Label-Free Evaluation of Myocardial Infarction and Its Repair by

Jun 10, 2014 - using the hypothesis that the myocardium in the course of myocardial infarction and its repair could be recognized by spontaneous Raman...
0 downloads 0 Views 6MB Size
Article pubs.acs.org/ac

Label-Free Evaluation of Myocardial Infarction and Its Repair by Spontaneous Raman Spectroscopy Nanae Nishiki-Muranishi,†,‡ Yoshinori Harada,† Takeo Minamikawa,† Yoshihisa Yamaoka,† Ping Dai,† Hitoshi Yaku,‡ and Tetsuro Takamatsu*,† †

Department of Pathology and Cell Regulation and ‡Department of Cardiovascular Surgery, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, 465 Kajii-cho Kawaramachi-Hirokoji, Kamigyo-ku, Kyoto 602-8566, Japan ABSTRACT: Raman spectroscopy, which provides information about molecular species and structures of biomolecules via intrinsic molecular vibrations, can analyze physiological and pathological states of tissues on the basis of molecular constituents without staining. In this study, we analyzed Raman spectra of myocardial infarction and its repair in rats using the hypothesis that the myocardium in the course of myocardial infarction and its repair could be recognized by spontaneous Raman spectroscopy on the basis of chemical changes in myocardial tissues. Raman spectra were acquired from unfixed frozen cross sections of normal and infarcted heart tissues upon excitation at 532 nm. Raman spectra of the infarcted tissues were successfully obtained at characteristic time points: days 2, 5, and 21 after coronary ligation, at which the main components of the infarcted region were coagulation necrosis, granulation tissue, and fibrotic tissue, respectively. The latent variable weights calculated by a multivariate classification method, partial least-squares-discriminant analysis (PLS-DA), revealed fundamental information about the spectral differences among the types of tissues on the basis of molecular constituents. A prediction model for the evaluation of these tissue types was established via PLS-DA. Cross-validated sensitivities of 99.3, 95.3, 96.4, and 91.3% and specificities of 99.4, 99.5, 96.5, and 98.3% were attained for the discrimination of normal, necrotic, granulation, and fibrotic tissue, respectively. A two-dimensional image of a marginal area of infarction was successfully visualized via PLS-DA. Our results demonstrated that spontaneous Raman spectroscopy combined with PLS-DA is a novel label-free method of evaluating myocardial infarction and its repair.

S

cells to necrotic tissue, induction of angiogenesis, and proliferation of fibroblasts occur; that is, granulation tissue is developed. Collagen maturates in the infarcted area, and formation of fibrotic tissue is completed after several weeks. This process comprises chemical changes of biomolecules in the myocardial tissues; however, vibrational signatures reflected in Raman spectra are still unclear. In this study, we sought to analyze Raman spectra of myocardial infarction and its repair in rats in detail using the hypothesis that the myocardium in the course of myocardial infarction and its repair could be recognized by spontaneous Raman spectroscopy on the basis of chemical changes in myocardial tissues. We obtained Raman spectra corresponding to the pathological types of infarcted tissues and elucidated the spectral features of each type of tissue. We also clarified spectral signatures indicating the differences among the Raman spectra obtained from each type of tissue using a multivariate classification method, partial least-squares-discriminant analysis (PLS-DA).20 We constructed a multivariate discrimination

pontaneous Raman spectroscopy can provide quantitative chemical information about samples according to specific vibrational signatures of chemical bonds1,2 and permits noninvasive analysis of biomolecules without staining. Recently, spontaneous Raman spectroscopy has been applied to molecular analysis of cells and tissues. For example, the noninvasiveness of Raman spectroscopy made possible cell imaging,3−8 and medical applications of Raman spectroscopy have confirmed its accuracy in identifying molecules composing tissues.9−18 We previously demonstrated, for the first time, that Raman spectra of cardiomyocytes and old infarcted scar lesions in the proximity of cardiomyocytes were clearly recorded without staining by analyzing features of Raman spectra of cytochrome c and collagen type I in an experiment with rat hearts.19 This innovative study targeted normal and completely fibrous myocardial tissues. However, the myocardium at earlier time points in the course of myocardial infarction and its repair has not been visualized. For the precise evaluation of the infarcted myocardium, more information about changes of chemical components is required. The characteristic process of the healing of injury and maturation of infarcted tissue is as follows: The sustained myocardial ischemia induces necrosis of cardiomyocytes. After 1−2 days, infiltration of inflammatory © 2014 American Chemical Society

Received: February 12, 2014 Accepted: June 10, 2014 Published: June 10, 2014 6903

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

Figure 1. Histological examination and Raman spectra derived from cardiac tissues. (A) H&E-stained specimen of normal heart. (B−D) H&Estained specimens of myocardial infarction obtained at various postligation time points: day 2 (B), day 5 (C), and day 21 (D1). (D2) Azan staining of myocardial infarction at day 21. Histology shows coagulation necrosis (B), granulation tissue (C), and fibrotic scar (D1 and D2). (E) Raman spectra (means ± one standard deviation) of normal tissue [n = 2; number of spectra (NOS) = 150], necrotic tissue on day 2 (n = 3; NOS = 150), necrotic tissue on day 5 (n = 4; NOS = 150), granulation tissue on day 5 (n = 4; NOS = 500), and fibrotic tissue on day 21 (n = 2; NOS = 150). The scale bars are 50 μm.

a freezing mixture (frozen carbon dioxide with acetone), and stored at −80 °C until they were examined. Samples for Raman analysis were cut at a 5 μm thickness and subsequently mounted on 0.17 mm thick quartz flats (Daico MFG, Kyoto, Japan). Serial sections for hematoxylin and eosin (H&E) staining were used to confirm the histology, and Azan staining was also used for evaluating fibrotic tissue. For day 2 samples, we obtained spectra from the areas of necrosis confirmed with serial sections of H&E-stained specimens. For day 5 samples, we obtained spectra from the areas of necrotic and granulation tissue because necrotic tissue was still present in day 5 tissue. For day 21 samples, we obtained spectra from the areas of fibrotic tissue confirmed with serial sections of H&E-stained specimens and Azan-stained specimens. We also independently created an infarcted heart 5 days after coronary ligation for PLS-DA imaging. The heart was used only for imaging and not for model establishment. Preprocessing for the heart, including perfusion and freezing, was conducted in the same way described above. Measurement of Raman Spectra. Raman spectra of hearts were recorded with a laser Raman confocal microscope [RAMAN-11 (Nanophoton, Osaka, Japan)] that has been described previously.4,13,19 A frequency-doubled Nd:YAG laser operating at 532 nm was employed for excitation. The samples

model with PLS-DA. The accuracy of the estimation results confirmed the importance of the spectral signatures calculated by PLS-DA, which provided data for crucial aspects of the differences in molecular components among the stages of myocardial infarction and its repair. The results showed that the label-free analytical method of Raman spectroscopy with PLSDA was effective for the accurate evaluation of myocardial infarction and its repair.



EXPERIMENTAL SECTION Sample Preparation. All animal experiments were conducted with the approval of and in accordance with guidelines from the Committee for Animal Research of the Kyoto Prefectural University of Medicine. All surgeries were performed under general anesthesia with sodium pentobarbital, and animal distress was kept to a minimum. Myocardial infarction was created in nine young adult Wistar rats (female, 8 weeks old, 160−180 g) by complete ligation of the left descending coronary artery for 2 days (3 rats), 5 days (4 rats), and 21 days (2 rats).21,22 Normal hearts (2 rats) and infarcted hearts at each postligation time point were excised and perfused in Tyrode’s solution to wash out blood. The excised hearts were immediately embedded in frozen section compound [FSC 22 (Leica, Wetzlar, Germany)], snap-frozen in 6904

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

well-known classic process. In rats, a little faster than in humans, the infarcted muscles appeared to be wavy and became increasingly eosinophilic after 24 h, with a loss of crossstriations, while the nuclei underwent fragmentation (Figure 1B). By days 4−7, macrophages, fibroblasts, and capillaries appeared at the margin of the infarction and migrated toward the center of the infarcted lesion, and granulation tissue formed (Figure 1C). Phagocytosis of necrotic cardiomyocytes and maturation of granulation tissue continued over the next several weeks. Typically, by the end of the third week, almost all of the necrotic cardiomyocytes had been resorbed and most infarct had been replaced by scar tissue (Figure 1D1,2).22,26 We obtained Raman spectra of infarcted heart tissues at each of these three time points. Raman Spectra of Normal and Infarcted Hearts. The spectra of the normal hearts had strong Raman bands at 643, 691, 750, 1130, 1314, and 1587 cm−1 (Figure 1E). The strong bands at 750, 1130, and 1587 cm−1 corresponded to the spectra of both reduced cytochrome c and reduced cytochrome b5.19 There were no strong bands at 643, 691, or 1314 cm−1 for cytochrome b5, but there were for cytochrome c.19 Therefore, the contribution of cytochrome c to the spectra of normal cardiac tissue seemed to be larger than that of cytochrome b5. In viable cardiomyocytes, electrons are always transported from an electron donor to an electron acceptor. Therefore, a certain number of oxidized cytochrome c molecules must exist. However, the spectra for normal cardiac tissues did not show prominent oxidized cytochrome c, because the spectrum of oxidized cytochrome c is much weaker than that of reduced cytochrome c.5,8,27 The peak position of the ν4 mode of heme also indicated a contribution of the reduced form of cytochrome c larger than that of the oxidized form. The spectra of reduced cytochromes have a peak of the ν4 mode at 1364 cm−1, whereas those of oxidized cytochromes have a peak at 1375 cm−1.28 The spectra of normal cardiac tissues had a peak of the ν4 mode at 1364 cm−1 (Figure 1E). Thus, we concluded that the main contributor to the spectra of normal cardiac tissue was reduced cytochrome c. The spectra of the necrotic tissue did not show the spectral features of reduced cytochrome c detailed above remarkably. They did not have the peaks at 643 and 691 cm−1 and had weaker than normal peaks at 750 and 1314 cm−1 [p < 0.001 (Figures 1E and 2)]. It may be partly because the amount of reduced cytochrome c is decreased or reduced cytochrome c is converted to other forms in necrotic tissues. In necrotic tissues, the plasma membranes of cardiomyocytes were injured and cellular components had leaked and been disrupted. Additionally, the ratio of reduced to oxidized cytochrome c would have changed because the process of electron transport was disrupted in the necrotic cardiomyocytes. The spectral differences between normal and necrotic tissues seemed to reflect these biological changes in cardiac tissues. The spectra of granulation tissues had peaks at 750 cm−1 [p < 0.001 (Figures 1E and 2A)] weaker than those of the normal hearts. The peak position at 1306 cm−1 for the granulation tissues was slightly different than that at 1314 cm−1 for the normal and necrotic tissues. The peaks are assigned to the ν4 mode of heme, and the peak at 1306 cm−1 is assigned to hemoglobin and the peak at 1314 cm−1 to cytochrome c.28 Although hemoglobin is thought to be removed from the vascular channels because of the perfusion of the hearts with Tyrode’s solution, it may sometimes be extravasated and trapped in white blood cells in the granulation tissues.

were illuminated with the line-shaped laser beam through a water-immersion objective lens [UPLSAPO 60×, NA = 1.2 (Olympus, Tokyo, Japan)], and one-dimensional Raman spectral images (one-dimensional space and Raman spectra) were simultaneously obtained with a thermoelectrically cooled CCD camera [Pixis 400BR, −70 °C, 400 pixels × 1340 pixels (Princeton Instruments, Trenton, NJ)]. Two-dimensional Raman images were obtained by scanning the line-shaped laser focus with 5 μm steps. The entrance slit width of the spectrometer installed in the Raman microscope was set to 100 μm. The irradiated laser intensity at the sample plane and exposure time for each line were 0.021−0.13 mW/μm2 and 10 s, respectively. Each spectrum was obtained by averaging in a square-shaped area (25 μm × 25 μm) of normal or infarcted heart tissues. Data Processing. Nanophoton software was utilized for preprocessing each individual Raman spectrum. All raw Raman spectra were calibrated using the known lines of ethanol. The broad fluorescence background was removed by fitting a modified least-squares fifth-order polynomial to the Raman spectra and subtracting this polynomial from the spectra.23 The intensity of each spectrum was normalized by the peak intensity of the Raman spectrum around 2935−2941 cm−1 because the bands existed in the spectra of all types of tissues and shapes of the bands are only slightly different among the tissue types in this study. All calculations in PLS-DA were performed with PLS toolbox (eigenvector Research, Wenatchee, WA) in MATLAB (Mathworks, Natick, MA). PLS-DA is widely used to build a quantitative model for classification in spectral analysis by projecting the predicted and observable variables into a new space to maximize the covariance between the response and independent variables.24 PLS-DA has been applied in some reported studies of Raman spectroscopy.12 In this study, we applied PLS-DA for accurate evaluation of heart tissues. We constructed a multivariate regression model with orthogonal condition and calculated the latent variable weights (LVs) and scores. We used ranges of Raman spectra from 706 to 765 cm−1 and from 1091 to 1700 cm−1 of each individual spectrum to exclude the effect of the quartz plate.25 This process was done prior to PLS-DA. For verifying the predictive accuracy of the regression model constructed by using PLS-DA, we employed leave-one-out cross validation analysis.11,12,20 The leave-one-out analysis also provided an estimate of the optimal number of LVs used in the regression model. For the comparison of the intensities of spectral peaks and the scores of PLS-DA among four tissue types, we employed Kruskal−Wallis analysis and a multiple-comparison test with a 99.9% confidence interval after analysis of variation in MATLAB for a test of the significant difference. For reconstruction of two-dimensional (2D) images in the estimation of tissue types by PLS-DA, we obtained averaged spectra in an area of 25 μm × 25 μm as a pixel of the 2D image. Tissue types were estimated on the basis of the spectrum of each pixel with the PLS-DA model that was established with the other rats as shown above. A 2D pseudocolor image representing tissue types was created using ImageJ (National Institutes of Health, Bethesda, MD).



RESULTS AND DISCUSSION A histological image of normal heart tissue composed of healthy cardiomyocytes is shown in Figure 1A. In general, after the development of ischemia, myocardial infarction occurs by a 6905

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

very related to these tissue types. Thus, LV1 seems to contribute to the difference between the spectra of the granulation and fibrotic tissues. LV1 shown in Figure 3A represented the bands at 1170, 1227, and 1306 cm−1 in a negative direction, which were assigned to the resonance Raman bands of heme (i.e., ν30, ν13, and ν4, respectively30), and these bands were detected in the spectra of granulation tissue. In particular, the band at 1306 cm−1 was found in the spectra of the granulation tissue specifically, as noted above. LV1 also represented the band at 1690 cm−1 in an upward direction. In the spectra of the fibrotic tissue, there was a broad Raman band around 1670 cm−1, which was assigned to amide I. This result indicated that the spectral features of the difference between fibrotic tissue and other tissues were apparent, especially at 1690 cm−1 in the broad Raman band of amide I. As a result, LV1 reflected the differences between the spectra of the granulation tissue characterized by the resonance Raman band of heme and those of the fibrotic tissue characterized by the Raman bands assigned to amide I. In the score plot of LV2, the normal tissue had the largest score and the necrotic tissue had the second-largest score in a positive direction. In contrast, the scores of the granulation and fibrotic tissues had negative values (Figure 4B). LV2 shown in Figure 3B included the peaks at 750, 1130, 1314, and 1587 cm−1, assigned to reduced cytochrome c,5,19 in an upward direction, and we found that the normal and necrotic tissues, the tissues derived from cardiomyocytes, could be characterized by the bands assigned to reduced cytochrome c. As noted above, the peak intensities of Raman bands at 750, 1130, and 1314 cm−1 in the necrotic tissue were smaller than those in the normal tissue (Figure 1E). Thus, the contribution of reduced cytochrome c represented as the Raman bands at 750, 1130, and 1314 cm−1 was smaller in the necrotic tissue than in the normal tissue, as indicated in the score plot of LV2. However, the intensity of the Raman peak at 1587 cm−1 in the spectra of necrotic tissues was not significantly smaller than that in normal cardiac tissues (Figure 1E). This suggests that other molecules may contribute to the formation of the spectra of the necrotic tissues. In addition to reduced cytochrome c, the molecules including a heme ring, such as oxidized cytochrome c, oxygenated myoglobin, and oxygenated hemoglobin, also possess a strong band at 1587 cm−1.8,19,27 Whether those molecules contribute to the spectra of the necrotic tissues is unclear at present. In granulation and fibrotic tissues, the negative scores of LV2 indicated that the contribution of the Raman bands assigned to reduced cytochrome c was significantly small. Cardiomyocytes are rich in mitochondria, and the amount of cytochrome c contained in mitochondria in the tissues composed of cardiomyocytes was greater than that in the granulation and fibrotic tissues. Furthermore, LV2 included the peaks at 1260, 1450, and 1679 cm−1, which were assigned to amide III, CH2 and CH3 deformation, and amide I, respectively, in a downward direction. These peaks were found in the spectra of fibrotic tissues, which were assigned to collagen type I. In the mean spectra of the granulation tissues, these bands were not apparent. However, in the observation of each individual spectrum of the granulation tissue, there were spectra that had small bands of the resonance Raman bands of heme and comparatively large bands derived from collagen type I (data not shown). In the granulation tissue, we surmised that the connective tissue was growing and the development was spatially inhomogeneous; that is, the variation of spectra was large. Thus, these bands in LV2, assigned to collagen type I,

Figure 2. Intensities of the Raman peaks at 750 cm−1 (A) and 1314 cm−1 (B). An asterisk means that there is a significant difference (p < 0.001) with Kruskal−Wallis analysis and a multiple-comparison test.

Furthermore, it is reported that hemoglobin is produced de novo in granulation tissue regardless of myelopoiesis.29 Hemoglobin has strong absorption at a wavelength of 532 nm, which greatly enhances the Raman signal due to the resonance Raman scattering effect, and may contribute to the spectra of granulation tissues in part. Of course, other substances may also contribute to the spectra of granulation tissues because granulation tissue consists of a wide variety of types of cells and components. The spectra on day 21 exhibited the features of collagen type I, which indicates fibrotic tissue, as we have previously described.19 Fibrotic tissue is richer in mature collagen than other tissue types (Figure 1).26 The spectra also showed the peak of CH3 stretching located at 2941 cm−1, while the other tissue types showed the peak at 2935 cm−1 (Figure 2B). This peak shift characterizes the spectra of fibrotic tissue, as we also noted previously.13 PLS-DA of the Raman Spectra. Figure 3 shows the LVs calculated with PLS-DA. The LVs were derived from the observed spectra via PLS-DA for discrimination of tissue types and provided important information about the differences among the spectra of the various tissue types. PLS-DA with leave-one-out cross validation employed six LVs for the building model, which is the most appropriate number as defined by the minimum of the root-mean-square error of cross validation. The LVs have implications for assignment of Raman bands when combined with scores of PLS-DA (Figure 4). In Figure 4A, the score of LV1 in the granulation tissue had a negative value and that of the fibrotic tissue had a positive value. The scores in normal and necrotic tissues were almost zero, and the information contained in LV1 did not seem to be 6906

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

Figure 3. Plots of LVs resulting from PLS-DA. Essential spectral features for diagnostic classification of tissues are shown. Contributions of LV1 (A), LV2 (B), LV3 (C), LV4 (D), LV5 (E), and LV6 (F) to Raman spectra are 43.7, 26.1, 17.1, 3.93, 2.19, and 1.14%, respectively.

also characterized the spectra of fibrotic tissues and a part of those of the granulation tissues. The score of LV3 had large standard deviations (SDs), but there were significant differences between the normal or necrotic tissues and the granulation or fibrotic tissues (Figure 4C). LV3 represented the peaks at 1170, 1306, 1375, and 1640 cm−1 in a downward direction (Figure 3C). These peaks were resonance Raman bands of heme and apparent in the spectra of the granulation tissues. In particular, the band at 1306 cm−1 was a specific band in the granulation tissue, as noted above, and LV3 seemed to reflect the differences between normal or necrotic tissues and granulation tissues. However, in the score plot of LV3, the granulation tissue and the fibrotic tissue were not significantly different, though the bands in LV3 were not apparent in the spectra of the fibrotic tissues. The reason for this is unclear. In our analysis of LV4, the importance of reduced cytochrome c as a characteristic of the spectra of normal tissues was emphasized again. The score of LV4 of the normal tissues was much larger than those of other tissue types (Figure 4D). In LV4, the peaks at 750 and 1314 cm−1 were apparent in

an upward direction (Figure 3D). They were assigned to reduced cytochrome c, as noted above. Thus, our study of LV4 suggested that the Raman spectra of normal cardiac tissue is characterized by reduced cytochrome c, and in particular, the bands at 750 and 1314 cm−1, assigned to ν15 and ν4, respectively, of the heme ring,30 are important to indicate the spectral features of the normal cardiac tissues. The score for fibrotic tissues was significantly larger than those for the necrotic and granulation tissues in the negative direction because LV4 includes the band at 1683 cm−1. In LV5, the spectral difference between the necrotic and granulation tissue was emphasized. In the score plot of LV5, the granulation tissue had a large positive value and the necrotic tissue had a large negative value (Figure 4E). LV5 included the peaks at 1170, 1314, and 1590 cm−1, assigned to heme resonance Raman bands (ν30, ν4, and ν2, respectively30) in a downward direction (Figure 3E). Thus, these bands might suggest the spectral features of the necrotic tissues. The bands at 1170 and 1590 cm−1 were found in the spectra of necrotic and granulation tissues, whereas the band at 1314 cm−1 was found only in the spectra of necrotic tissues (Figure 1E). As 6907

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

Figure 4. Plots of scores of individual LVs. Scores of LV1 (A), LV2 (B), LV3 (C), LV4 (D), LV5 (E), and LV6 (F). An asterisk denotes that there is significant difference (p < 0.001) with Kruskal−Wallis analysis and a multiple-comparison test.

noted above, the ν4 band of heme was at 1314 cm−1 in the spectra of necrotic tissues, reflecting the influence of cytochrome c; in contrast, it was at 1306 cm−1 in those of the granulation tissues influenced by hemoglobin. The scores of normal and fibrotic tissues were near zero, and these tissue types seemed not to be much affected by LV5. Thus, our analysis of LV5 suggested the spectral features of the necrotic tissues, and the band at 1314 cm−1 seems to be especially important for detecting the spectral difference between the necrotic and granulation tissues on the basis of molecular components. Although the scores of LV6 had large SDs, there were significant differences between the necrotic and normal or granulation tissues, and LV6 might reflect the spectral differences among them. In the LV6 plot, the bands at 1199

and 1273 cm−1 were apparent (Figure 3F). We could not derive any meaningful information about the spectral differences from the LV6 plot. Classification Capacity of PLS-DA. Table 1 shows the results of classification of Raman spectra derived from each cardiac tissue obtained with PLS-DA. High cross-validated sensitivities and specificities were obtained for the discrimination of normal, necrotic, granulation, and fibrotic tissue. These high sensitivities and specificities reveal that the spectral features indicated by LVs obtained by using PLS-DA provide important information about differences of the spectra acquired from individual tissue types. These results also demonstrate that the process of myocardial infarction and its repair can be evaluated by using the PLS-DA model on the basis of the constituent molecules with high accuracy. 6908

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

Table 1. Discrimination Results of Cardiac Tissues Obtained via PLS-DA Raman prediction tissue type

normal

necrosis

granulation

fibrosis

normal necrosis granulation fibrosis sensitivity (%) specificity (%)

149 3 2 0 99.3 99.4

0 286 4 0 95.3 99.5

1 7 482 13 96.4 96.5

0 4 12 137 91.3 98.3

Two-Dimensional Imaging of the Marginal Area of Infarction. A 2D image of the marginal area of infarction in a section of day 5 cardiac tissue is shown in Figure 5. In most of the imaged area, the determination of each tissue type obtained via PLS-DA is consistent with that obtained with the H&Estained serial section. In the margin between normal and granulation tissue, there is a small area determined to be necrosis with PLS-DA, while the area shows no apparent coagulation necrosis in the H&E-stained section. PLS-DA revealed the normal and necrotic tissues based chiefly on the Raman spectra derived from cytochrome c, as discussed above. In the marginal area, it is surmised that a certain amount of ischemic damage had been sustained. Actually, in the H&Estained section, inflammatory cells were found to have infiltrated to the marginal area of infarction. It is possible that degradation of the cardiomyocytes in the marginal area affected the Raman spectra. As a result, in the margin between the normal and granulation tissue, PLS-DA may reveal degradation of cardiomyocytes, which was not apparent in the H&E-stained section. Our results show that Raman spectroscopy with PLS-DA can potentially be used to examine myocardial infarction and its repair without staining. Precise evaluation of myocardial tissues is essential for treatments of ischemic hearts. Various techniques, such as magnetic resonance imaging, are now clinically available for the assessment of myocardial tissues. However, these modalities require contrast agents and have some limitations in real-time measurement and spatial resolution. A higher resolution is needed to quantitatively assess the spatial distribution of normal, necrotic, granulation, and fibrotic tissues, because they are mixed together in infarcted hearts. Raman spectroscopy can facilitate minimally invasive and real-time tissue diagnosis, as it does not necessitate chemical labeling. It can be incorporated into catheters or probes by using optical fibers.12,31−35 Although further analysis of human hearts under in vivo conditions is required, clear discrimination of normal tissues from infarcted tissues would allow a specific differentiation of myocardial tissues by Raman spectroscopy in the future. We anticipate that this method could be applicable as an aid during open cardiac surgery, especially in the case of surgical ventricular restoration (SVR). SVR is effective for refractory heart failure due to ischemia by excluding infarct legions and reducing the ventricular volume.36,37 For determining which area of the left ventricle should be excluded, intraoperative evaluation of myocardial infarction is essential.38 Although further studies are needed, we believe that this study will serve as a basis for the development of new medical equipment evaluating the myocardium noninvasively during cardiac surgery.

Figure 5. Two-dimensional image of a section of day 5 tissue (border area of normal, necrosis, and granulation tissues). (A) PLS-DA image. (B) H&E staining (B): (a) necrosis, (b) granulation, and (c) normal tissue. The scale bar is 50 μm.



CONCLUSION In this study, we investigated the Raman spectral properties of unfixed frozen cross sections of normal and infarcted rat cardiac tissues and observed significant differences in Raman spectra among normal, necrotic, granulation, and fibrotic tissues. Combined with the PLS-DA method, Raman spectroscopy successfully revealed tissue types of cardiac samples accurately 6909

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910

Analytical Chemistry

Article

and automatically. This is the first report demonstrating Raman spectra derived from sequential stages of myocardial infarction and its repair. By using LVs extracted with the PLS-DA, we clarified the differences in the spectra of individual tissue types. A 2D image of the tissue types in the marginal area, including normal, necrotic, and granulation tissues, was successfully obtained via PLS-DA. This label-free technique is a novel methodology for analysis of nonfixed tissue sections of myocardial infarction and during the repair process. After the resolution of several issues, we are hopeful that this intriguing methodology will be used for the accurate evaluation of the heart in vivo in the future.



(14) Ji, M.; Orringer, D. A.; Freudiger, C. W.; Ramkissoon, S.; Liu, X.; Lau, D.; Golby, A. J.; Norton, I.; Hayashi, M.; Agar, N. Y.; Young, G. S.; Spino, C.; Santagata, S.; Camelo-Piragua, S.; Ligon, K. L.; Sagher, O.; Xie, X. S. Sci. Transl. Med. 2013, 5, 201ra119. (15) Harada, Y.; Takamatsu, T. Curr. Pharm. Biotechnol. 2013, 14, 133−140. (16) Cals, F. L. J.; Bakker Schut, T. C.; Koljenović, S.; Puppels, G. J.; de Jong, R. J. B. J. Raman Spectrosc. 2013, 44, 963−972. (17) Kong, K.; Rowlands, C. J.; Varma, S.; Perkins, W.; Leach, I. H.; Koloydenko, A. A.; Williams, H. C.; Notingher, I. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 15189−15194. (18) Draga, R. O.; Grimbergen, M. C.; Vijverberg, P. L.; van Swol, C. F.; Jonges, T. G.; Kummer, J. A.; Ruud Bosch, J. L. Anal. Chem. 2010, 82, 5993−5999. (19) Ogawa, M.; Harada, Y.; Yamaoka, Y.; Fujita, K.; Yaku, H.; Takamatsu, T. Biochem. Biophys. Res. Commun. 2009, 382, 370−374. (20) Tan, Y.; Shi, L.; Tong, W.; Hwang, G. T.; Wang, C. Comput. Biol. Chem. 2004, 28, 235−244. (21) Matsushita, T.; Oyamada, M.; Fujimoto, K.; Yasuda, Y.; Masuda, S.; Wada, Y.; Oka, T.; Takamatsu, T. Circ. Res. 1999, 85, 1046−1055. (22) Takamatsu, T. Pathol. Int. 2008, 58, 533−543. (23) Lieber, C. A.; Mahadevan-Jansen, A. Appl. Spectrosc. 2003, 57, 1363−1367. (24) So, C. F.; Choi, K. S.; Chung, J. W.; Wong, T. K. Med. Eng. Phys. 2013, 35, 172−177. (25) Oshima, Y.; Shinzawa, H.; Takenaka, T.; Furihata, C.; Sato, H. J. Biomed. Opt. 2010, 15, 017009. (26) Kumar, V.; Abbas, A. K.; Fausto, N.; Aster, J. Robbins and Cotran Pathologic Basis of Disease, 8th ed.; Elsevier Saunders: Philadelphia, 2008. (27) Brazhe, N. A.; Treiman, M.; Faricelli, B.; Vestergaard, J. H.; Sosnovtseva, O. PLoS One 2013, 8, e70488. (28) Strekas, T. C.; Spiro, T. G. Biochim. Biophys. Acta 1972, 278, 188−192. (29) Tommila, M.; Stark, C.; Jokilammi, A.; Peltonen, V.; Penttinen, R.; Ekholm, E. J. Mol. Cell Biol. (Oxford, U.K.) 2011, 3, 190−196. (30) Desbois, A. Biochimie 1994, 76, 693−707. (31) Ogawa, M.; Harada, Y.; Yamaoka, Y.; Fujita, K.; Takamatsu, T. Proc. SPIE 2009, 7169, 71690H. (32) Soares, J. S.; Barman, I.; Dingari, N. C.; Volynskaya, Z.; Liu, W.; Klein, N.; Plecha, D.; Dasari, R. R.; Fitzmaurice, M. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 471−476. (33) Agenant, M.; Grimbergen, M.; Draga, R.; Marple, E.; Bosch, R.; van Swol, C. Biomed. Opt. Express 2014, 5, 1203−1216. (34) Bergholt, M. S.; Zheng, W.; Ho, K. Y.; Teh, M.; Yeoh, K. G.; So, J. B.; Shabbir, A.; Huang, Z. J. Biophotonics 2013, 6, 49−59. (35) Zavaleta, C. L.; Garai, E.; Liu, J. T.; Sensarn, S.; Mandella, M. J.; Van de Sompel, D.; Friedland, S.; Van Dam, J.; Contag, C. H.; Gambhir, S. S. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, E2288−E2297. (36) Buckberg, G.; Athanasuleas, C.; Conte, J. Nat. Rev. Cardiol. 2012, 9, 703−716. (37) Matsui, Y. Circ. J. 2009, 73 (Suppl. A), A13−A18. (38) Ogawa, M.; Doi, K.; Yamada, Y.; Fukumoto, A.; Okawa, K.; Kan’bara, T.; Koushi, K.; Itoh, H.; Nishimura, T.; Yaku, H. General Thoracic and Cardiovascular Surgery 2007, 55, 149−157.

AUTHOR INFORMATION

Corresponding Author

*Department of Pathology and Cell Regulation, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, 465 Kajii-cho Kawaramachi-Hirokoji, Kamigyo-ku, Kyoto 602-8566, Japan. Telephone: +81-75-251-5322. Fax: +81-75-251-5353. E-mail: [email protected]. Author Contributions

N.N-M., Y.H., and T.M. contributed equally to this work. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Dr. Mitsugu Ogawa (John Hunter Hospital, New South Wales, Australia) for his helpful discussion. A portion of this work was supported by the Adaptable and Seamless Technology Transfer Program through Target-driven R&D (AS2321506F) from the Japan Science and Technology Agency. T.M. acknowledges support from a Grant-in-Aid for JSPS Fellows (23-8134) from the Japan Society for the Promotion of Science (JSPS).



REFERENCES

(1) Raman, C. V.; Krishnan, K. S. Nature 1928, 121, 501−502. (2) Puppels, G. J.; de Mul, F. F.; Otto, C.; Greve, J.; Robert-Nicoud, M.; Arndt-Jovin, D. J.; Jovin, T. M. Nature 1990, 347, 301−303. (3) van Manen, H. J.; Lenferink, A.; Otto, C. Anal. Chem. 2008, 80, 9576−9582. (4) Harada, Y.; Dai, P.; Yamaoka, Y.; Ogawa, M.; Tanaka, H.; Nosaka, K.; Akaji, K.; Takamatsu, T. Histochem. Cell Biol. 2009, 132, 39−46. (5) Okada, M.; Smith, N. I.; Palonpon, A. F.; Endo, H.; Kawata, S.; Sodeoka, M.; Fujita, K. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 28−32. (6) Pascut, F. C.; Kalra, S.; George, V.; Welch, N.; Denning, C.; Notingher, I. Biochim. Biophys. Acta 2013, 1830, 3517−3524. (7) Inya-Agha, O.; Klauke, N.; Davies, T.; Smith, G.; Cooper, J. M. Anal. Chem. 2007, 79, 4581−4587. (8) Brazhe, N. A.; Treiman, M.; Brazhe, A. R.; Find, N. L.; Maksimov, G. V.; Sosnovtseva, O. V. PLoS One 2012, 7, e41990. (9) Haka, A. S.; Volynskaya, Z.; Gardecki, J. A.; Nazemi, J.; Lyons, J.; Hicks, D.; Fitzmaurice, M.; Dasari, R. R.; Crowe, J. P.; Feld, M. S. Cancer Res. 2006, 66, 3317−3322. (10) Chan, J. W.; Taylor, D. S.; Lane, S. M.; Zwerdling, T.; Tuscano, J.; Huser, T. Anal. Chem. 2008, 80, 2180−2187. (11) Magee, N. D.; Villaumie, J. S.; Marple, E. T.; Ennis, M.; Elborn, J. S.; McGarvey, J. J. J. Phys. Chem. B 2009, 113, 8137−8141. (12) Bergholt, M. S.; Zheng, W.; Lin, K.; Ho, K. Y.; Teh, M.; Yeoh, K. G.; So, J. B.; Huang, Z. Analyst 2010, 135, 3162−3168. (13) Minamikawa, T.; Harada, Y.; Koizumi, N.; Okihara, K.; Kamoi, K.; Yanagisawa, A.; Takamatsu, T. Histochem. Cell Biol. 2013, 139, 181−193. 6910

dx.doi.org/10.1021/ac500592y | Anal. Chem. 2014, 86, 6903−6910