagnola, 2003). Owing to the number of SHG-active mac-
romolecules, second-harmonic imaging has allowed the extraction of a substantial
amount of structural information on cancerous cells and general tissue organization
which has solidified its impact in biomedical research (Jay et al., 2015; Pfeffer, Olsen,
Ganikhanov, & L�egar�e, 2011; Wu et al., 2015).
THG known as frequency tripling also results from the summation of photons
induced by specific physical properties of the sample (Brakenhoff et al., 1998;
Weigelin et al., 2016). THG triples the energy and generates a single photon with
a wavelength one-third of the incident photons. Unlike SHG though, THG occurs
mainly at interfaces between materials of differing excitability, such as the refraction
index mismatch between water–lipid or water–protein interfaces, or resonant en-
hancement by chromatic materials such as hemoglobin. Since the presence of a non-
centrosymmetric structure is not required for THG, its applications are more wide
ranging than SHG. These have included studying collagen and elastic fibers
(Rehberg et al., 2011), calcified teeth (Chen, Hsu, & Sun, 2008), cellular and nuclear
membranes (Aptel et al., 2010; Chen, Wu, & Sun, 2009), adipocytes (Tsai et al.,
2013), and neurites (Farrar, Wise, Fetcho, & Schaffer, 2011; Lim et al., 2014). In
addition, this lower specificity makes manipulation of the THG signal easier for
591 Introduction
visualizing separate structures. For example, altering the axial position of the focal
volume allows for the illumination of different cellular components in Drosophila
embryos thus facilitating discrimination of the THG signal produced by external
membrane vs internal organelles (Aptel et al., 2010; Mahou et al., 2011). Another
example is changing the angle between the sample interface and the incident laser
can facilitate detection of different orientations of fibrillar collagen (Cheng &
Xie, 2002; Weigelin et al., 2016). All these features contribute to the increased ver-
satility THG microscopy enjoys as opposed to its SHG relative.
A common setup for a MHIM system can be routinely achieved by proper mod-
ification of a commercial laser-scanningNLOmicroscope equipped with a titanium
sapphire mode-locked laser as the illumination source. Acquisition of the SHG
signal (with a frame rate and pixel density similar to that of a laser-scanning con-
focal microscope) can be routinely achieved by the selection and installation of an
appropriate dichroic mirror and band-pass filter combination in front of the photo-
detector. However, for THG, both the illumination objective lens and detection
light path need to be optimized for the highest sensitivity. The availability of
high-quantum efficiency gallium arsenide phosphide (GaAsP) photomultipliers
drastically improves signal-to-noise ratios and thus lowers the barrier for THG
imaging on biological specimens. Given its increased popularity, integration of
nonlinear SHG and THG imaging with two-photon excited fluorescence (TPEF)
has become a common practice that offers complementary information. For exam-
ple, in myofibers, SHG signals mainly result from myosin that is localized in the -
A-band of the sarcomere, while THG signals result from the actin-rich I-bands that
reside between sarcomeres (Rehberg et al., 2011; Rehberg, Krombach, Pohl, &
Dietzel, 2010; Weigelin et al., 2012). Conversely, 3D SHG/THG imaging of
a collagen fiber network, data illustrate complementary patterns between the
SHG and the THG signals that result due to the orientation of the fibrillar collagen
(Weigelin et al., 2016).
In this chapter, we will focus on describing the theoretical principles of MHIM,
the practical implementation of the approach to image cross-sections of mouse aorta
along with the necessary steps for suitable sample preparation. One of the first im-
aging paradigms that paired SHG with two-photon excitation (TPE) was that of
Zoumi, Lu, Kassab, and Tromberg (2004). They made use of an excitation wave-
length of 800nm and successfully separated SHG and TPEF signals from coronary
artery in ranges of 390–410 and 500–540nm. This approach allowed the capture of
the morphology of collagen fibrils in the outer layer (adventitia) by SHG and elastin
fibrils in the middle layer (medial) by TPEF. This approach was later modified for
imaging arterial microstructures under 860nm excitation (Boulesteix et al., 2006).
We describe a further modification of the earlier approach by demonstrating how
to induce and capture the augmented third-harmonic signal by reflectance of the for-
wardly propagating THG signal by a dielectric mirror. Such a methodology has been
used previously to enhance MHIM signals as it serves to improve the signal-to-noise
ratio on the backward scattered detector (Rehberg et al., 2010). We have utilized this
60 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
paradigm for evaluating the microstructure of the mouse aorta. Since the polymeric
materials collagen and elastin are two well-characterized components present in
blood vessels that aid in maintaining healthy function, MHIM paired with TPEF im-
aging serves as a powerful, label-free microscopy method for the study of the mor-
phology and pathology of the vascular system.
2 METHODS2.1 THEORETICAL PRINCIPLES OF MHGThe nonlinear polarization of a given material can be expressed by a Taylor series
expansion of the dielectric polarization density (dipole moment per unit volume) P(t)at time t in terms of electrical field E(t):
P tð Þ¼ E0 w 1ð ÞE tð Þ + w 2ð ÞE2 tð Þ+ w 3ð ÞE3 tð Þ+⋯� �
(1)
where the coefficients w(n) are the nth order nonlinear susceptibilities of the material,
w(1) describes the normal absorption and reflection processes; w(2) describes SHG,sum, and difference frequency generation; and w(3) describes THG, multiphoton ab-
sorption, and light scattering (Barad, Eisenberg, & Horowitz, 1997; Campagnola
et al., 1999).
SHG, which dominates the second-order term, is highly restricted to
materials with noncentrosymmetric molecular organization. The susceptibility ten-
sor of second-order w(2) is a bulk property that is related to molecular hyperpolariz-
ability by:
w 2ð Þ ¼Ns bh i (2)
where the Ns is the molecular density of the material and hbi is expectation value ofthe molecular orientations in the material, i.e., the orientational average. The require-
ment for noncentrosymmetric symmetry is necessary to prevent hbi from vanishing
for an isotropic distribution of dipole moments, i.e., to keep w(2) nonzero. Type-I/IIcollagen and acto–myosin complexes are the two main macromolecular assemblies
that satisfy this requirement. In contrast, nonfibrillar collagen, such as type-IV, fails
to produce any appreciable SHG signal due to the lack of a noncentrosymmetric
structure. In the case of THG, however, the signal is generated from the third-order
susceptibility factor (w(3)). Thus in tissues, THG occurs predominantly at water–lipidor water–protein interfaces, or within media that exhibits significant optical hetero-
geneity, such as lipid-containing organelles and cells, the myelin sheath surrounding
nerve fibers as well as cellular membranes and intracellular vesicles. Unlike SHG,
which is highly specific to macromolecular structures that are noncentrosymmetric,
the intensity of the generated THG signal is highly dependent on the interaction be-
tween the focal volume and interface of changing refractive index or w(3).
612 Methods
The energy-level schema ofMHG and TPE are depicted in the Jablonsky diagram
shown in Fig. 1A. In the process of TPE, two lower energy photons (typically in the
infrared region of the electromagnetic spectrum) near-simultaneously excite a fluor-
ophore from its ground electronic state (S0) to its first- and second-excited electronic
FIG. 1
Principle of multiharmonic generation. (A) Jablonsky diagram comparing resonance
enhanced two-photon excitation (TPE), second-harmonic generation (SHG), and third-
harmonic generation (THG). Dashed lines, virtual states; S0, ground state; S1 and S2,
excitation states; thin lines, vibrational states. (B) The diagram indicating the range of
nondescanned detector (NDD), multiharmonic generation (MHG), and autofluorescence of
elastin (TPE). Two PMTs of the NDD were equipped with these filters: 380–430 and
600–750nm (green and pink boxes). SHG under 820nm excitation and THG under 1230nm
excitation are shown as blue box (around 410nm). The range of autofluorescence from
elastin is indicated by dark red box, induced by TPE under either 820nm or 1230nm.
62 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
states (S1 and S2). This fluorophore, by virtue of intramolecular vibrational redistri-
bution dissipates some of this excited state energy as heat but releases the majority of
it by emission of a photon of light (typically in the visible region of the electromag-
netic spectrum) as it relaxes back to its ground electronic state. Instead of the char-
acteristic absorption and emission processes that are seen in TPE, SHG, and THG are
induced by the presence of a very intense field that results in the production of a non-
linear polarization inside the focal volume. Because the SHG and THG signals are
coherent waves at double or triple the incident frequency, molecules that contribute
to either the SHG signal or THG signal are no longer considered to be point emission
sources as in the case of a fluorescing object. As such, the direction of the nonlinear
harmonics critically depends on the spatial distribution of the molecules as well as
the dipole moment of the driving field. Thus the generated harmonics propagate in a
direction that is parallel to the incident beam.
In order to settle on an incident wavelength for MHIM, one must find the balance
between light scattering and absorption, especially in the case of THG imaging. From
a theoretical perspective, longer wavelengths (>1350nm) serve to reduce light scat-
tering, but they suffer from a reduced tissue penetration due to the absorption effects
of water. Conversely, shorter wavelengths (<1100nm) would induce THG in the ul-
traviolet (UV) range, which is highly scattered by tissue and has low transmission
through optical glass components. Hence, the ideal wavelength to induce THG
would be by using infrared excitation (1200–1300nm) thus generating THG signal
in the blue region of the electromagnetic spectrum (400–450nm). With this partic-
ular wavelength, one can also induce SHG simultaneously in the red region of the
electromagnetic spectrum (600–650nm) (Jay et al., 2015, 2008; Rehberg et al.,
2011;Wu et al., 2015). In our method described below, we have chosen a wavelength
of 820nm for SHG induction and 1230nm for THG induction (Fig. 1B). This enables
the simultaneous acquisition of the MHG/TPE signals on a pair of nondescanned de-
tectors (NDDs), which are equipped with a dichroic mirror and two band-pass filters,
which serve to separate and capture the spectral ranges of 380–430 and 580–670nm,
respectively.
2.2 IMPLEMENTING MHIM2.2.1 InstrumentationWe implemented this methodology on a commercial Zeiss LSM 880 NLO two-
photon microscope system installed on an Axio Observer Z.1 inverted microscope
frame (Carl Zeiss Microscopy GmbH, Jena, Germany) equipped with a pulsed, tun-
able infrared laser source (ChameleonDiscovery, Coherent Inc., CA,USA) (Fig. 2A).
The Chameleon Discovery laser generates emission in the range 680–1300nm with a
pulse duration of�100fs at a repetition rate of 80�0.5MHz. The output at 1230nm
was �1100mW. The average intensity at the sample plane was 8.96�0.09mW
at 820nm and 35.8�0.07mW at 1230nm. A Plan-Apochromat 63� 1.4NA oil
immersive objective lens with a working distance of 0.19mm was used for imaging
(Carl Zeiss Microscopy GmbH, Jena, Germany). During MHIM, both SHG and
THG signals were detected by two nondescanned gallium arsenide phosphide photo-
multiplier tubes (GaAsP PMTs) (Zeiss BiG-2, Carl Zeiss Microscopy GmbH, Jena,
632 Methods
FIG. 2
Multiharmonic imaging microscopy. (A) Schematic overview of Zeiss LSM 880 NLO imaging
system. The tunable femtosecond laser is equipped for NLO imaging on a laser scanning
system, controlled by the galvano mirror. Emission fluorescent or harmonic signals
propagated backward to the high sensitivity PMTs in the nondescanned detector (NDD) or the
spectral PMT in the descanned detector. Enhancement by the dielectric mirror reflection can
be optional. (B) The reflection of SHG and THG signals when mirror is presence.
This setup was adapted from a method described by Rehberg, M., Krombach, F., Pohl, U., & Dietzel, S., 2010.
Signal improvement in multiphoton microscopy by reflection with simple mirrors near the sample. Journal of
Biomedical Optics, 15(2), 026017. https://doi.org/10.1117/1.3374337.
64 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
Germany). The incoming light to the BiG-2 detector assembly was split by a 468nm
short pass dichroic mirror and filtered by two band-pass filters, one in the blue
(380–430nm) and one in the far-red (580–670nm) (Semrock, NY, USA) channels.
To confirm the spatial pattern of autofluorescence elicited by TPE, we applied
820nm laser excitation and detected the fluorescence emission in multiple ranges
using a descanned GaAsP spectral detector (QUASAR array) equipped within
the Zeiss LSM 880 NLO scan head. To enhance the MHG signals, a broadband
dielectric mirror (BB2-E02, 400–750nm, Thorlabs Inc., NJ, USA) was placed on
the back of sample slide to reflect the forward scattered signal to the backward
facing detectors (Fig. 2B).
2.2.2 Image acquisitionSingle image planes and z-stack image data were acquired using the ZEN software
(ZEN 2.3 SP1 Black, Carl ZeissMicroscopy, Jena, Germany). To enhance the signal-
to-noise ratio, the scan speed was set to 1, which corresponds to a pixel dwell time to
65.94ms/pixel, and the frame size was set to either 512 �512 with an optical zoom
factor of 1.5� (0.18mm/pixel) or 1024 �1024 without no optical zoom (0.13mm/
pixel). For multiharmonic imaging using the NDD, increasing the digital gain to
8.0 significantly improved the quality of the acquired THG image. However, care
must be taken when using digital gain as it is simply a multiplication factor. While
the signal intensity will be enhanced, so will any noise present in the image. In con-
trast, the SHG and the TPE signals were much stronger than the THG signal, so the
use of digital gain was not required. Also, due to the low signal obtained from THG,
the higher dynamic range provided by a 16-bit depth acquisition resulted in more
robust data for postprocessing and analysis. The voltage of GaAsP PMTs was set
to 900 and 600V for the acquisition of THG and SHG/TPE signals, respectively.
Owing to the high sensitivity of the GaAsP PMTs, it is especially important to main-
tain a dark imaging environment with proper blackout material. Furthermore, the
slides were secured onto the slide holders by an orthodontic wax to prevent any lat-
eral drift caused by the weight of the dielectric mirror. For TPE spectral imaging
using the descanned QUASAR spectral detector, the voltage was set between 700
and 900V with digital gain set to 1. For all imaging, no digital offset was applied.
The theoretical resolution of an optical system can be estimated by the Rayleigh
criterion:
dmin ¼ 0:61�l=NAobjective 3
where l is the wavelength, NAobjective is the numerical aperture of the objective lens,
and dmin is the smallest spatial distance that can be resolved. With a Plan-
Apochromat 63� 1.4NA oil immersion objective lens (Carl Zeiss Microscopy, Jena,
Germany) and the knownwavelength fromMHG, dmin is 174nm in the range of SHG
and THG induced at �800 and �1200nm, respectively, and 261nm for the TPE
autofluorescence emission at 600nm. No oversampling was considered as the image
pixel sizes were greater than a factor of 1/3 to 1/2 of dmin.
652 Methods
2.2.3 Image and data processingAll image data were saved as both proprietary Zeiss CZI as well as exported to TIFF
format for subsequent processing in ImageJ (Schindelin, Rueden, Hiner, & Eliceiri,
2015) and/or Fiji (Schindelin et al., 2012). The plugins necessary for visualization
purposes include Bio-Formats (Linkert et al., 2010) and ClearVolume (Royer
et al., 2015). All figures included in this chapter were created using the Adobe Cre-
ative Cloud Illustrator App (Adobe, CA, USA).
2.3 SAMPLE PREPARATION2.3.1 Materials1. Coverslip—Microscope cover glasses, size: 22 �50mm, thickness: #1.5
(160–190mm), Cat. no. 12-544-D (Thermo Fisher Scientific, MA, USA).
2. Slide—Microscope slides, precleaned, Cat. no. 12-550-003 (Thermo Fisher
Scientific, MA, USA).
3. Fixative—4% Paraformaldehyde (EM grade) in phosphate-buffered saline
(PBS).
4. Mounting media—ProLong® Gold (Thermo Fisher Scientific, MA, USA).
2.3.2 Sample collection and fixationMice at 2 months of age (maintained on a 12-h/12-h light/dark cycle with food and
water ad libitum) were used in preparing the aorta sections. Animals were first anes-
thetized with Avertin (Sigma-Aldrich Co., MO, USA), and then the heart and lungs
were then exposed by opening the chest cavity. The ascending aorta or carotid artery
was removed after several washes in PBS. Once dissected, an identified target region
in the tissue was first washed in the PBS to remove excess blood prior to fixation in
4% paraformaldehyde inside of microcentrifuge tubes stored at 4°C for 20h, fol-
lowed by two washes in PBS for 15min each. Samples were then dehydrated sequen-
tially in a graded ethanol series (30%, 50%, and 70%) for 30min each, and embedded
in paraffin with a Leica HistoCore Arcadia H (Leica Biosystems Inc., IL, USA). Tis-
sue sectioning was performed using a manual rotary microtome (RM2235, Leica
Biosystems Inc., IL, USA).
3 IMAGING THE MICROSTRUCTURE OF MICE AORTA USINGMHIMThe aorta is the main artery that is responsible for the systemic circulation of oxy-
genated blood from the heart to the circulatory system. In order to sustain the large
volume of blood that flows through it, the aorta consists of sheets (lamellae) of elastin
and bundles of fibrous collagen, which serve to provide and maintain the elasticity of
the arterial wall. Any change in the quantity or quality of the matrix components will
result in both a mechanical and functional alteration, which potentially develops into
disease pathology. To date, the architecture of the aorta has been most commonly
66 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
studied using pigment-based stains such as Sirius Red and Verhoeff’s solution
(Boulesteix et al., 2006; Jiao et al., 2017) which target elastin and collagen. Such
staining approaches are easy to perform but yield a product that can only be visual-
ized using brightfield optical microscopy, which lacks the ability to optically section.
Previous work have shown that by utilizing a transgenic animal expressing fluores-
cent reporters or by combining SHG with TPE microscopy under a singular excita-
tion wavelength (in the range 800–860nm) it is possible to acquire the 3D spatial
organization of both collagen and elastin (Boulesteix et al., 2006; Zoumi et al.,
2004). The strong autofluorescence signal from elastin and SHG signal that is in-
duced by fibrillar collagen can be captured simultaneously and used to elucidate
the microstructure of the aorta. We have recapitulated this combined SHG/TPE par-
adigm in our commercial NLO imaging system. We also demonstrate the possibility
for a combined SHG/THG/TPE imaging approach using a mirror-based augmenta-
tion methodology.
3.1 DETECTING MIRROR-AUGMENTED MHG SIGNALS FROM THEMOUSE AORTAAs described above, MHG signals typically propagate in the direction of the incident
beam, thus forward-scattered SHG and THG photons are substantially more preva-
lent than backward-scattered photons. To successfully detect MHG signals on a de-
tector in the backward-scattered direction, a straightforwardmethod is to enhance the
signal by reflecting forwardly propagating signal back to the backward-scattered de-
tector with a simple mirror. Such a methodology was first described in the imaging of
mouse cremaster muscle by Rehberg et al. In their experiment, SHG and THG signals
were increased by factors of 9.6 and 7.6 times, respectively, by the presence of an
aluminum-coated coverslip located behind the sample plane (Rehberg et al.,
2010). To reproduce this method on our system, we placed a broadband dielectric
mirror on top of the sample, which generated a significant enhancement of the
MHG signal. Using 820nm light, SHG signal induced by fibrillar collagen in the ad-
ventitia of a mouse aorta was easily detected in the spectral range 380–430nm(Fig. 3A). Despite this signal already being sufficient for detection, its intensity in-
creased approximately twofold to threefold in the center of the collagen bundles
(Fig. 3B). In contrast, the induced THG signal was so low without the presence
of the reflecting mirror (Fig. 3C), that it increased up to sevenfold when the
forward-scattered signal was reflected (Fig. 3D). The TPE autofluorescence that
was detected by in the spectral range 600–750nm did not exhibit significant en-
hancement upon mirror augmentation irrespective of 820nm or 1230nm excitation.
The only increase in autofluorescent signal originated from elastic laminae under
820nm excitation, which was clearly separated from other autofluorescent sources
such as smooth muscle cells containing NAD(P)H and flavin molecules. Interest-
ingly, the same enhancement was not observed under 1230nm excitation.
Note to reader: The broadband dielectric mirror used for reflection of the
forward-scattered MHG signal was placed directly on top of the sample slide.
673 Imaging the microstructure of mice aorta using MHIM
FIG. 3
The mirror-augmented MHG signals from mice aorta. (A) Representative cross-sectional
images of the SHG/TPE signals at 820nm from mice aorta, with and without the reflection of
forward light propagation and (B) the intensity profiles inside the adventitia or across the wall
(white dash line). The intensity of SHG was increased twofold to threefold in the center of
collagen bundle with the dielectric mirror (arrows). However, the augmentation was not
homogenous and presented lower degree of enhancement on location where the SHG was
already strong without mirror. In contrast, the autofluorescence signals show nearly no
difference with and without mirror except the elastin lamina (arrows). (C) Representative
cross-sectional images of the THG/TPE signals at 1230nm from mice aorta and (D) the
intensity profiles across the wall (white dash line). The enhancement of THG was from twofold
to sevenfold (arrows), but no change of TPE signals was shown. A: adventitia (outer layer), M:
media (middle layer), and L: lumen. Scale bars: 20mm.
As such, all visible light between 400 and 750nm was reflected. Since detection of
MHG signals requires high sensitivity, it is essential that ambient illumination in
the imaging area is kept to an absolute minimum.
3.2 SPECTRAL COMPARISON OF MHG AND TWO-PHOTON EXCITEDAUTOFLUORESCENCE HIGHLIGHTS DIFFERENCES IN TISSUEMICROSTRUCTUREBoth the mirror-augmented MHG and TPE signals were compared using the NDD
GaAsP PMTs and the GaAsP QUASAR spectral detector in the same spectral bands.
First, the SHG signal was captured successfully by both the external NDD and the
internal descanned spectral detector and showed, as expected, a highly correlated
pattern. However, the optical loss that the emission suffered on the descanned light
path to the spectral detector was sufficient to preclude detection of the expected, al-
beit much weaker THG signal at 1230nm excitation (Fig. 4A and B). The images
captured at 820nm excitation using an NDD coupled with a red band-pass filter
(580–670nm) and those captured by the descanned spectral detector are quite sim-
ilar. The spatial patterns of lamellar structures are consistent with those previously
reported (Boulesteix et al., 2006; Zoumi et al., 2004). The observed autofluorescence
potentially being contributed by desmosine and isodesmosine crosslinks between
polypeptide chains, which can be visualized at ease with an excitation wavelength
in the range 800–860nm. However, the autofluorescent signature was not limited
to lamellar structures, in fact 820nm excitation also elicited autofluorescence from
the interlaminar spaces as well as the adventitia, which are highlighted in the green
and red regions of the spectra using the spectral detector. However, using 1230nm
excitation showed a stark difference in the autofluorescence spectrum with the ma-
jority of the previously observed green signal under 820nm being completely absent.
This phenomenon served to significantly increase the contrast of the elastic laminae.
Two-photon excited autofluorescence spectra from three different aortic micro-
structures, the adventitia, the elastic laminae, and the interlaminar space were ana-
lyzed (Fig. 4C). Regions of interest defined in the spectral data stacks obtained at
both 820 and 1230nm excitation, revealed that the elastic laminae exhibit a peak
in their autofluroescent spectra in the range 660–670nm that was absent from regions
defined in both the adventitia and the interlaminar spaces. This supports the hypoth-
esis that lamellar structures preferentially contribute to the autofluroescence signal in
the far-red regions of the electromagnetic spectrum. At 820nm excitation, strong
autofluorescence signal was observed in the spectral range 530–580nm in the re-
gions of interest highlighting the adventitia. This signal most likely is contributed
from either smooth muscle cells containing NAD(P)H and/or flavin compounds.
These species may also contribute the moderate autofluorescence present in the elas-
tic laminae and the interlaminar spaces. These signals, however, were exhibited at
1230nm excitation, which served to significantly increase the relative contrast of la-
mellar structures.
693 Imaging the microstructure of mice aorta using MHIM
FIG. 4
See figure legend on next page.
70 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
3.3 VISUALIZING THE MICROSTRUCTURES OF MOUSE AORTA USINGNONLINEAR OPTICAL MICROSCOPYBecause both MHG and TPE happen inside the same spatial excitation volume,
each technique is inherently able to optically section to generate 3D data. By ac-
quiring the MHG/TPE signals under 820nm excitation in 3D, it is possible to vi-
sualize both the ribbon-like collagen bands and lamellar elastin structures, thus
providing detailed information on the microstructure of the aortic wall (Fig. 5A,
top). Images acquired at different axial depths within the sample allow a 3D recon-
struction to be built (Fig. 5B, left), which highlights the distinct structures of elastin
and collagen present in the aortic wall. The same principle can be applied to the
THG/TPE signals under 1230nm excitation (Fig. 5A, bottom, and Fig. 5B, right).
In contrast to the specificity of SHG signals that arise from fibrillar collagen, the
source of THG in the mice aorta is not well known. Potential sources are smooth
muscle, endothelial cells, fibroblast-like cells, and extracellular matrix compo-
nents. Because of the inherent complexity of the microstructure present, heteroge-
neous THG signals were shown in both adventitia and media, and especially in the
interlaminar spaces (Fig. 6A). In the lumen, the resonant enhancement of the THG
signal due to the presence of hemoglobin was observed and allowed the observation
of the morphological shape of blood cells, consistent with previous reports (Chang,
Yu, & Sun, 2010; Rehberg et al., 2011). It is clear though that further investigation
will be required in order to be able to identify the nonspecific sources of THG sig-
nals. In contrast, the TPE signal under 1230nm excitation shows both high contrast
and high specificity which aids in subsequent reconstruction and segmentation.
This added benefit could have potential benefits for investigating the microstruc-
ture of elastic laminae. For example, the surface of elastic laminae contains fenes-
trae, which are small holes on the order 1–10mm in diameter, which also vary
between different age and species. Their exact purpose and physiological function
Mirror-augmented multiharmonic and TPE autofluorescent signals. (A) Representative
images of second- and third-harmonic generation (top, filter: 380–430nm), TPE
autofluorescence (middle, 580–670nm), and merged image (bottom) under the excitation of
820 and 1230nm wavelength, captured by the nondescanned detector (NDD). Filters:
380–430 and 580–670nm. (B) The images of averaged intensity collected by the
QUASAR spectral detector at spectra between 380–430, 499–552, 579–668nm and merge
(from top to bottom). The intensity of image at 380–430nm was doubled for presentation.
(C) Emission spectra of autofluorescent in different region of interests (ROIs) in the adventitia,
elastic laminae, and interlaminar spaces: Left, the representative image with five selected
ROIs (1.75mm �1.75mm) in each type of structures. Right, average intensity of emission
signals under the excitation of 820 and 1230nm wavelength. Resolution at emission
wavelength: 8.9nm. Range of emission wavelength: 430–690nm. Note that emission signals
between 380 and 430nm were collected by a separated PMT and were not analyzed in
this figure. Scale bars: 20mm.
713 Imaging the microstructure of mice aorta using MHIM
is not well known, but has been linked in facilitating nutrient transport between
cells in the interlaminar spaces (Wong & Langille, 1996), and could potentially
be aberrant in diseases like Marfan syndrome (Lopez-Guimet, Andilla, Loza-
Alvarez, & Egea, 2017). In our experimental system, we successfully observed
FIG. 5
3D multiharmonic microscopy with mirror augmentation. (A) Representative images of SHG/
TPE and THG/TPE under designated excitation wavelength. The maximum Z-projection
over Top: 10mm deep 3D stack at 820nm excitation wavelength; Bottom, 6.25mm deep
3D stack at 1230nm excitation wavelength. Scale bars: 20mm. (B) The 3D rendering of the
aortic wall from SHG/TPE and THG/TPE images.
72 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
fenestrae with high contrast using the autofluorescence of elastin under 1230nm
excitation (Fig. 6B). Taking into consideration the additional penetration of
1230nm vs 820nm light into scattering milieu, this could be useful for studying
microstructural changes that arise when the elastic laminae are disrupted.
FIG. 6
Structural details delineated by MHIM and TPE. (A) Representative THG/TPE images from
adventitia, media, and lumen. Scale bars: 10mm. (B) Fenestrae (arrows) were identified
under the 3D MHIM in the stacks of four images acquired every 1.25mm. Scale bars: 10mm.
733 Imaging the microstructure of mice aorta using MHIM
4 CONCLUSIONSIn this method, we have presented an integrated approach to allow investigators to
differentiate the structural components of mouse aortic wall by combining two NLO
techniques: MHIM and TPE microscopy. Under 820nm excitation, SHG and TPE
signals that primarily correspond to collagen and elastin can be spectrally separated
and exhibit identical patterns in the adventitia, the elastic laminae, and the interla-
minar spaces. These results are consistent with previously published work that sug-
gest that the SHG/TPE imaging can be used as a replacement for conventional
histological methods for identifying microstructures in the mouse aorta
(Boulesteix et al., 2006; Zoumi et al., 2004). The tunable pulsed femtosecond laser
is a powerful tool for us to examine the THG/TPE imaging paradigm on the same
samples. We have demonstrated that under 1230nm excitation, the elastic laminae
exhibit higher contrast in TPE imaging. This could have an added benefit as images
with enhanced signal-to-noise ratios make the postprocessing and analysis of fine
structures such as fenestrae more feasible. In addition, our approach is not limited
to blood vessels, rather it may facilitate the study of other collagen- or elastin-rich
tissues such as elastic cartilage, skin, or the bladder.
Furthermore, we achieved THG imaging with the same infrared laser wavelength
(i.e., 1230nm) on cross-sections of mouse aorta by reflecting the forwardly propa-
gating THG signal to the backward detector. To the best of our knowledge, this is the
first report showing the acquisition of THG signals obtained frommouse aorta tissue.
However, further investigation is required to successfully identify the source of THG
signals since refractive index mismatching can significantly occur inside the micro-
environment of blood vessels. THG signals most likely originate from a combination
of smooth muscle and/or additional extracellular matrix components that may alter
under a specific pharmacological treatment or genetic modification. Therefore, com-
putational approaches may be necessary to evaluate the spatial pattern of the THG
signals, which have been used previously to characterize the microarchitecture of
collagen fibrils (Bayan et al., 2009; Boudaoud et al., 2014; Osman et al., 2013). Fi-
nally, by integration of SHG/TPE and THG/TPE imaging data with other labeling
techniques, such as immunofluorescence staining or fluorescent fusion proteins, it
will be possible to provide a more complete morphological picture and help in un-
derstanding more about the physiological functions and pathological progress of dis-
ease in the mouse aorta.
ACKNOWLEDGMENTSWe would like to thank Bob Mecham at Washington University for the provision of frozen
sections of mouse aorta for imaging studies as well as fruitful discussion and helpful comments
on the manuscript. This work was supported by grants from the Children’s Discovery Institute
of Washington University and St. Louis Children’s Hospital (CDI-CORE-2015-505) and the
Foundation for Barnes Jewish Hospital (3770) to J.A.J.F.
74 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
REFERENCESAptel, F., Olivier, N., Deniset-Besseau, A., Legeais, J.-M., Plamann, K., Schanne-Klein, M.-C.,
et al. (2010). Multimodal nonlinear imaging of the human cornea. Investigative Ophthal-mology & Visual Science, 51(5), 2459–2465. https://doi.org/10.1167/iovs.09-4586.
Barad, Y., Eisenberg, H., & Horowitz, M. (1997). Nonlinear scanning laser microscopy by
third harmonic generation. Applied Physics Letters, 70, 922. https://doi.org/10.1063/1.118442.
Bayan, C., Levitt, J. M., Miller, E., Kaplan, D., & Georgakoudi, I. (2009). Fully automated,
quantitative, noninvasive assessment of collagen fiber content and organization in
thick collagen gels. Journal of Applied Physics:105(10). 102042. https://doi.org/
10.1063/1.3116626.
Boudaoud, A., Burian, A., Borowska-Wykręt, D., Uyttewaal, M., Wrzalik, R.,
Kwiatkowska, D., et al. (2014). FibrilTool, an ImageJ plug-in to quantify fibrillar struc-
tures in rawmicroscopy images.Nature Protocols, 9(2), 457–463. https://doi.org/10.1038/nprot.2014.024.
Boulesteix, T., Pena, A. M., Pages, N., Godeau, G., Sauviat, M. P., Beaurepaire, E., et al.
(2006). Micrometer scale ex vivo multiphoton imaging of unstained arterial wall structure.
Cytometry, Part A: The Journal of the International Society for Analytical Cytology, 69(1),20–26. https://doi.org/10.1002/cyto.a.20196.
Brakenhoff, G. J., Squier, J. A., Wilson, K. R., & M€uller, M. (1998). Third harmonic gener-
ation microscopy. Optics Express, 3(9), 315–324. https://doi.org/10.1364/OE.3.000315.Campagnola, P. J., & Loew, L. M. (2003). Second-harmonic imaging microscopy for visual-
izing biomolecular arrays in cells, tissues and organisms. Nature Biotechnology, 21(11),1356–1360. https://doi.org/10.1038/nbt894.
Campagnola, P. J., Millard, A. C., Terasaki, M., Hoppe, P. E., Malone, C. J., & Mohler, W. A.
(2002). Three-dimensional high-resolution second-harmonic generation imaging of en-
dogenous structural proteins in biological tissues. Biophysical Journal, 82(1), 493–508.https://doi.org/10.1016/S0006-3495(02)75414-3.
Campagnola, P. J., Wei, M. D., Lewis, A., & Loew, L. M. (1999). High-resolution nonlinear
optical imaging of live cells by second harmonic generation. Biophysical Journal, 77(6),3341–3349. https://doi.org/10.1016/S0006-3495(99)77165-1.
Chang, C.-F., Yu, C.-H., & Sun, C.-K. (2010). Multi-photon resonance enhancement of third
harmonic generation in human oxyhemoglobin and deoxyhemoglobin. Journal of Biopho-tonics, 3(10�11), 678–685. https://doi.org/10.1002/jbio.201000045.
Chen, S.-Y., Hsu, C.-Y. S., & Sun, C.-K. (2008). Epi-third and second harmonic generation
microscopic imaging of abnormal enamel. Optics Express, 16(15), 11670–11679.Chen, X., Nadiarynkh, O., Plotnikov, S., & Campagnola, P. J. (2012). Second harmonic
generation microscopy for quantitative analysis of collagen fibrillar structure. NatureProtocols, 7(4), 654–669. https://doi.org/10.1038/nprot.2012.009.
Chen, C.-H., & Sun, C.-K. (2009). Third and second harmonic generation imaging of human
articular cartilage. In A. Periasamy & P. T. C. So (Eds.), SPIE: Vol. 7183. Presented at theSPIE BiOS: Biomedical optics (pp. 71831V–10). https://doi.org/10.1117/12.809968.
Chen, S. Y., Wu, H. Y., & Sun, C. K. (2009). In vivo harmonic generation biopsy of human skin.
Journal of Biomedical Optics, 14, 060505. https://doi.org/10.1117/1.3269676.Cheng, J.-X., & Xie, X. S. (2002). Green’s function formulation for third-harmonic generation
microscopy. Journal of the Optical Society of America B: Optical Physics, 19(7),1604–1610. https://doi.org/10.1364/JOSAB.19.001604.
75References
Chu, S.-W., Chan, M.-C., Tai, S.-P., Keller, S., DenBaars, S. P., & Sun, C.-K. (2005). Simul-
taneous four-photon luminescence, third-harmonic generation, and second-harmonic
generation microscopy of GaN. Optics Letters, 30(18), 2463–2465.D�ebarre, D., Olivier, N., Supatto, W., & Beaurepaire, E. (2014). Mitigating phototoxicity dur-
ing multiphoton microscopy of live drosophila embryos in the 1.0-1.2 mm wavelength
range. PLoS One:9(8). e104250. https://doi.org/10.1371/journal.pone.0104250.Farrar, M. J., Wise, F. W., Fetcho, J. R., & Schaffer, C. B. (2011). In vivo imaging of myelin in
the vertebrate central nervous system using third harmonic generation microscopy. Bio-physical Journal, 100(5), 1362–1371. https://doi.org/10.1016/j.bpj.2011.01.031.
Freund, I., Deutsch, M., & Sprecher, A. (1986). Connective tissue polarity. Optical second-
harmonic microscopy, crossed-beam summation, and small-angle scattering in rat-tail
tendon. Biophysical Journal, 50(4), 693–712. https://doi.org/10.1016/S0006-3495(86)
83510-X.
Friedl, P., Wolf, K., von Andrian, U. H., & Harms, G. (2007). Biological second and third
harmonic generation microscopy. Current Protocols in Cell Biology:chapter 4. Unit4.15–4.15.21. https://doi.org/10.1002/0471143030.cb0415s34.
Gannaway, J., & Sheppard, C. J. R. (1978). Second-harmonic imaging in the scanning optical
microscope. Optical and Quantum Electronics, 10(5), 435–439. https://doi.org/10.1007/BF00620308.
Hellwarth, R., & Christensen, P. (1974). Nonlinear optical microscopic examination of struc-
ture in polycrystalline ZnSe. Optics Communications, 12(3), 318–322. https://doi.org/10.1016/0030-4018(74)90024-8.
Jay, L., Bourget, J.-M., Goyer, B., Singh, K., Brunette, I., Ozaki, T., et al. (2015). Character-
ization of tissue-engineered posterior corneas using second- and third-harmonic genera-
tion microscopy. PLoS One, 10(4), e0125564–13. https://doi.org/10.1371/journal.
pone.0125564.
Jay, L., Brocas, A., Singh, K., Kieffer, J.-C., Brunette, I., & Ozaki, T. (2008). Determination of
porcine corneal layers with high spatial resolution by simultaneous second and third har-
monic generation microscopy. Optics Express, 16(21), 16284–16293.Jiao, Y., Bersi, M. R., Li, Q., Mecham, R. P., Humphrey, J. D., & Tellides, G. (2017). Deficient
circumferential growth is the primary determinant of aortic obstruction attributable to par-
tial elastin deficiency. Arteriosclerosis, Thrombosis, and Vascular Biology, 37(5),930–941. https://doi.org/10.1161/ATVBAHA.117.309079.
Kasischke, K. A., & Hyman, B. T. (2003). Uniform polarity microtubule assemblies imaged in
native brain tissue by second-harmonic generation microscopy. Proceedings of theNational Academy of Sciences of the United States of America, 100(12), 7081–7086.https://doi.org/10.1073/pnas.0731953100.
Lim, H., Sharoukhov, D., Kassim, I., Zhang, Y., Salzer, J. L., & Melendez-Vasquez, C. V.
(2014). Label-free imaging of Schwann cell myelination by third harmonic generation mi-
croscopy. Proceedings of the National Academy of Sciences, 111(50), 18025–18030.https://doi.org/10.1073/pnas.1417820111.
Linkert, M., Rueden, C. T., Allan, C., Burel, J.-M., Moore, W., Patterson, A., et al. (2010).
Metadata matters: Access to image data in the real world. The Journal of Cell Biology,189(5), 777–782. https://doi.org/10.1083/jcb.201004104.
Lopez-Guimet, J., Andilla, J., Loza-Alvarez, P., & Egea, G. (2017). High-resolution morpho-
logical approach to analyse elastic laminae injuries of the ascending aorta in a murine
76 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins
model of Marfan syndrome. Scientific Reports, 7(1), 1505. https://doi.org/10.1038/
s41598-017-01620-8.
Mahou, P., Olivier, N., Labroille, G., Duloquin, L., Sintes, J.-M., Peyri�eras, N., et al. (2011).Combined third-harmonic generation and four-wave mixing microscopy of tissues and
embryos. Biomedical Optics Express, 2(10), 2837–2849. https://doi.org/10.1364/
BOE.2.002837.
Marangoni, M., Lobino, M., & Ramponi, R. (2006). Simultaneously phase-matched second-
and third-harmonic generation from 1.55 microm radiation in annealed proton-exchanged
periodically poled lithium niobate waveguides. Optics Letters, 31(18), 2707–2709.Mohler, W., Millard, A. C., & Campagnola, P. J. (2003). Second harmonic generation imaging
of endogenous structural proteins. Methods, 29(1), 97–109.Osman, O. S., Selway, J. L., Harikumar, P. E., Stocker, C. J., Wargent, E. T.,
Cawthorne, M. A., et al. (2013). A novel method to assess collagen architecture in skin.
BMC Bioinformatics, 14(1), 260. https://doi.org/10.1186/1471-2105-14-260.Pfeffer, C. P., Olsen, B. R., Ganikhanov, F., & L�egar�e, F. (2011). Imaging skeletal muscle
using second harmonic generation and coherent anti-stokes Raman scattering microscopy.
Biomedical Optics Express, 2(5), 1366–1376. https://doi.org/10.1364/BOE.2.001366.Rehberg, M., Krombach, F., Pohl, U., & Dietzel, S. (2010). Signal improvement in multipho-
ton microscopy by reflection with simple mirrors near the sample. Journal of BiomedicalOptics, 15(2). 026017. https://doi.org/10.1117/1.3374337.
Rehberg, M., Krombach, F., Pohl, U., & Dietzel, S. (2011). Label-free 3D visualization of cel-
lular and tissue structures in intact muscle with second and third harmonic generation mi-
croscopy. PloS One:6(11). e28237. https://doi.org/10.1371/journal.pone.0028237.Royer, L. A., Weigert, M., G€unther, U., Maghelli, N., Jug, F., Sbalzarini, I. F., et al. (2015).
ClearVolume: Open-source live 3D visualization for light-sheet microscopy. NatureMethods, 12(6), 480–481. https://doi.org/10.1038/nmeth.3372.
Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., et al.
(2012). Fiji: An open-source platform for biological-image analysis. Nature Methods,9(7), 676–682. https://doi.org/10.1038/nmeth.2019.
Schindelin, J., Rueden, C. T., Hiner, M. C., & Eliceiri, K. W. (2015). The ImageJ ecosystem:
An open platform for biomedical image analysis. Molecular Reproduction and Develop-ment, 82(7–8), 518–529. https://doi.org/10.1002/mrd.22489.
Tai, S.-P., Kung, C.-T., Yu, H.-C., Huang, H.-J., & Chan, Y.-F. (2007). In vivo and ex vivo
imaging of intra-tissue elastic fibers using third-harmonic-generation microscopy. OpticsExpress, 15(18), 11167–11177.
Tsai, C.-K., Wang, T.-D., Lin, J.-W., Hsu, R.-B., Guo, L.-Z., Chen, S.-T., et al. (2013). Virtual
optical biopsy of human adipocytes with third harmonic generation microscopy. Biomed-ical Optics Express, 4(1), 178–186. https://doi.org/10.1364/BOE.4.000178.
Weigelin, B., Bakker, G.-J., & Friedl, P. (2012). Intravital third harmonic generation micros-
copy of collective melanoma cell invasion. IntraVital, 1(1), 32–43. https://doi.org/
10.4161/intv.21223.
Weigelin, B., Bakker, G. J., & Friedl, P. (2016). Third harmonic generation microscopy of
cells and tissue organization. Journal of Cell Science, 129(2), 245–255. https://doi.org/10.1242/jcs.152272.
Wong, L. C., & Langille, B. L. (1996). Developmental remodeling of the internal elastic lam-
ina of rabbit arteries: Effect of blood flow. Circulation Research, 78(5), 799–805.
77References
Wu, P.-C., Hsieh, T.-Y., Tsai, Z.-U., & Liu, T.-M. (2015). In vivo quantification of the struc-
tural changes of collagens in a melanoma microenvironment with second and third har-
monic generation microscopy. Scientific Reports, 5, 8879. https://doi.org/10.1038/
srep08879.
Zoumi, A., Lu, X., Kassab, G. S., & Tromberg, B. J. (2004). Imaging coronary artery micro-
structure using second-harmonic and two-photon fluorescence microscopy. BiophysicalJournal, 87(4), 2778–2786. https://doi.org/10.1529/biophysj.104.042887.
78 CHAPTER 3 Nonlinear optical imaging of extracellular matrix proteins