The simultaneous detection of third harmonic (THG), and multiphoton excitation fluorescence (MPF) or second harmonic (SHG) from the same focal volume has led us to the development of a nonlinear multimodal microscopic biological imaging tool. The multimodal microscope has been applied for imaging of isolated live cardiomyocytes, and investigation of structural origin of the THG and SHG signals has been performed. By employing the different image contrast mechanisms, differentiation of structures inside a single live adult rat cardiomyocyte has been achieved. Based on structural crosscorrelation image analysis between NAD(P)H fluorescence and THG, and morphology of cardiomyocytes we were able to assign large part of the structure revealed by THG to the mitochondria. The crosscorrelation of THG with fluorescence of tetramethylrhodamine methyl ester (TMRM) labeled cardiomyocytes confirmed the mitochondrial origin of THG. The SHG generated structures were anticorrelated with THG and possessed the characteristic pattern of the myofibrils in the myocyte in accordance with the literature. Possible visualization of mitochondria with THG microscopy appeared due to enhancement of the third harmonic by multilayer arrangement of cristae.
©2005 Optical Society of America
Co-localization of different intracellular structures and simultaneous monitoring of their dynamics is one of several very successful ways to investigate subcellular interactions and functional dynamics inside a cell. Frequently, simultaneous investigations of several structures are accomplished by specifically labeling them with dyes that fluoresce at different wavelengths. Multiphoton fluorescence (MPF) microscopy is often employed for such studies, where several fluorescence labels can be efficiently excited with spectrally broad femtosecond pulses . Often more than two fluorescence channels with different detection wavelengths are utilized.
Besides multiphoton excitation fluorescence, other non-linear optical processes such as second harmonic generation (SHG) and third harmonic generation (THG) may take place during excitation. Simultaneous detection of these harmonic signals together with fluorescence is a very efficient way of imaging and enables direct comparison of structures that are revealed by the different contrast mechanisms. In previous studies, simultaneous MPF and SHG was used to image labeled lipid vesicles , labeled neurons , labeled neuroblastoma cells , and muscle and tubulin structures . Simultaneous MPF and THG was applied to imaging glial human cells . Simultaneous SHG and THG microscopy was used to monitor mitosis in a live zebrafish embryo  and study non-linear anisotropy of muscle cell . The first demonstration of simultaneous collection of three responses was accomplished by recording the full spectrum from each microvolume of an object and reconstructing images of different contrast mechanisms based on the spectral regions of each non-linear response . In this paper we describe the imaging of cardiomyocytes by the collection of MPF, SHG, and THG simultaneously into separate detection channels (see also ). This method provides fast data acquisition that enables to perform in vivo dynamic investigations of cellular activity.
The SHG and THG signals provide different contrast mechanisms for biological investigations and neither requires staining. The harmonics can be generated at non-resonance conditions i.e. the excitation energy is not deposited into the sample if the excitation wavelength is selected to be outside the absorption bands of the target molecules. The SHG can be generated in a volume of a non-linear material that has a non-centrosymmetric arrangement [5, 11, 12] or contains molecules with high second order non-linear susceptibility. Interfaces also provide symmetry breaking where SHG can be generated . The SHG is very efficiently generated in collagen [14, 15] actin-myosin complexes , tubulin , and chloroplasts . The third harmonic can be efficiently generated in a volume containing molecules with high third order non-linear susceptibility or interfaces of materials with different refractive indices . Additionally, multilayer arrangements can enhance THG . In biological samples the third harmonic is efficiently generated in rhizoids from the green algae , chloroplasts , erythrocytes , glial human cells , fixed epithelial, neuron and muscle cells , and sea urchin larval spicules .
In this study, we investigated isolated live cardiomyocytes with simultaneous MPF, SHG, and THG detection, and elucidated the structural origin of THG and SHG signals. Cardiomyocytes have been studied with MPF microscopy by recording NAD(P)H fluorescence as well as by labeling with tetramethylrhodamine methyl ester (TMRM). The MPF studies showed that mitochondria can be visualized with NAD(P)H fluorescence . TMRM labeling of mitochondria additionally revealed mitochondrial structures as well as labeled sarcoplasmic reticulum . Cardiomyocytes were also imaged by SHG and THG microscopy. SHG revealed periodic structures of sarcomeres and originated from anisotropic bands of myofibrils. THG of muscular myocytes was also recorded, and the origin of the signal was assigned to isotropic bands of myofibrils [8, 24]. In this study we have determined that very strong THG signal is generated in mitochondria probably due to THG enhancement from the multilayer structures. We investigated the co-localization of structures revealed by THG, SHG, and fluorescence in the rather highly ordered interleaved organization of mitochondria and myofibrils in the cardiomyocyte.
2. Materials and methods:
2.1 Ti:Sapphire laser
A home built femtosecond Ti:sapphire oscillator with tunable laser wavelength from 760 nm to 840 nm was used for imaging NAD(P)H multiphoton excitation fluorescence of cardiomyocytes. The oscillator was pumped by a Millenia V cw-laser from Spectra-Physics. The cavity layout was based on a standard delta-cavity design with one of the cavity arms containing an additional focus to accommodate an acousto-optic Bragg cell. This allowed us to cavity dump the laser beam and conveniently vary the repetition rate of the pulses from 200 kHz to 4 MHz. The base laser pulse repetition rate was 76 MHz. Dispersion compensation within the cavity was done by a fused silica prism pair. The pulse duration was about 25 fs at an average output power of 120 mW (output coupling 5%, pump power 1.5 W).
2.2 Nd:glass laser
For 1064 nm excitation, we used a diode-pumped, Nd:glass laser. The home-built Nd:glass laser was based on the work of Kopf et al. . An inhomogeneously broadened silicate glass was cw-pumped by one diode laser (Spectra Diode Laboratories SDL-2472, 3.0 W). We achieved self-starting with an output coupling of 3%. The pulse duration was ~150 fs, with an average output power of 100 mW at a repetition rate of 94 MHz. For stable mode-locking an intracavity semiconductor saturable absorber mirror (SESAM)  was used.
2.3 Microscope setup:
Simultaneous imaging with THG, SHG and MPF signals was accomplished by a home built microscope. The basic outline of the microscope is presented in Fig. 1. The Ti:sapphire or Nd:glass laser was employed for imaging. For obtaining 2-dimensional images, the laser beam was rastered by the two closed-loop galvanometric mirrors (GSI Lumonics, VM2000 Series). The scanned beam was appropriately relayed to the high numerical aperture objective via a 1:1 telescope as shown in Fig. 1. The IR-Achroplan 63× 0.9NA water immersion objective (Zeiss) was used for the excitation. The harmonics in forward direction were
2.4 Image analysis
The acquired stacks of two-dimensional images were reconstructed into a 3-dimensional data set and deconvoluted using a point spread function (PSF) obtained from the 3D image of 0.1 μm diameter fluorescing polystyrene beads. 3D-Doctor software (Able Software Corp.) was used for the deconvolution. Deconvolution was necessary to avoid artificial overlap of closely located structures that would obscure structural correlation image analysis (see below). Deconvoluted, 3D images were reconstructed into the stacks of 2D frames. The 2D frames of THG, SHG and MPF were correlated between each other. Initially, Pearson’s correlation coefficient was calculated for each slice. The coefficient, which is a representation of how linear the relationship is between two signals, is given by:
where Ai represents the intensity of the i-th pixel from channel A, and Bi is the intensity of the i-th pixel from channel B. Note that pixels with intensity less than the threshold value were set to zero where the threshold values were determined based on the intensity level for regions of the image not containing the cardiomyocyte. N is the number of pixels in the slice that have either one or both channels with non-zero intensity after thresholding. Pixels which have zero intensity in both channels are not included in the calculation. The output of Pearson’s coefficient for a comparison of two 2D frames is a single number between -1 and +1 that represents the relationship between two areas. A coefficient of +1 represents a perfect correlation while -1 denotes a perfect negative relationship (anti-correlation). The Pearson coefficient is useful in providing information regarding two acquired images, but it does not allow one to compare individual structures within the images. Pearson’s coefficient is also limited in that if the single output number is not either +1 or -1, there is no indication as to which parts of the images overlap and which do not. In order to examine the overlap of individual structures between two signals, a different kind of correlation was necessary.
We carried out a structural cross correlation image analysis (SCIA), which is a modified version of image cross correlation analysis. The SCIA software was written with LabView (National Instruments) interface. A more detailed description of SCIA can be found in Greenhalgh et al . The SCIA compares signal intensities of simultaneously recorded 2D slices from two separate channels. The comparison is performed on a pixel by pixel basis, and correlated 2D images are reconstructed into a 3D output of structures that appear in the same spatial location in both images. The algorithm for the SCIA is given by:
where the pixel intensities from the two different channels are represented by a and b. The subscript i denotes the i-th pixel of the image. The min denotes the threshold value. The threshold values were determined for each slice by measuring the maximum background intensity from the regions not containing the cardiomyocyte. The max denotes the maximum signal value determined by examining the histogram of the pixel intensities of each frame. The examination was performed in order to remove high intensity outliers. The renormalization of pixel intensities with the maximum signal value of the slice enabled a full use of the dynamic range of the intensity spectrum. The C represents the correlation value. The correlation value was calculated for each pixel and assumed non zero value if signals higher than background were present in both images, otherwise the Ci value was set to zero as described in Eq. (2).
By assigning a different color to both of the uncorrelated 3D structures and the correlated structure and combining these structures, a single, tri-colored co-localized structure could be generated. In this case, the two uncorrelated structures were constructed from the original thresholded images (as described above) by assigning zero values to the correlated pixels. The generated 3D structure was rendered for analysis and understanding of structural relations and spatial correlations of the investigated images. Rendering was carried out using a home built program as well as with the Volume J plug-in of ImageJ.
2.5 Cardiomyocyte sample preparation
Experiments were performed in accordance with institutional guidelines and the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Cardiomyocytes of adult Sprague-Dawley rats were isolated as previously described  and generously provided by the laboratories of Wilbur Y. W. Lew (Cardiology Section, Department of Medicine, V.A. San Diego Healthcare System and UCSD) and of Wolfgang H. Dillman (Division of Endocrinology & Metabolism, Department of Medicine, School of Medicine, UCSD). Briefly, rats were anesthetized with sodium pentobarbital, the heart excised and perfused with collagenase B and protease. Cardiomyocytes were plated on sterile custom-made plastic dishes with a laminin-coated coverslip of fused silica and maintained in Dulbecco’s modified Eagle’s medium at room temperature for 2 to 24 hours before experiments. A typical white light image of an isolated cardiomyocyte, superimposed with a rectangle, showing a usual region scanned with the nonlinear multimodal microscope is depicted in Fig. 2. The thickness of the plated cardiomyocytes was on average around 30 μm. The cardiomyocytes were imaged in the same medium at ambient environment. Imaging cells with 1064 nm excitation continuously for several hours produced no obvious toxicity and structural alterations were not observed during imaging unless or until the cell proceeded to hypercontraction. The probablility of hypercontraction of imaged and non-imaged cells was the same. In some experiments, cardiomyocytes were labeled with the mitochondria-specific fluorophore TMRM (0.27 nM) to correlate mitochondrial fluorescence with THG and SHG signals. The average thickness of cardiomyocytes was about 30μm. More than ten labeled and more than ten unlabeled cells were imaged with each combination of the two contrast mechanisms.
3.1. THG and NAD(P)H fluorescence
The 3-D imaging of live cardiomyocytes was performed by simultaneously collecting THG and MPF on two separate detection channels. Images were recorded with 837 nm peak excitation from the cavity dumped Ti:Sapphire laser. The femtosecond laser spectral bandwidth was about 30 nm providing effective NAD(P)H excitation as well as flavin adenine dinucleotide (FAD) and lipoamide dehydrogenase (LipDH) excitation. The emission comprised mainly of NAD(P)H, FAD and LipDH fluorescence . The NAD(P)H multiphoton excitation fluorescence is observed in the 400 – 600 nm region while FAD and LipDH fluorescence lays in the 460 – 600 nm range. For imaging, we collected the fluorescence between 435 and 500 nm which originated predominantly from NAD(P)H . The highest fluorescence of NAD(P)H is usually confined to mitochondria, where the tricarboxylic cycle takes place . Figure 3(a) shows the NAD(P)H MPF of a single image slice while Fig. 3(b) shows the corresponding image obtained with THG. Both images appear very similar. The intensity profiles of the line in Fig. 3(a) and 3(b) is depicted in Fig. 3(c). In general, they follow a similar path. The 2D SCIA result for the same nonprocessed slice is shown in Fig. 3(d), where we see the majority of the sample is correlated, shown in red. The cardiomyocyte structures obtained with 3-D fluorescence imaging (Fig. 3(e)) revealed typical granular structures that can be attributed to mitochondria ordered in rows along the myofibrils . The mitochondria are approximately 1-2 μm in diameter. The image generated from the third harmonic that was recorded simultaneously with the fluorescence is presented in Fig. 3(f). The THG shows granular structures with periodicity and spatial location similar to those generated from the fluorescence.
For comparison we co-localized the THG and NAD(P)H MPF images (see Fig. 3(g)). The red colored structure represents the correlated signal, while blue is uncorrelated THG, and green is the uncorrelated fluorescence. Figure 3(h) presents only the correlated volume of the cardiomyocyte. We perform structural cross correlation image analysis only on pixels that had substantial signal in at least one of the channels. In this way we could avoid correlation of the image backgrounds, which do not belong to the highlighted THG or MPF structures but would obscure the structural cross-correlation analysis. The SCIA is outlined in Materials and Methods. The cross-correlation image (Fig. 3(h)) once again shows an ordered granular pattern arranged along the myofilaments. There are only few pixels of uncorrelated THG (blue) visible in the co-localized image (Fig. 3(g)). Those pixels have typically lower THG signal intensity. Some of the fluorescence signal is not correlated with the THG signal. The uncorrelated fluorescence (Fig. 3(g)) appears to envelop the correlated structure, or shows up as a separate row of granular structures arranged along the myofilaments. The envelope structure emerges because the point spread function is broader for two-photon excitation than for THG. This blurs fluorescing structures in the image more than THG structures. The deconvolution procedure minimizes the blur; however some of it still remains giving envelope-like artifacts in the correlated image. The uncorrelated fluorescent granular structures, most probably, represent mitochondria that have their lamellae oriented predominantly parallel to the laser beam rendering unfavorable conditions for THG.
Unfortunately, around 800 nm excitation wavelength the live cardioyocytes can only be imaged for several minutes before hypercontraction occurs. Damage of the cardiomyocytes is most likely inflicted by the multiphoton absorption of NAD(P)H, FAD and LipDH. The lowest phototoxicity with Ti:sapphire excitation has been achieved by tuning the oscillator to the lower absorption region of NAD(P)H around 837 nm wavelengths, and adjusting the pulse repetition rate to 3.1 MHz and pulse energy to 0.6 nJ at the sample while keeping the average laser power constant. With these conditions two-photon excitation fluorescence in the 435 -500 nm range was still at measurable level and excitation pulse energy was high enough to generate appreciable intensity of THG.
3.2 THG and 1064 nm excitation autofluorescence
Live cardiomyocytes can be imaged for many hours with the Nd:glass laser excitation that peaks at 1064nm. At this excitation wavelength NAD(P)H as well as FAD and LipDH are not excited  i.e. the deposition of the excitation energy into the cardiomyocyte is drastically reduced. This enables higher laser powers, which in turn, generate intense THG and SHG signals. The ~ 20 mW excitation power at the sample was used, that resulted in 0.2nJ energy per pulse with the 94MHz repetition rate laser. Figure 4(a) shows a single raw image slice revealed by autofluorescence emitted at 630-700nm while the 3D structures of the cardiomyocyte are revealed in the deconvoluted rendered volume displayed in Fig. 4(c). Only a few bright spots spanning through the cardiomyocyte perpendicularly to the myofibrils are visible. Similar structures have been observed by  with 900 nm excitation. The origin of the revealed structure and the fluorophore is unknown.
The simultaneously recorded 2D and 3D THG images are shown in Fig. 4(b) and 4(d), respectively. The structure reveals characteristic granular rows of mitochondria aligned along the myofibrils. The THG image shows similar structures to those recorded at 837 nm. The structural cross-correlation analysis is presented in Fig. 4(e). It is clearly visible that fluorescing structures overlap with corresponding structures revealed by THG (overlap colored in red). The remaining structures in Fig. 4(e) originate from the uncorrelated THG signal and are depicted in blue. Uncorrelated fluorescence structures are coded in green and are practically absent in the co-localized image. The correlated structures are presented separately in Fig. 4(f). It is interesting to note that correlated structures appear as continuous oval shape volumes with the longest axes oriented perpendicular to myofibrils and parallel to the direction of the laser beam propagation. The structures are several times larger in axial direction (~8 μm) than the diffraction limit of the imaging system. THG would not be detected inside an isotropic media, suggesting that the observed structures contain a multi membrane arrangement. The correlated structures appear elongated laterally in the scanning direction. The elongated shape most likely is an artifact that comes from lateral dragging of the structure optically trapped by the laser during scanning. Apparently, those structures are much more susceptible to trapping compared to mitochondria.
3.3. THG and TMRM fluorescence excited with 1064 nm
We also compared structures revealed by MPF and THG from TMRM labeled cardiomyocytes. TMRM is a lipid membrane permeable cationic dye. It accumulates in the compartments with high negative potential according to the Nernst equilibrium. This dye accumulates preferentially in energized mitochondria and endoplasmic reticulum, revealing them in the fluorescence image . We imaged TMRM labeled cardiomyocytes with 1064 nm excitation pulses by simultaneously recording THG and MPF into two separate channels. The TMRM MPF raw data image and rendered volume are shown in Fig. 5(a) and 5(e), respectively. The image reveals spherical structures organized in a grid like pattern that has a characteristic spatial arrangement of mitochondria in cardiomyocytes . The 2D and 3D THG images recorded simultaneously with fluorescence are shown in Fig. 5(b) and 5(f), respectively. The THG images are similar to the previous images in Fig. 4(b) and 4(d). Note, that the images in Fig. 5 have more than twice the magnification of Fig. 4. The THG image has the same structural similarities as the fluorescence image. The intensity profile of the line in Fig. 5(a) and 5(b) is shown Fig. 5(c); as with Fig. 3(c), there are similarities in the profiles of THG and MPF.
The same structural cross-correlation analysis that had been done previously with THG and NAD(P)H fluorescence was carried out on the images of THG and TMRM fluorescence. The co-localized 2D and 3D images are shown in Fig. 5(d) and 5(g), respectively. The uncorrelated THG is presented in blue and uncorrelated fluorescence is presented in green. The correlated part is shown in red in Fig. 5(d) and 5(g), and presented separately in Fig. 5(h). We found only partial correlation between images of TMRM fluorescence and THG. The correlated image shows oval structures arranged in rows positioned along myofibrils. As viewed from the side of the three-dimensional image, the correlation appears in the upper part of the cardiomyocyte. The lower part of the cardiomyocyte, (part that is closest to the plated surface,) is mostly dominated by the THG signal. The correlation provides evidence of the THG signal being generated in the mitochondria. More uncorrelated signals are observed between TMRM fluorescence and THG images than between NAD(P)H fluorescence and THG. The uncorrelated fluorescence appears possibly due to accumulation of TMRM not only in mitochondria, but also in the sarcoplasmic reticulum. The other two factors of THG dependence: (i) the membrane orientation with respect to the laser beam, and (ii) the differences in MPF and THG point spread functions, contribute to the amount of uncorrelated fluorescence signal in the co-localized image Fig. 5(g). The uncorrelated THG can originate from membranous cellular structures other than mitochondria. The uncorrelated THG can also appear due to non-uniform staining that might have occurred in our sample, because less fluorescence was observed from the structures located closest to the plated side of the cardiomyocytes where access of dyes is more restricted. We used an extremely low TMRM concentration of 0.27 nM. Furthermore, the TMRM excitation may produce oxygen radicals leading to depolarization of connected groups of mitochondria. Therefore, scanning of upper layers may depolarize the mitochondria in the lower layers of the cardiomyocytes prior to the imaging. In addition to TMRM fluorescence, auto-fluorescence contributed to the MPF and correlation images as described in the previous paragraph (Fig. 4).
3.4. THG and SHG
Three-dimensional images of cardiomyocytes were obtained with a simultaneous recording of THG and SHG into two separate detection channels. For imaging, we used the mode-locked Nd:glass laser with 1064 nm excitation. The SHG image of the cardiomyocyte is presented in Fig. 6(a) with the rendered structures shown in Fig. 6(c). Myofibrils of the cardiomyocyte are clearly visible. They appear as rows of alternating high-low intensity structures. By now it is well established that SHG is generated from the non-centrosymmetric microcrystalline structures  corresponding to anisotropic bands of the myofilaments. The distance between two intense structures is about 1.8 μm. That relates well with the size and periodicity of sarcomeres . The myofibril-like structures revealed by SHG have on average 5 μm in 35]. It is difficult to resolve individual myofibrils with an optical microscope when they are bundled together in close proximity.
The THG image of a cardiomyocyte is presented in Fig. 6(b) with the rendered volume shown in Fig. 6(d). The granular structures in the THG image appear along the myofibrils. The image closely resembles features of the THG image recorded with the Ti:sapphire laser (Fig. 3) and can be assigned primarily to mitochondria.
The image of co-localized SHG and THG structures is presented in Fig. 6(e). The uncorrelated SHG structures are presented in green and uncorrelated THG structures are presented in blue. The correlated structures that appear very seldom in the image are shown in red. There is almost no visible correlation between structures revealed by SHG and THG (Fig. 6(e)). However, it should be mentioned that there is weak correlation from the SHG and THG signals generated at the sarcolemma (see Fig. 7) where low correlation accumulates over the whole image area to a significant value of Pearson’s coefficient but is thresholded in the pixel by pixel correlation analysis. This shows that SHG and THG are generated from completely different formations.
The co-localized image (Fig. 6(e)) showed that oval shape structures revealed by THG were preferentially positioned at the maximal SHG intensity of sarcomeres. This corresponds well with localization of mitochondria close to the anisotropic bands of myofibrils. The SHG highlighted myofibrils appeared in a central part of the cardiomyocyte. In contrast, the THG revealed mitochondria were localized mainly along the sides of the cell. The interface orientation dependence of THG efficiency most likely resulted in preferential visualization of mitochondria with lamellar membranes oriented perpendicular to the laser beam propagation.
3.5. Linear image crosscorrelation at different depth of cardiomyocytes
Pixel-by-pixel analysis of SCIA used in the previous paragraphs renders correlated and two uncorrelated 3-D structures that give visual information about morphology of the structure. The traditional image cross-correlation analysis of the optical sections obtained with two different signals is also useful to perform for quantitative comparison. We carried out image cross-correlation analysis of optical sections at different depths of cardiomyocytes by using Pearson’s coefficient. Pearson’s coefficient provides a measure of the linear relationship between two images. The two signals that are being generated within the same structure will result in a positive correlation. If, however, a structure generates one signal and not the other then the correlation will be negative. Positive correlation also occurs for regions with background intensities in both images. By excluding pixels that have background intensity for both channels, we are able to eliminate correlation of the area around the cardiomyocyte as well as regions inside the cardiomyocyte that do not generate appreciable signal intensity. The results of this analysis, shown in Fig. 7, reveal that THG and MPF show mainly a positive correlation while THG and SHG appear negatively correlated. This agrees with the initial pixel-by-pixel structural cross-correlation of THG, SHG, and MPF. The cell to cell variations of structural cross-correlations revealed by different contrast mechanisms varies by about 10 to 15%. Figure 7 also shows that the correlation coefficient approaches zero in the middle of the cardiomyocyte, showing the presence of both correlated and anticorrelated structures in the image.
The third harmonic can be generated from any structural interface and depends on the differences in the refractive index of two media and differences in the third order nonlinear susceptibility of the materials . In addition, the THG signal generation efficiency depends on the orientation of the interface with respect to the excitation beam . The largest THG signal is generated when an interface is positioned perpendicular to the propagation of the beam.
There are many interfaces inside cardiomyocytes that potentially could generate THG signals. Besides mitochondria, outer membranes, sarcoplasmic reticulum, myofibrils and other membrane containing structures are good candidates for visualization by THG. Our data indicates that a large part of the thresholded THG signal generated in the cardiomyocytes originates from mitochondria. In fact, a very low intensity THG signal revealed isotropic bands of myofibrils that appear as a striped pattern anticorrelating with SHG (see also ). Our analysis involves thresholding the THG. This eliminates very low intensity signals, which in turn enables discrimination of the mitochondria that generate substantially higher THG intensity. The main reasons for THG enhancement in mitochondria most likely appears due to the multilamelar structure of the densely folded cristae, and outer and inner membrane in the mitochondria. Tsang in his work  showed this effect for dielectric multilayer structures. In the multilayer structure, in addition to the orientation of the membranes with respect to the laser propagation direction, the spacing between layers plays a significant role in the THG process. THG wave fronts generated at each membrane may interfere constructively or destructively depending on the distance between the membranes. It is therefore not surprising that some mitochondria observed in NAD(P)H or TMRM labeled MPF image do not appear in THG. The most intense THG comes from subsarcolemmal mitochondria that are along the sides of the cardiomyocyte. The mitochondria along the sides have their cristae oriented approximately perpendicular to the laser beam propagation resulting in significant enhancement of THG, while the mitochondria which are situated between the myofibrils, the interfibrillar mitochondria, are likely to have a less favorable cristae orientation or spatial periodicity. Although there are no known structural differences between subsarcolamellal and interfibrillar mitochondria , some may still exist that could also be affecting the THG efficiency. In any case, the biochemical properties of subsarcolemmal and interfibrillar mitochondria from rat cardiac muscle appear to differ significantly . It is very likely therefore that we are experiencing THG enhancement from the multilamellar structure of subsarcolemmal mitochondria. Therefore THG appears to be a wonderful tool for mitochondrial visualization. THG does not require staining, which can influence structural organization of mitochondria, ion distribution, and free radical generation during laser scanning.
The development of an SHG, THG, and MPF microscope has enabled us to visualize simultaneously different subcellular structures, spatially correlate them and position these structures in 3-D images. The new structural cross-correlation analysis helped us to construct the correlated and uncorrelated 3-D images and showed if the signals originate from the same or two different structures. By applying simultaneous MPF, SHG, and THG detection microscope for imaging isolated live cardiomyocytes we were able to assess the structural origin of THG and SHG signals. Large part of the THG signal was generated from the mitochondria possibly due to the signal enhancement from multilayer arrangement of the crista. The SHG signal was generated from anisotropic bands of myofibrils, consistently with the literature . Our investigation has shown that simultaneous imaging technique offers new possibilities in obtaining complementary information on cellular structures and colocalization between cellular organelles.
The authors are grateful to Dr. Lew and Dr. Dillman, Department of Medicine at UCSD, for both providing the preparation of the cardiomyocytes and for stimulating discussions. This work was supported by the National Institute of Health under grant R21 EB001722-01, the National Science and Engineering Research Council of Canada grants and grant D94.016 of the Netherlands Heart Foundation. C. Greenhalgh also acknowledges the financial support of NSERC.
References and links
1. K. Konig, “Multiphoton microscopy in life sciences,” J Microsc-Oxford 200, 83–104 (2000). [CrossRef]
2. L. Moreaux, O. Sandre, and J. Mertz, “Membrane imaging by second-harmonic generation microscopy,” J Opt Soc Am B 17(10), 1685–1694 (2000). [CrossRef]
4. P. J. Campagnola, M. D. Wei, A. Lewis, and L. M. Loew, “High-resolution nonlinear optical imaging of live cells by second harmonic generation,” Biophys J 77(6), 3341–3349 (1999). [CrossRef] [PubMed]
5. P. J. Campagnola, A. C. Millard, M. Terasaki, P. E. Hoppe, C. J. Malone, and W. A. Mohler, “Three-dimensional high-resolution second-harmonic generation imaging of endogenous structural proteins in biological tissues,” Biophys. J. 82(1), 493–508 (2002). [CrossRef]
6. R. Barille, L. Canioni, S. Rivet, L. Sarger, P. Vacher, and T. Ducret, “Visualization of intracellular Ca2+ dynamics with simultaneous two-photon-excited fluorescence and third-harmonic generation microscopes,” Appl Phys Lett 79(24), 4045–4047 (2001). [CrossRef]
7. S. W. Chu, S. Y. Chen, T. H. Tsai, T. M. Liu, C. Y. Lin, H. J. Tsai, and C. K. Sun, “In vivo developmental biology study using noninvasive multi-harmonic generation microscopy,” Opt Express 11(23), 3093–3099 (2003). http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-23-3093 [CrossRef] [PubMed]
8. S. W. Chu, S. Y. Chen, G. W. Chern, T. H. Tsai, Y. C. Chen, B. L. Lin, and C. K. Sun, “Studies of x((2))/x((3)) tensors in submicron-scaled bio-tissues by polarization harmonics optical microscopy,” Biophys. J. 86(6), 3914–3922 (2004). [CrossRef] [PubMed]
9. S. W. Chu, I. H. Chen, T. M. Liu, P. C. Chen, C. K. Sun, and B. L. Lin, “Multimodal nonlinear spectral microscopy based on a femtosecond Cr : forsterite laser,” Opt Lett 26(23), 1909–1911 (2001). [CrossRef]
10. V. Barzda, C. Greenhalgh, J. Aus der Au, J. A. Squier, S. Elmore, and J. H. van Beek, “Second- and third-harmonic generation and multiphoton excitation fluorescence microscopy for simultaneous imaging of cardiomyocytes,” in Commercial and Biomedical Applications of Ultrafast Lasers IV, J. Neev, C. B. Schaffer, and A. Ostendorf eds, Proc. of SPIE 5340, 96–103 (2004). [CrossRef]
11. R. W. Boyd, Nonlinear optics, 2nd ed. (Academic Press, San Diego, CA, 2003).
13. L. Moreaux, O. Sandre, M. Blanchard-Desce, and J. Mertz, “Membrane imaging by simultaneous second-harmonic generation and two-photon microscopy,” Opt Lett 25(5), 320–322 (2000). [CrossRef]
14. I. Freund, M. Deutsch, and A. Sprecher, “Connective-Tissue Polarity - Optical 2nd-Harmonic Microscopy, Crossed-Beam Summation, and Small-Angle Scattering in Rat-Tail Tendon,” Biophys J 50(4), 693–712 (1986). [CrossRef] [PubMed]
15. Y. C. Guo, P. P. Ho, H. Savage, D. Harris, P. Sacks, S. Schantz, F. Liu, N. Zhadin, and R. R. Alfano, “Second-harmonic tomography of tissues,” Opt. Lett. 22(17), 1323–1325 (1997). [CrossRef]
16. J. A. Squier, M. Muller, G. J. Brakenhoff, and K. R. Wilson, “Third harmonic generation microscopy,” Opt Express 3(9), 315–324 (1998). http://www.opticsexpress.org/abstract.cfm?URI=OPEX-3-9-315 [CrossRef] [PubMed]
18. M. Muller, J. Squier, K. R. Wilson, and G. J. Brakenhoff, “3D microscopy of transparent objects using third-harmonic generation,” J Microsc-Oxford 191, 266–274 (1998). [CrossRef]
19. A. C. Millard, P. W. Wiseman, D. N. Fittinghoff, K. R. Wilson, J. A. Squier, and M. Muller, “Third-harmonic generation microscopy by use of a compact, femtosecond fiber laser source,” Appl Optics 38(36), 7393–7397 (1999). [CrossRef]
20. D. Yelin, D. Oron, E. Korkotian, M. Segal, and Y. Silbergerg, “Third-harmonic microscopy with a titanium-sapphire laser,” Appl Phys B-Lasers O 74, S97–S101 (2002). [CrossRef]
23. C. Xu, W. Zipfel, J. B. Shear, R. M. Williams, and W. W. Webb, “Multiphoton fluorescence excitation: New spectral windows for biological nonlinear microscopy,” P Natl Acad Sci USA 93(20), 10763–10768 (1996). [CrossRef]
24. 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-Oxford 208, 190–200 (2002). [CrossRef]
25. D. Kopf, F. X. Kartner, U. Keller, and K. J. Weingarten, “Diode-Pumped Mode-Locked Nd-Glass Lasers with an Antiresonant Fabry-Perot Saturable Absorber,” Opt Lett 20(10), 1169–1171 (1995). [CrossRef] [PubMed]
26. U. Keller, K. J. Weingarten, F. X. Kartner, D. Kopf, B. Braun, I. D. Jung, R. Fluck, C. Honninger, N. Matuschek, and J. A. derAu, “Semiconductor saturable absorber mirrors (SESAM’s) for femtosecond to nanosecond pulse generation in solid-state lasers,” IEEE J. Sel. Top. Quantum Electron. 2(3), 435–453 (1996). [CrossRef]
27. C. Greenhalgh, R. Cisek, N. Prent, A. Major, J. Aus der Au, J. Squier, and V. Barzda, “Time and structural crosscorrelation image analysis of microscopic volumes, simultaneously recorded with second harmonic generation, third harmonic generation, and multiphoton excitation fluorescence microscopy,” in Photonic Applications in Biosensing and Imaging, W.C.W. Chan, K. Yu, U.J. Krull, R.I. Hornsey, B.C. Wilson, and R.A. Weersink. eds., Proc. of SPIE 5969, 557–564 (2005).
28. H. L. Li, J. Suzuki, E. Bayna, F. M. Zhang, E. D. Molle, A. Clark, R. L. Engler, and W. Y. W. Lew, “Lipopolysaccharide induces apoptosis in adult rat ventricular myocytes via cardiac AT(1) receptors,” Am J Physiol-Heart C 283(2), H461–H467 (2002).
29. 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. USA 100(12), 7075–7080 (2003). [CrossRef] [PubMed]
30. K. Blinova, C. Combs, P. Kellman, and R. S. Balaban, “Fluctuation analysis of mitochondrial NADH fluorescence signals in confocal and two-photon microscopy images of living cardiac myocytes,” J. Microsc.-Oxford 213, 70–75 (2004). [CrossRef]
31. J. Eng, R. M. Lynch, and R. S. Balaban, “Nicotinamide Adenine-Dinucleotide Fluorescence Spectroscopy and Imaging of Isolated Cardiac Myocytes,” Biophys. J. 55(4), 621–630 (1989). [CrossRef] [PubMed]
32. D. L. Farkas, M. D. Wei, P. Febbroriello, J. H. Carson, and L. M. Loew, “Simultaneous Imaging of Cell and Mitochondrial-Membrane Potentials,” Biophys J 56(6), 1053–1069 (1989). [CrossRef] [PubMed]
33. D. N. Romashko, E. Marban, and B. O’Rourke, “Subcellular metabolic transients and mitochondrial redox waves in heart cells,” P Natl Acad Sci USA 95(4), 1618–1623 (1998). [CrossRef]
34. S. Inouâe and R. J. Walter, Video microscopy (Plenum Press, New York, 1986).
35. F. Martini and W. C. Ober, Fundamentals of anatomy & physiology, 5th ed. (Prentice Hall, Upper Saddle River, N.J. ; Toronto, 2001), p. 1 v. (various pagings).
36. L. Skarka and B. Ostadal, “Mitochondrial membrane potential in cardiac myocytes,” Physiol Res 51(5), 425–434 (2002). [PubMed]
37. J. W. Palmer, B. Tandler, and C. L. Hoppel, ”Biochemical Properties of Subsarcolemmal and Inter-Fibrillar Mitochondria Isolated from Rat Cardiac-Muscle,” J. Biological Chem. 252(23), 8731–8739 (1977).