Multicolor fluorescence imaging has been widely used by neuroscientists to simultaneously observe different neuropathological features of the brain. However, these optical modalities rely on exogenous labeling. Here, we demonstrate, for the first time, a label-free additive-color multi-harmonic generation microscopy to elucidate, concurrently with different hues, Alzheimer’s disease (AD) neuropathological hallmarks: amyloid β (Aβ) plaques and neurofibrillary tangles (NFT). By treating third harmonic generation (THG) and second harmonic generation (SHG) as two primary colors, our study can simultaneously label-free differentiate AD hallmarks by providing different additive colors between Aβ plaques, NFT, and neuronal axons, with weaker THG presentation from NFT in most places of the brain. Interestingly our pixel-based quantification and Pearson’s correlation results further corroborated these findings. Our proposed label-free technique fulfills the unmet challenge in the clinical histopathology for stain-free slide-free differential visualization of neurodegenerative disease pathologies, with a sub-femtoliter resolution in a single image field-of-view.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
In our nervous system, neurons construct complex networks, forming connectomes with thousands of synapses possibly overlapping in a tiny volume . Analyzing and understanding such complex systems in time and space need sophisticated visualization tools which can provide a display distinguishing individual neuropathological components. And perhaps the most useful visual modality for such purposes is color, considering our vision can process multivariate information present in a complex visual field.
In this context, multicolor fluorescence imaging using multiple probes has been developed and provides an important capability to fluorescence microscopy for in vivo optical imaging [2,3]. Among the techniques of exogenous fluorescence labeling of neurological structures, the brianbow strategy has been widely utilized for genetic cell-labeling with primarily red, green and blue fluorescent proteins (FPs) . This method is based on the fact that primary colors red, green, and blue (RGB) can combine to generate hundreds of different hues. Brainbow can achieve such effects by expressing different ratios of FPs within cells. The color combinations are unique in a category of cells, and can be used as cellular identification tags under a light microscope. Although over the years, brainbow technologies have found firm places in the genetic toolbox of neuroscientists, the major concern regarding its utility in clinical pathology practice, to study neurodegenerative disorders like Alzheimer’s disease (AD), involves the use of exogenous FPs; along with color imbalance and discrimination .
Moreover, in clinical practice, the gold standard for definitive confirmation and diagnosis of AD comes from the histopathology and/or immunohistochemical staining procedures of AD brain tissues to reveal its neuropathological hallmarks . However, these protocols require long fixation time, embedding and staining procedures which make the sample analysis a time-consuming process. Considering such a scenario, there is an urgent need of a new technique which can provide diagnostically relevant information, quickly and reliably through a visual display of the AD disease hallmarks in a label-free slide-free approach ex vivo and in particular in vivo.
Pathologically, AD is associated with the aggregation of insoluble forms of amyloid-β (Aβ) in plaques (Aβ plaques) in extracellular space and microtubule protein tau in neurofibrillary tangles (NFT) in neurons . In the development of label-free tools for AD pathology, autofluorescence of the diseased brain tissues was evaluated; and had shown that autofluorescence can detect senile plaques and NFT in the brain tissues from human subjects [7–9]. Both the senile plaques and NFT generate blue emissions (plaques at >430 nm; while NFT at ∼ 460 nm) when excited with ultraviolet light , and hence this limits the simultaneous differentiation of these two features in brain tissues. In addition, these studies were also limited to wide-field, or confocal microscopy on superficial areas, or thinly sliced sections [7–9]. More recently, hyperspectral Raman imaging was used for the identification of neuritic plaques and NFT along with water, lipids, and proteins . However, this technique is severely limited by its low spatial resolution, image acquisition speeds, and most importantly it did not provide a way to distinguish distinctly the plaques and NFTs from other lipid or protein structures. On the other hand, coherent anti-stokes Raman (CARS) imaging was only able to see the lipids associated the plaques; and showed no evidence regarding NFT . In general, these techniques depend on the differences in the vibrational spectra; and are time consuming due to their requirements to compare the measured spectra with a library of reference spectra. Recently, polarization sensitive optical coherence tomography (OCT) has also been successful in identifying only plaques , with very low spatial resolutions and as such was unable to provide information regarding axonal networks. No reported work has been published regarding OCT in identifying NFT. Most significantly, all these works have failed to provide a visual display, like that of the brianbow strategy, to discriminate the different neuropathological hallmarks of AD simultaneously in a single image field-of-view with ultrahigh resolution.
It is widely known that brain tissues contain various endogenous proteins/structures that can generate optical higher harmonic generation signals . In this study, we have explored the potential of concurrently using the label-free nonlinear optical microscopy techniques: second and third harmonic generation (SHG and THG) microscopy to provide a dual contrast additive multi-color mechanism to display the various structural features of AD brain tissues simultaneously with different hues. By treating third harmonic generation (THG) and second harmonic generation (SHG) as two primary colors, our study indicates different additive colors between neuronal axons, Aβ plaques, and NFT. Our experimental observation supports our additive-color multi-harmonics approach as a label-free slide-free multicolor imaging platform for differential visualization of AD hallmark structures based on an excitation wavelength in the optical window of brain.
1.1 Principle of additive-color multi-harmonic generation microscopy
Optical multi-harmonic generation, such as SHG and THG, is a nonlinear coherent polarization process induced by inorganic or organic structures with specific physical properties, molecular arrangements, and order [10,14–16]. In the SHG process, two incident photons are converted into one photon with twice the energy and half the wavelength; while in THG, three incident photons are converted into a photon of thrice the energy and one third of the wavelength (Fig. 1(a)) . No excess energy is deposited in the organic structures during the harmonic generation process, while no bleaching and phototoxicity effects are expected. The schematic of the imaging setup to generate and collect simultaneously the SHG and THG signals is shown in Fig. 1(b). A femtosecond Cr:forsterite laser with a central wavelength of 1262 nm was used as an excitation source for SHG/THG.
SHG relies on the second order non-linear susceptibility χ(2) of the tissue and arises from well-organized non-centrosymmetric molecules such as collagen [16,17]. Structural studies of Aβ plaques and NFT shows their beta sheet structures (Fig. 1(c)) [18,19], and as such, can provide SHG signals. Kwan et al. has already proved this fact by demonstrating SHG emissions from Aβ plaques in relatively thick, native brain tissues from a AD transgenic mice model . Moreover, even though it is not from tissues, SHG emissions were further observed from NFT in a neuronal cell-culture study . On the other hand, THG occurs at structural interfaces such as the spatial variation of third order nonlinear susceptibility χ(3), or the change in the refractive indices [14,15], while refractive indices and χ(3) directly reflect the nature of molecules regardless of their higher order arrangement. In addition to these two major factors, the THG signal intensity also varies significantly with size and organization of the signal-originating structures at the micrometer scale; and hence depends on the focal volume of the imaging system. In addition to all, molecules such as lipids have very high value of χ(3) and consequently generate strong THG signals under 1200 nm excitation . It has been reported that the typical diameter of axons and dendrites ranges in between 0.3 to 2 µm, and the axons structures are composed of lipids (myelin sheath) (Fig. 1(d)); hence, these can produce strong THG signals. On the other hand, neuronal soma is generally of diameter ∼ 10-25 µm and thickness about ∼10-14 µm. In addition, somatal area is mostly composed a large nucleus; and hence can be considered homogeneous with uniform susceptibility. Considering the confocal parameter of our microscope [23,24] and the physical dimension and properties of soma, THG signal will be extremely weak, or none when the laser beam will be focused inside them . Due to this, the neuronal cell bodies will appear dark in the THG images [25–27]. Considering all such conditions and the varying size and organization of Aβ plaques and NFT, we would find that different signal strengths of THG can arise from these two AD pathological structures. Plaques are usually associated lipids  and in the advanced AD mice model, large plaque structures (5-50 µm) with inhomogeneous densely aggregated Aβ structural organization can be found everywhere in the brain, and hence, being expected to provide stronger THG signals . NFT in most cases was found as an intercellular entity with a much smaller size and loosely bound aggregated structures of tau proteins; and is expected to produce weaker THG signals (Fig. 1(d)).
Taking into account of the above facts, here we propose and adopt the additive-color principle to the label-free multi-harmonic generation microscopy to display the various AD pathologies in mice brain. Providing SHG and THG with different primary colors, we can obtain an image with different hues showing different structures of the AD brain tissues (Fig. 1(e)), reflecting different combinations of the molecular nature and structural arrangements. In this study thus combined THG-green and SHG-red pseudo-colors generate a color map ranging from green to different hues of yellow to red as shown in Fig. 1(e). As a result, without labeling, we successfully generate a visual display where, green color shows axons and dendrites, yellow color shows Aβ plaques, and mostly orange-red color shows NFT (Fig. 1(e)).
2. Materials and methods
In this study, we used one wild-type 4 months old C57BL/6 mouse and 4 (one 8 months and three 13 months old) triple transgenic (3xTg) AD mice. The results of this study primarily focus on demonstrating the capability of the present technique to differentiate AD pathologies through label-free microscopic images. The results do not depend on statistical methods to determine the sample size. Here, the animals were maintained following the guidelines approved in the Codes for Experimental Use of Animals of the Council of Agriculture of Taiwan, based on the Animal Protection Law of Taiwan. All the mice were kept on a 12-hour light/12-hour dark cycle and fed ad libitum. The food and water were always available for them. All the experimental procedures were performed after the approval from The Institutional Animal Care and Use Committee (IACUC) of National Taiwan University, Taipei, Taiwan (IACUC Approval number-NTU106-EL-00203).
The C57BL/6 mouse was purchased from Charles River (Sulzfeld, Germany); while the 3xTg mice were from the Jackson Laboratory (Maine, USA). All the animals were maintained in the local animal facilities at the Department of Life Science (for C57BL/6), National Taiwan University, Taipei and National Taiwan University Hospital Animal Center (for 3xTg), Taipei.
2.2 Perfusion and brain tissue slicing
Before the perfusion, the mice were anesthetized with a mixture of 10 mg/kg xylazine (Bayer Vital, Germany) and 100 mg/kg ketamine (Bela-pharm GmbH & CO. KG, Germany). Using a peristaltic pump, the mice were subsequently perfused transcardially with 0.9% normal saline, followed by 4% paraformaldehyde in 0.1M phosphate-buffered saline (PBS). All the solutions were prepared fresh and kept in ice before perfusion. The brains were removed from the skull and further post-fixed in 4% paraformaldehyde overnight at 4°C. After this, the brains were kept in PBS containing 0.01% sodium azide at 4°C.
Prior to the brain tissue slicing, it was kept submerged in 20% sucrose solution overnight at 4°C. Subsequently, the brain was sectioned coronally using a freezing microtome at a thickness of ∼ 50 µm. These 50µm-thick tissue slices were first imaged by THG and SHG microscopy; and then, were stained following immunohistochemical procedures for comparison with harmonic generation microscopy images.
2.3 Immunohistochemical (IHC) staining of the brain tissues
The immunohistochemical staining of the brain tissue sections were performed following standard protocols. In brief, the tissue sections were first washed in 1x PBS containing 0.1% TritonX-100 (PBST) three times each for 10 mins. To avoid unspecific binding of the primary antibody, the tissue sections were incubated in blocking buffer (PBST and 5% normal goat serum; NGS) for 90 mins on a shaker at room temperature. After blocking, the sections were incubated in primary antibody at 1:1000 in PBST containing 1% NGS overnight at 4°C. Following day, the sections were washed three times (each for 10 mins) in PBST at room temperature, and were, subsequently, incubated in secondary antibody (Alexa Fluor 594-conjugated goat anti-mouse antibody, Jackson ImmunoResearch, PA, USA) at 1:1000 prepared in 1%NGS/PBST for 120 mins at room temperature. After washing the sections as before, the brain tissue sections were mounted on poly-l-lysine coated glass coverslips and further used for fluorescence imaging under Leica LSM SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany). The primary antibodies used for staining Aβ-plaques, and NFT were anti-β-amyloid produced in mouse (BioLegend, San Diego, USA), and monoclonal anti-PHF1 antibody (Sigma-Aldrich, St. Louis, USA), also produced in mouse, respectively.
2.4 Experimental set up of multi-harmonic generation microscopy
Figure 1(b) shows the experimental set up of the multi-harmonic generation microscopy. The excitation source of our system was a home-built Cr:forsterite laser, capable of producing femtosecond pulses (38 fs) with a repetition rate of 105 MHz at a central wavelength of 1262 nm (bandwidth: 91 nm) . All the experiments were performed on an inverted Olympus microscope IX71 (Olympus, Tokyo, Japan). The dispersion of the optical components for lengthening the laser temporal pulsewidth in tissues was compensated by a pair of double chirp-mirrors [24,29]. For our system, a galvo-resonant scanning head (MPM-2PKIT, Thorlabs) was used for the two-dimensional (2D) scanning of the sample. A scan lens and tube lens inside IX71 were used to fill the back aperture of a microscope objective. A 40x, 1.15 numerical aperture (NA) water-immersion objective (UAPON 40XW340, Tokyo, Olympus) focused the scanning beam onto the sample. For z scanning, the objective was attached to an electronically controllable stage (TSDM40-15X, SIGMA KOKI, Japan). A neutral density (ND) filter was used to control the laser power at the sample focal plane; and an average power of 20-30 mW was maintained to image the samples. The higher harmonic generation signals (SHG and THG) were epi-collected by the same objective and focused on photomultiplier tubes (PMTs) after reflected by a dichroic beamsplitter (DM1) (FF705-Di01, Semrock, New York, USA). Back-reflection laser lights were blocked by a color filter (CG-KG-5-50, CVI Optics, Albuquerque, USA). The SHG and THG signals were separated by another dichroic beamsplitter (DM2) (FF458-Di02-25 × 36, Semrock, New York, USA), and further filtered by bandpass filters (F2), FF01-417/60 for THG (Semrock, New York, USA) and FF01-618/50 for SHG (F1) (Semrock, New York, USA). Commercially available ThorImageLS GUI (ThorImageLS 1.3, Thorlabs) was used for signal recording and real-time display. All the images were collected with 1024×1024 pixels and 10 image frames were accumulated as one image for better image quality. Based on our recent measurement of the system resolution, the 3D spatial resolution of THG and SHG will be 0.31 and 0.57 femtoliter respectively .
2.5 Data analysis
Free image analysis software, FIJI Image J, was used for display purpose and performing Otsu’s thresholding operation. Moreover, standard defined functions of MATLAB 2016B were used for Pearson’s correlation calculations and pixel-by-pixel analysis. In image acquisitions, and analysis, we only excluded data whose quality was compromised by identifiable reasons, such as flawed sample preparations, and poor signal-to-noise ratio. Moreover, the gain (700 V) of the PMTs for the acquisitions of the SHG and THG images was kept fixed for all the imaging sessions. Following this, the acquired SHG and THG images were added without any further scaling/normalization. The only signal processing done is that the background PMT noise was subtracted from each image before addition of the SHG and THG images. To check the normal distribution of our data set, Chi-squared goodness-of-fit test was performed with level of significance 0.05.
3. Results and discussion
3.1 Experimental proof-of-principle
In this study, 3xTg AD mice model was used to demonstrate our proposed principle of additive-color multi-harmonic generation microscopy in discerning the neuropathology of AD (Fig. 2). We performed our SHG and THG imaging sessions within 24 hours of the sacrifice of the mice to minimize the effect (maintaining as fresh condition) of fixative agents in the structural loss of the samples for longer storage times. We used brain tissues from a 13 months old 3xTg AD mouse as our sample. We typically repeat the imaging sessions independently with identical conditions for brain slices from three 13-months old 3xTg mice to ensure the reproducibility of the results. The details of the sample preparation can be seen in the “Materials and methods” section.
In general, the SHG and THG images were recorded as gray labeled images (Figs. 2(a) and 2(b)). However to distinguish them with colors for better visualization, we can provide pseudo-primary-colors to these images. As a result, in our case, the SHG images are shown in red (Fig. 2(c)), while the THG images are shown in green (Fig. 2(d)). When these two colors overlap additively, it appears as yellow; or with different hues of yellow. Figure 2(e) shows the overlapped image of Figs. 2(c) and 2(d); displaying different hues of colors with green for the axons; yellow for the big-clump-like structures, and intraneuronal red structures. This additive-color multi-harmonic generation image represents a 240µmx240µm en face optically-sectioned area of the cortex of the 13 months old 3xTG mice.
To further observe the small red-specks-like structures in Fig. 2(e), we also imaged other areas of the brain cortex slice from a 13 months old 3xTG mouse. Figures 2(h) and 2(i) shows the gray labeled images of SHG and THG respectively. The corresponding pseudo-colored SHG and THG images are shown in Figs. 2(j) and 2(k); and their merged additive image in Fig. 2(l). From Fig. 2(l), we can easily see that the intraneuronal structures appear mostly as red; with some places at different hues of orange.
To study the contrast origin of SHG and THG related to the bright-yellow colored big clump extraneuronal structures (as indicated by arrows in Fig. 2(f)) and the intraneuronal red colored structures (as indicated by arrows in Fig. 2(m)), the multi-harmonic generation microscopy (HGM)-imaged AD brain tissues were stained with anti-β-amyloid to identify the plaques while the antibody PHF-1 was used for NFT. As both the antibodies had the same host in our case, we stained two different HGM-imaged tissue sections, corresponding to Figs. 2(e) and 2(l), respectively for plaque and NFT to avoid non-specific fluorescence imaging. Our immunohistochemical (IHC) stained confocal microscopy images of the same tissue regions are shown in Figs. 2(g) and 2(n) in cyan hot color. Comparing the IHC staining results for Aβ plaques (Fig. 2(g)) with our additive-color multi-harmonic generation image, identified with a dotted box in 2(e) and shown as an enlarged view in Fig. 2(f), we can confirm that the yellow-colored structures with combined THG and SHG contrasts are plaques (indicated by arrows in Figs. 2(f) and 2(g)). In addition, the IHC staining for NFT (Fig. 2(n)) also confirmed the red and orange-red intraneuronal specks as NFT as indicated in the dotted box in Fig. 2(l) with its corresponding enlarged view in Fig. 2(m). The arrows in Fig. 2(m)-(n) indicate the same distribution of NFTs with a high structural similarities in the brain tissue sample before and after the sample processing for staining. All these observations corroborate the hypothesis shown in Fig. 1(e) that additive color multi-harmonic generation microscopy can enable one to visualize concurrently different hallmark structures of AD pathology depending on color. It is noted that the additive-color multi-harmonic generation microscopy images are optically-sectioned images with a vertical thickness on the order of 3 µm . The sectioned thickness in the z-direction is different from the corresponding confocal image. It is hard to control the THG/SHG images to be virtually-sectioned exactly at the same depth and tilt as the IHC confocal images inside the tissue. It is further noted that SHG, THG, and linear IHC microscopy all have different signal intensity dependency on the molecule density. THG and SHG depend not only nonlinearly on the amount of molecules, but also on phase matching conditions. Combining all the above facts, we do not expect to observe 100% match between the HGM and the IHC confocal images of the same area. It is also highly possible that the extensive IHC sample staining procedure, like the shaker, might modify the sample structures and their distributions. However, the comparison study of the same sample successfully assisted the identification of the contrast origins of the yellow and orange structures as plaques and NFTs.
3.2 Quantitative evaluation of the contrast of SHG and THG in AD brain tissues
To quantitatively understand the correlation of THG and SHG signals originating from plaques and NFT, we used the Pearson’s correlation (R) coefficient  as below:
In the pixel-by-pixel analysis, we determine the exact number in percentage of overlapping THG pixels with SHG pixels. As the Pearson correlation study, the pixel analysis was performed after denoising and Otsu’s thresholding. Figure 2(p) summarizes the calculations. Each datum point in Fig. 2(p) represents the overlapping THG pixels with SHG pixels for an Aβ plaque (or, NFT). For analysis, 9 different image planes from 3 different image stacks were considered. For this analysis, the number of areas (n) under this study, representing each datum point in Fig. 2(p), for Aβ plaques and NFT are 20 and 17 respectively. The data is normally distributed. From our results, it can be easily inferred that approximately 91% THG pixels overlapped with SHG pixels for plaques; while nearly 48% of the same were observed for NFT. From this significant difference, we can conclude that for Aβ plaques, additive color for green (THG) and red (SHG) will provide distinct yellow color; while it will be predominantly red or orange for NFT.
3.3 Evolution of AD pathology in different aged 3xTg mice
We have further used our developed technique to observe the evolution of AD pathology in 3xTg mice. For this we have compared the brain tissues from the 8 months old and the other 13 months old 3xTg AD mice. Moreover, we have compared our results with a control C57BL/6 mouse.
Figure 3(a), (b), and (c) shows the additive-color multi-harmonic generation microscopy images of the control C57BL/6 mouse brain slices from cortex, striatum, and hippocampal layer respectively. From the images, it can be observed that there is no contribution of SHG (with red color) from any structures of the brain. The axons and dendrites appear strongly with psudo-green hue of THG; while the cell body of neurons appears as dark shadow. Our study confirms SHG as a background free imaging technique for plaques and NFTs in mouse brain.
The 3xTg AD mice are widely used as a model system for AD related studies as they exhibit plaque and tau pathology which is the characteristics of human form. 12 months, or older 3xTg mice show plaque and tau pathology in cortical and hippocampus regions of the brain . Accordingly, our additive-color multi-harmonic generation microscopy images for the 8 months old 3xTg mouse in cortex (Fig. 3(d)), striatum (Fig. 3(e)), and the hippocampal layer (Fig. 3(f)) shows predominantly THG signals with a green hue. However, very few and small yellow plaques can be observed in the zoomed images (Fig. 3(g)) of the cortex of the 8 months old 3xTg mouse. Moreover, intracellular red signals can also be observed sparsely for this mouse in the hippocampal and cortex regions. This can be considered as the early evolution of NFT and plaques inside the cell body. However, further studies are required to verify this hypothesis.
At last, Figs. 3(j), (k), and (l) show the additive-color multi-harmonic generation microscopy images of cortex, striatum, and the hippocampal layer respectively from the 13 months old 3xTg mice. From these figures, our additive-color multi-harmonic generation microscopy scheme can easily distinguish the Aβ plaques (yellow color), NFTs (red-orange color), and axons (green color). Our additive images of brain slices show intracellular NFTs and plaques more dominantly distributed as compared to the 8-months old mouse, agreeing well with the AD development in 3xTg mice.
To add to the versatility of our proposed methodology, the additive-color multi-harmonic generation microscopy can also identify flame like NFTs in cortex of the 13 months old 3xTg mice. SHG, THG, and additive-color images of this kind of NFTs are shown respectively in Figs. 3(p), (q), and (r). From the additive-color image (Fig. 3(r)), it is further confirmed that THG presentation in NFT is weaker; following which some NFTs also appeared as red in color. So, in a nutshell, we can say that our proposed additive-color multi-harmonic generation microscopy can visually differentiate NFTs, plaques, and axons/dendrites.
In this work, we have treated the well-studied THG and SHG as two primary colors for label-free simultaneous differentiation of AD pathologies: Aβ-plaques and NFT in fixed mouse brain tissues. A femtosecond 1262 nm laser with an extremely wide bandwidth was used as the excitation source. Using this ∼1300nm wavelength provides us with several advantages including deeper penetration depth in brain tissues  and minimal autofluorescence background signals. While efficient SHG/THG excitation might be achieved by the ultrashort pulsewidth, the broad spectral bandwidth of the pulse would experience extensive dispersion in optics, and thus should be managed properly. It is worth to notice that for most optical glass materials, 1262 nm is close to chromatic-dispersion-free wavelength, in sharp contrast to the high dispersion 800 nm wavelength range.
With a wider laser spectrum, a much shorter transform-limited pulse duration $\Delta t$ can be achieved. At the same time, the actual pulse duration after the tissues and optics might experience severe chromatic dispersion, which can be described by group delay dispersion (GDD) $\emptyset $ . Based on a Gaussian pulseshape, the broadened pulse duration tout can be described and estimated as 34] device (FrogScan, MesaPhotonics, Santa Fe, NM, USA), we characterized our laser output after three silver mirrors. As shown in Fig. 4(a), the frequency-dependent phase of the laser pulse can be retrieved, and thus the laser pulsewidth. With the resolved phase, one can then calculate the GDD curve (Fig. 4(c)) with a GDD value of 301 fs2 retrieved at the laser central wavelength of 1262 nm. It can be found that the FROG measured laser pulse duration 38 fs is not transformed limited. The transformed limited pulse duration can be obtained by Fourier-transforming the output spectrum. Twenty-seven femtosecond transform-limited pulse duration can be obtained as shown in Fig. 4(d).
We used double-chirp mirrors (DCM)  to compensate the system GDD. By using FROG, one can measure the component GDD at 1262 nm as follows: one solver mirror + 12.5 fs2; water immersion objective + 350 fs2; scan lens and tube lens system + 600 fs2; one bounce DCM −700 fs2. With 11 silver mirrors, two bounces of DCM, one scan lens tube lens pair, the objective, the dichroic mirror, and a thin water layer before the imaged sample, the total GDD provided by the optical system can be estimated to be −323 fs2. After considering the initial GDD of 263.5 fs2, after deducting the contribution from 3 silver mirrors, the pulse GDD in front of the sample will be −60 fs2. Following Eq. (2), we thus expect a shortened pulsewidth tout of 28 fs, with a pulseshape almost the same as the transform-limited pulse as shown in Fig. 4(d). Considering the brain samples, the major components of brain tissues are water and lipid. Even though we do not find any literature for chromatic dispersion data of brain tissues around 1260 nm, it has been reported that the chromatic dispersion value of water with lipid at 1300 nm is zero . With a group velocity dispersion value of water at 1262 nm as −47 fs2/mm , we thus expect that the GDD due to 50-µm-thick brain tissues at 1260 nm is negligible as compared to our optical components.
In addition to chromatic dispersion, another concern regarding the use of tightly focused femtosecond lasers for bioimaging is the issue of photodamage and altering structural integrity of biological samples. In this study, we have used 1262 nm light for THG/SHG microscopy. This wavelength belongs to the optical window of brain tissues, and suffers least attenuation. Chu et al. has already shown the minimal attenuation constant in porcine brain at the 1260-1300 nm wavelength range . Previous studies involving HGM using femtosecond 1230 nm excitation light for in vivo optical virtual biopsy of human oral mucosa, with 100 mW average laser powers after the objective, showed significant enhancement of THG/SHG signals with much reduced multi-photon tissue autofluorescence . These studies also reported no biological damage of the tissues at such excitation powers. Another study in brain tissues with 1280 nm femtosecond light also showed the benefit of imaging with the effective natural optical filtering of the intrinsic fluorescence background . In addition, by using 1230 nm as excitation wavelength, we have shown in our previous HGM studies that embryos, even after 100 mW 20 hours continuous irradiation (total exposure 7200 J/embryo), maintain their viability [38,39] and DNA expression for normal development with no observable biodamages. Even for buccal tissues, after 150 mW 3-hours continuous irradiation under 1230 nm femtosecond light for THG imaging, histopathological follow-ups reported normal squamous epithelium and sub-epithelial stroma with no evidence of coagulation necrosis . In addition to all, the dominant absorber in tissues at 1260 nm is water (absorption coefficient: 10 cm−1) . With less water content in brain and with reduced average power compared to previous studies [42,43], we expect negligible sample temperature rise in the planned ex vivo studies. In brief, we expect no photo-induced structure modification or autofluorescence background in this study with an average power of 20-30 mW at 1260 nm. Following all these discussions, we can expect that our proposed technique can be further used, without producing any artifacts, for imaging of mice brain tissues.
In clinical practice and neuroscience, the gold standard of visualizing the histopathological features of brain tissues involves staining procedures of fixed tissues, followed by microscopy analysis. The sample preparation procedure is time consuming and the IHC staining cannot be applied for in vivo observation. Label-free optical imaging techniques are thus widely explored for their possible potential to be applied in vivo. In this study, we have successfully demonstrated the capability of additive-color multi-harmonic generation microscopy to distinguish different pathological hallmark features of AD in fixed brain tissues without staining. This label-free technique can be a potential all-optical imaging platform for in vivo or intravital animal studies, as well as ex vivo clinical diagnosis. It is noted that transgenic fluorophores based AD mice model has been widely used for in vivo study of the disease [44–46]. Two-photon excitation at 750 nm was explored to study such kind of in vivo experiments . Our discovery might provide an alternative approach to study the disease progression without the need of fluorophores. Though the present study has been performed in ex vivo fixed brain samples, a previous study on fixed human skin sample  using THG/SHG with the 1230 nm excitation, also from a Cr:forsterite laser, shows the same THG/SHG contrast as the later in vivo imaging studies . Using a light source of 1260 nm, THG can provide us detailed structural information such as soma and axonal networks of the brain tissues with a sub-femtoliter resolution, as lipid structures in axons provide strong THG contrast. In addition, SHG provides background-free information on two most important pathological hallmarks of AD: Aβ-plaques and NFT. SHG augmented with THG further distinguishes these two hallmarks. As a result, additive-color multi-harmonic generation microscopy can provide a visual display of different pathologies of AD with different hues. Moreover, THG and SHG both rely on nonlinear optical excitation phenomenon and hence can provide natural optical sectioning with 3D volumetric images without affecting viability [38,48]. Considering this, our demonstrated approach, as well, can be held with a potential technical value in understanding neuronal networks and AD disease pathologies at millimeter depths  in intact whole mice brain.
Ministry of Science and Technology, Taiwan (MOST 106-2221-E-002 -156 -MY3, MOST 106-2321-B-002-017).
The authors would like to acknowledge Prof. Chung-Tung Yen, Department of Life Sciences, National Taiwan University, Taiwan for generously providing the laboratory space for the sample preparations and constructive suggestions.
A patent (Inventor: C.-K.S., & S.C.) application, describing the idea related to the presenting technique in this manuscript, will be applied.
1. J. W. Lichtman and W. Denk, “The big and the small: challenges of imaging the brain’s circuits,” Science 334(6056), 618–623 (2011). [CrossRef]
2. N. Yamamoto, H. Tsuchiya, and R. M. Hoffman, “Tumor imaging with multicolor fluorescent protein expression,” Int. J. Clin. Oncol. 16(2), 84–91 (2011). [CrossRef]
3. J. Licea-Rodriguez, A. Figueroa-Melendez, K. Falaggis, M. P. Sanchez, M. Riquelme, and I. Rocha-Mendoza, “Multicolor fluorescence microscopy using static light sheets and a single channel detection,” J. Biomed. Opt. 24(01), 1 (2019). [CrossRef]
4. T. A. Weissman and Y. A. Pan, “Brainbow: new resources and emerging applications for multicolor genetic labeling and analysis,” Genetics 199(2), 293–306 (2015). [CrossRef]
5. S. Love, “Neuropathological investigation of dementia: A guide for neurologists,” J. Neurol., Neurosurg. Psychiatry 76(suppl_5), v8–v14 (2005). [CrossRef]
6. A. Burns and S. Iliffe, “Alzheimer’s disease,” BMJ 338, b158 (2009). [CrossRef]
7. J. H. Dowson, “A sensitive method for demonstration of senile plaques in the dementing brain,” Histopathology 5(3), 305–310 (1981). [CrossRef]
8. D. R. Thal, E. Ghebremedhin, C. Haass, and C. Schultz, “UV light-induced autofluorescence of full-length abeta-protein deposits in the human brain,” Clin. Neuropathol. 21(1), 35–40 (2002).
9. M. Diez, “Neuropeptides in hippocampus and cortex in transgenic mice overexpressing V717F beta-amyloid precursor protein-initial observations,” Neuroscience 100(2), 259–286 (2000). [CrossRef]
10. W. R. Zipfel, R. M. Williams, R. Christie, A. Y. Nikitin, B. T. Hyman, and W. W. Webb, “Live tissue intrinsic emission microscopy using multiphoton-excited native fluorescence and second harmonic generation,” Proc. Natl. Acad. Sci. 100(12), 7075–7080 (2003). [CrossRef]
11. R. Micheal, A. Lenferink, G. F. J. M. Vrensen, E. Gelpi, R. I. Barraquer, and C. Otto, “Hyperspectral Raman imaging of neuritic plaques and neurofibrillary tangles in brain tissue from Alzheimer’s disease patients,” Sci. Rep. 7(1), 15603 (2017). [CrossRef]
12. J. Kiskis, H. Fink, L. Nyberg, J. Thyr, J.-Y. Li, and A. Enejder, “Plaque-associated lipids in Alzheimer’s diseased brain tissue visualized by nonlinear microscopy,” Sci. Rep. 5(1), 13489 (2015). [CrossRef]
13. B. Baumann, A. Woehrer, G. Ricken, M. Augustin, C. Mitter, M. Pircher, G. G. Kovacs, and C. K. Hitzenberger, “Visualization of neuritic plaques in Alzheimer’s disease by polarization-sensitive optical coherence microscopy,” Sci. Rep. 7(1), 43477 (2017). [CrossRef]
14. C.-K. Sun, “Higher Harmonic Generation Microscopy,” J. Struct. Biol. 147(1), 19–30 (2004). [CrossRef]
15. B. Weiglein, G.-J. Bakker, and P. Friedl, “Third harmonic generation microscopy of cells and tissue organization,” J. Cell Sci. 129(2), 245–255 (2016). [CrossRef]
16. S.-W. Chu, I.-H. Chen, T.-M. Liu, C.-K. Sun, S.-P. Lee, B.-L. Lin, P.-C. Cheng, M.-X. Kuo, D.-J. Lin, and H.-L. Liu, “Nonlinear bio-photonic crystal effects revealed with multimodal nonlinear microscopy,” J. Microsc. 208(3), 190–200 (2002). [CrossRef]
17. P. Campagnola, “Second harmonic generation imaging microscopy: applications to disease diagnostics,” Anal. Chem. 83(9), 3224–3231 (2011). [CrossRef]
18. G.-F. Chen, “Amyloid beta: structure, biology and structure-based therapeutic development,” Acta Pharmacol. Sin. 38(9), 1205–1235 (2017). [CrossRef]
19. V. Daebel, S. Chinnathambi, J. Biernat, M. Schwalbe, B. Habenstein, A. Loquet, E. Akoury, K. Tepper, H. Muller, M. Baldus, C. Griesinger, M. Zweckstetter, E. Mandelkow, V. Vijayan, and A. Lange, “β-sheet core of tau paired helical filaments revealed by solid-state NMR,” J. Am. Chem. Soc. 134(34), 13982–13989 (2012). [CrossRef]
20. A. C. Kwan, K. Duff, G. K. Gouras, and W. W. Webb, “Optical visualization of Alzheimer’s pathology via multiphoton excited intrinsic fluorescence and second harmonic generation,” Opt. Express 17(5), 3679–3689 (2009). [CrossRef]
21. W. H. Stoothoff, B. J. Bacskai, and B. T. Hyman, “Monitoring tau-tubulin interactions utilizing second harmonic generation in living neurons,” J. Biomed. Opt. 13(6), 064039 (2008). [CrossRef]
22. Y.-C. Chen, H.-C. Hsu, C.-M. Lee, and C.-K. Sun, “Third harmonic generation susceptibility spectroscopy in free fatty acids,” J. Biomed. Opt. 20(9), 095013 (2015). [CrossRef]
23. T.-H. Tsai, S.-P. Tai, W.-J. Lee, H.-Y. Huang, Y.-H. Liao, and C.-K. Sun, “Optical signal degradation study in fixed human skin using confocal microscopy and higher-harmonic optical microscopy,” Opt. Express 14(2), 749–758 (2006). [CrossRef]
24. C.-K. Sun, C.-T. Kao, M.-L. Wei, S.-H. Chia, F. X. Kärtner, A. Ivanov, and Y.-H. Liao, “Slide-free imaging of hematoxylin-eosin stained whole-mount tissues using combined third-harmonic generation and three-photon fluorescence microscopy,” J. Biophotonics 12(5), e201800341 (2019). [CrossRef]
25. M.-R. Tsai, S.-Y. Chen, D.-B. Shieh, P.-J. Lou, and C.-K. Sun, “In vivo optical biopsy of human oral mucosa with harmonic generation microscopy,” Biomed. Opt. Express 2(8), 2317–2328 (2011). [CrossRef]
26. S. Witte, A. Negrean, J. C. Lodder, C. P. J. de Kock, G. T. Silva, H. D. Mansvelder, and M. L. Groot, “Label-free live brain imaging and targeted patching with third-harmonic generation microscopy,” Proc. Natl. Acad. Sci. 108(15), 5970–5975 (2011). [CrossRef]
27. N. V. Kuzmin, P. Wesseling, P. C. de Witt Hamer, D. P. Noske, G. D. galgano, H. D. Mansvelder, J. C. Baayen, and M. L. Groot, “Third harmonic generation imaging for fast, label-free pathology of human brain tumors,” Biomed. Opt. Express 7(5), 1889–1904 (2016). [CrossRef]
28. C.-F. Chang, H.-C. Chen, M.-J. Chen, W.-R. Liu, W.-F. Hsieh, C.-H. Hsu, C.-Y. Chen, F.-H. Chang, C.-H. Yu, and C.-K. Sun, “Direct backward third-harmonic generation in nanostructures,” Opt. Express 18(7), 7397–7406 (2010). [CrossRef]
29. N. Matuschek, F. X. Kartner, and U. Keller, “Theory of double-chirped mirrors,” IEEE J. Select. Topics Quantum Electron. 4(2), 197–208 (1998). [CrossRef]
30. D. E. Hinkel, W. Wiersma, and S. G. Jurs, Applied Statistics for the Behavioral Sciences, 2nd ed., (Houghton Mifflin Co.1988).
31. R. Sterniczuk, M. C. Antle, R. Laferla, and R. H. Dyck, “Characterization of the 3xTg-AD mouse model of Alzheimer’s disease: Part 2. Behavioral and cognitive changes,” Brain Res. 1348, 149–155 (2010). [CrossRef]
32. D. P. Brown, M. A. Walker, A. M. Urbas, A. V. Kildishev, S. Xiao, and V. Drachev, “Direct measurement of group delay dispersion in metamagnetics for ultrafast pulse shaping,” Opt. Express 20(21), 23082–23087 (2012). [CrossRef]
33. “Technical note: The effect of dispersion on ultrashort pulses”, https://www.newport.com/n/the-effect-of-dispersion-on-ultrashort-pulses (as accessed on 14th November, 2019).
34. D. J. Kane and R. Trebino, “Characterization of arbitrary femtosecond pulses using frequency-resolved optical gating,” IEEE J. Quantum Electron. 29(2), 571–579 (1993). [CrossRef]
35. B. Liu, E. A. Macdonald, D. L. Stamper, and M. E. Brezinski, “Group velocity dispersion effects with water and lipid in 1.3 µm optical coherence tomography system,” Phys. Med. Biol. 49(6), 923–930 (2004). [CrossRef]
36. G. M. Hale and M. R. Querry, “Optical constants of water in the 200-nm to 200-µm wavelength region,” Appl. Opt. 12(3), 555–563 (1973). [CrossRef]
37. D. Kobat, D. E. Michael, N. Nishimura, A. W. Wong, C. B. Schaffer, and C. Xu, “Deep tissue multiphoton microscopy using longer wavelength excitation,” Opt. Express 17(16), 13354–13364 (2009). [CrossRef]
38. C.-K. Sun, S.-W. Chu, S.-Y. Chen, T.-H. Tsai, T.-M. Liu, C.-Y. Lin, and H.-J. Tsai, “Higher harmonic generation microscopy for developmental biology,” J. Struct. Biol. 147(1), 19–30 (2004). [CrossRef]
39. S.-Y. Chen, C.-S. Hsieh, S.-W. Chu, C.-Y. Lin, C.-Y. Ko, Y.-C. Chen, H.-J. Tsai, C.-H. Hu, and C.-K. Sun, “Noninvasive harmonics optical microscopy for long-term observation of embryonic nervous system development in vivo,” J. Biomed. Opt. 11(5), 054022 (2006). [CrossRef]
40. S.-P. Tai, W.-J. Lee, D.-B. Shieh, P.-C. Wu, H.-Y. Huang, C.-H. Yu, and C.-K. Sun, “In vivo optical biopsy of hamster oral cavity with epi-third-harmonic-generation microscopy,” Opt. Express 14(13), 6178–6187 (2006). [CrossRef]
41. L. Kou, D. Labrie, and P. Chýlek, “Refractive indices of water and ice the 0.65- to 2.5-µm spectral range,” Appl. Opt. 32(19), 3531–3540 (1993). [CrossRef]
42. Y.-H. Liao, Y.-H. Su, Y.-T. Shih, W.-S. Chen, S.-H. Jee, and C.-K. Sun, “In vivo third-harmonic generation microscopy study on vitiligo patients,” J. Biomed. Opt. 25(1), 014504 (2019). [CrossRef]
43. S.-Y. Chen, S.-U. Chen, H.-Y. Wu, W.-J. Lee, Y.-H. Liao, and C.-K. Sun, “In vivo virtual biopsy of human skin by using noninvasive higher harmonic generation microscopy,” IEEE J. Sel. Top. Quantum Electron. 16(3), 478–492 (2010). [CrossRef]
44. S. E. Crowe and G. C. R. Ellis-Davies, “In vivo charactierization of a bigenic fluorescent mouse models of Alzheimer’s disease with neurodegeneration,” J. Comp. Neurol. 521(10), 2181–2194 (2013). [CrossRef]
45. E. Drummond and T. Wisniewski, “Alzheimer’s disease: experimental models and reality,” Acta Neuropathol. 133(2), 155–175 (2017). [CrossRef]
46. W. E. Klunk, B. J. Bacskal, C. A. Mathis, S. T. Kajdasz, M. E. McLellan, M. P. Frosch, M. L. Debnath, D. P. Holt, Y. Wang, and B. T. Hyman, “Imaging Aβ plaques in living transgenic mice with multiphoton microscopy and methoxy-X04, a systemically administered congo red derivative,” J. Neuropathol. Exp. Neurol. 61(9), 797–805 (2002). [CrossRef]
47. S.-Y. Chen, H.-Y. Wu, and C.-K. Sun, “In vivo harmonic generation biopsy of human skin,” J. Biomed. Opt. 14(6), 060505 (2009). [CrossRef]
48. M. J. Farrar, E. W. Wise, J. R. Fetcho, and C. B. Schaffer, “In vivo imaging of myelin in the vertebrate central nervous system using third harmonic generation microscopy,” Biophys. J. 100(5), 1362–1371 (2011). [CrossRef]
49. D. G. Ouzounov, T. Wang, M. Wang, D. D. Feng, N. G. Horton, J. C. Cruz-Hernandez, Y.-T. Cheng, J. Reimer, A. S. Tolias, N. Nishimura, and C. Xu, “In vivo three-photon imaging of activity of GCaMP6-labeled neurons deep in intact mouse brain,” Nat. Methods 14(4), 388–390 (2017). [CrossRef]