- Email: [email protected]
Infrared imaging of compositional changes in inflammatory cardiomyopathy
Vibrational Spectroscopy 38 (2005) 217–222 www.elsevier.com/locate/vibspec
Infrared imaging of compositional changes in inflammatory cardiomyopathy Qi Wang a,1, Wasiem Sanad b,1, Lisa M. Miller a,*, Antje Voigt b, Karin Klingel c, Reinhard Kandolf c, Karl Stangl b, Gert Baumann b a
National Synchrotron Light Source, Brookhaven National Laboratory, Building 725D, Upton, NY 11973-5000, USA b Charite´ Medical School, Humboldt University, Berlin, Germany c Pathology Institute, University of Tu¨bingen, Tu¨bingen, Germany Accepted 14 February 2005 Available online 28 April 2005
Abstract Myocarditis is an inflammatory disorder that affects heart muscle cells and results in their dysfunction. The group B coxsackieviruses are the most common agents of myocarditis. In most cases the disease is self-limiting; however, idiopathic dilated cardiomyopathy may in many instances represent the end stage of an immunologically mediated disease initiated by an episode of enteroviral myocarditis. Studies indicated that inflammatory dilated cardiomyopathy accounts for about 25% of the cases of heart failure in the United States. Earlier studies on a cardiomyopathic Syrian hamster model have shown that extracellular matrix remodeling plays a significant role in the cardiac dysfunction [I.M. Dixon, H. Ju, N.L. Reid, T. Scammell-La Fleur, J.P. Werner, G. Jasmin, J. Mol. Cell. Cardiol. 29 (7) (1997) 1837–1850] and previous FTIR microspectroscopy studies demonstrated elevated fibrillar collagen in the pathological myocardium [K.M. Gough, D. Zelinski, R. Wiens, M. Rak, I.M. Dixon, Anal. Biochem. 316 (2) (2003) 232–242]. The aim of this study was to investigate the chemical composition of the myocardium in an immuno-resistant (C57BL/6) and an immuno-permissive (ABY/SnJ) mouse model of cardiomyopathy, using Fourier transform infrared imaging (FTIRI). Cross-sections (8 mm thick) of heart muscle tissue from murine CVB-models were acquired at 4, 8, and 28 days post-inoculation with the virus, mounted on infrared-reflective glass slides, and FTIRI was performed (n = 2 for each stage). Agematched controls were also examined (n = 3). The composition and distribution of proteins and lipids were obtained at a pixel resolution of 6.25 mm. Specifically, the lipid/protein ratio (area ratio of 3000–2800/1700–1600 cm 1) and the collagen content (correlation analysis from 1400 to 1000 cm 1) were determined. Results from both mouse models showed that the lipid/protein ratio decreased as the collagen content increased, suggesting extracellular matrix remodeling. In the permissive model (ABY/SnJ), the lipid/protein ratio decreased and the collagen content increased continuously as the disease progressed. However, in the resistant model (C57BL/6), the lipid/protein ratio decreased and the collagen content increased in the early stages of the disease (4 and 8 days), but then reversed in the later stages (28 days), suggesting a recovery from the disease. These results demonstrate chemical differences between the inflammatory responses in these two mouse models, providing insight into why the disease can be self-limiting in some cases while fatal in others. # 2005 Elsevier B.V. All rights reserved. Keywords: Myocarditis; Cardiomyopathy; Infrared imaging; Extracellular matrix remodeling
1. Introduction
Abbreviations: IR, infrared; FTIRI, Fourier transform infrared imaging; L/P, lipid/protein ratio; CC, collagen correlation; dpi, days postinfection * Corresponding author. Tel.: +1 631 344 2091; fax: +1 631 344 3238. E-mail address: [email protected] (L.M. Miller). 1 Both authors contributed equally to this work. 0924-2031/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.vibspec.2005.02.011
Heart failure, a common clinical burden in adult medicine, is associated with debilitating limiting symptoms, even with optimal modern medical management [1]. Heart (or cardiac) failure is the pathophysiological state in which the heart is unable to pump blood at a rate commensurate with the requirements of the metabolizing
218
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
tissues (or can so only from an abnormally elevated filling pressure) [2]. About one fourth of the congestive heart failure in the United States is due to idiopathic dilated cardiomyopathy [3]. Dilated cardiomyopathy remains the principal indication for cardiac transplantation in children worldwide. The prognosis for dilated cardiomyopathy is around 60% at five years from presentation with a high attrition within 6 months of presentation [4]. The cause(s) of dilated cardiomyopathy remains unclear. Familial, genetic, viral, and immunological abnormalities have been speculated [5,6]. Familial transmission is relatively frequent with the disease being genetically heterogeneous [7–12]. Dilated cardiomyopathy is also believed by some to be mediated by an autoimmune reaction following an episode of viral myocarditis [3]. Coxsackievirus B3 (CVB3), a member of picornavirus, has been recognized to be a highly active etiologic agent in human beings and causes different acute and chronic diseases, such as myocarditis, meningoencaphalitis, hepatitis and pancreatitis [13]. It is found that murine models of viral myocarditis induced by CVB3 usually develop a similar pattern of pathological parameters as humans do at different disease phases [14]. Like in humans, CVB3 infection in the mouse depends on the immunocompetence of the host. A permissive host, in which the virus can evade immunological surveillance, develops chronic myocarditis associated with CVB3 virus persistence, whereas a resistant host eliminates the virus during acute myocarditis and completely recovers from CVB3 infection [15]. The development of myocarditis involves a complex of events at the molecular and cellular level [16,17]. Among them, extracellular matrix remodeling plays a significant role in the disease, where studies have demonstrated elevated fibrillar collagen in the pathological myocardium of cardiomyopathic hamster models [18]. Fourier transform infrared imaging (FTIRI) is an established technique for providing spatially resolved information about different chemical components in biological tissue [19, and references therein]. The structure and composition of proteins, lipids, nucleic acids and different organic/inorganic components in various normal and pathological tissues have been probed precisely and rapidly with this technique [20–23]. Specifically, chemical changes including collagen composition have been detected with FTIRI in cardiac models of myocarditis [24,25]. In this study, FTIRI has been extended to two coxsackievirus B3-induced mouse models of myocarditis: chronic (ABY/SnJ) and acute (C57BL/6) forms of the disease. The protein (specifically collagen) and lipid content of the heart tissue were examined at various time points after infection in order to characterize chemical differences as a function of immunocompetence.
2. Materials and methods 2.1. Experimental animal models and tissue preparation All animal experiments were carried out in accordance with European and German legal and ethical regulations. cDNA generated CVB3 (Nancy strain) [26] was grown and propagated in Vero cells (African green monkey kidney cells). Stock virus was prepared by three times freezing and thawing and further purified by sucrose gradient centrifugation. Nine 4–5 week-old inbred mice of the strains C57BL/6 and ABY/SnJ (both H-2b) were infected intraperitoneally (i.p.) with 1 105 PFU of purified CVB3. Animals were sacrificed at 4, 8, 28 days post-infection; non-infected animals of the two mouse strains were used as controls. Hearts were removed and frozen in liquid nitrogen. For analysis, 8 mm thick-heart tissue sections were mounted on MirrIR reflective microscope slides (Kevley Technologies, Chesterland, OH) and fixed for 20 min at 4 8C in 2% paraformaldehyde before use, which has been shown to preserve cellular components for FTIR microspectroscopy [27]. For each mouse model (ABY/SnJ and C57BL/6), nine animals were used; three were used as controls and six were infected with coxsackievirus B3. 2.2. FTIR imaging data collection IR images were collected using a Perkin Elmer Spectrum Spotlight FTIR Imaging System using the ‘‘image’’ mode of the instrument. For each tissue section, the entire area of the tissue was defined (approximately 5 mm in diameter) and an IR image was produced for this area using a liquid nitrogen cooled, 16-pixel mercury cadmium telluride (MCT-A) array detector at a 25 mm pixel resolution. Typically, 5000–10,000 spectra were collected for each tissue section. In addition, two or three smaller images were collected from each section at higher pixel resolution (6.25 mm). For the highresolution images, over 100,000 spectra were collected for each tissue section. Using a 25 mm pixel size (4 cm 1 resolution, 4 scans/pt.), a 4 mm 4 mm area of tissue took 110 min to image. Using a 6.25 mm pixel size (4 cm 1 resolution, 4 scans/pt.), imaging the same area of tissue would take approximately 29 h to collect. An absorbance spectrum was recorded for each pixel in trans-reflectance mode. Background spectra were collected from a clean region of the IR-reflective slide. All spectra were collected in the mid-infrared spectral region (4000– 720 cm 1). The spectral resolution was 4 cm 1 and eight scans were averaged for each spectrum. A visible image of the sample was also collected. A type I collagen sample (from bovine achilles tendon, 4 mg/ml in 0.02 M acetic acid [28]) was prepared on a MirrIR microscope slide and was used for a collagen reference spectrum. The spectrum was recorded using ‘‘point’’ mode of the Spectrum Spotlight with an aperture of 50 mm 50 mm, and a spectral resolution of 4 cm 1 in the mid IR spectral
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
region (4000–720 cm 1). The absorbance spectrum was acquired by co-adding 1024 scans in trans-reflectance mode. 2.3. FTIR imaging data evaluation and analysis Three FTIR spectral regions were used in the data evaluation (Fig. 1): (1) lipid content was determined by integrating the area between 3000 and 2800 cm 1 (linear baseline: 3000–2800 cm 1), which represents the C–H stretching region and is dominated by the long hydrocarbon chains of lipids; (2) protein content was determined by integrating the amide I band, 1700–1600 cm 1 (single-point linear baseline at 1800 cm 1), which is assigned to the carbonyl stretching mode of the protein amide backbone (–NH–C O); (3) collagen content was determined through correlation analysis of the region from 1400 to 1000 cm 1, which has a series of absorbance features unique to collagen [25]. The correlation analysis was performed against the reference spectrum of pure type I collagen, using the second derivative. Accordingly, two parameters were established to evaluate the FTIR images in this study: the lipid/protein (L/P) ratio and the collagen content. Images were plotted with a rainbow color scale where lowest values were blue and highest values were red. All images for each parameter (L/P ratio and collagen content) were plotted on the same scale for comparison.
3. Results and discussion Enteroviruses have been shown to induce an acute inflammation of the myocardium without cardiac dysfunction (i.e. myocarditis) or with cardiac dysfunction (i.e. inflammatory cardiomyopathy) that can transform to virus-negative cardiomyopathy. The three phases generally involved in this disease are: early, acute, and chronic stages. Extracellular
219
matrix remodeling plays a significant role in cardiac dysfunction. Lipid replacement by collagen deposition has been recognized as one of major indicators in the progression of cardiac dysfunction [29,30] and infrared imaging can be used to detect both collagen and lipid in the disease [28]. Fig. 1A shows a typical IR absorption spectrum of heart muscle tissue. The IR spectrum can be separated into three major regions related to this work. The peaks between 3000 and 2800 cm 1 are assigned to C–H stretching modes. Specifically, the asymmetric C–H stretching vibrations of the –CH3 and CH2 fall at 2959 and 2921 cm 1, respectively, while the symmetric modes occur at 2872 and 2852 cm 1, respectively [31]. This region is a sensitive indicator for lipid content. The absorbance peak centered near 1650 cm 1 originates from C O stretching vibration of the amide bond backbone in proteins [31]. The exact frequency of this mode depends on the secondary structure of the protein being examined. This peak was used to estimate the concentration of protein in tissue. Collagen content was determined by examining the fingerprint region from 1400 to 1000 cm 1. The spectrum of collagen can be seen in Fig. 1B. Collagen absorption features at 1338, 1283, 1240, 1204, 1082 and 1032 cm 1 are attributed to CH2 wagging, CH3 deformation, C–N stretching and C–OH stretching of collagen [24,25,28,32,33]. Fig. 2 shows FTIR images for the permissive mouse strain (ABY/SnJ) at different disease stages: early (4 dpi), acute (8 dpi) and chronic (28 dpi). The collagen distribution can be seen in Fig. 2A, as determined by the collagen correlation from 1400 to 1000 cm 1. All images are plotted on the same scale. Results show that the diseased tissue has a higher correlation with the pure collagen reference spectrum, indicating that collagen levels are elevated in this tissue compared to the control tissue. Specifically, the collagen correlation coefficients are 0.416 0.011, 0.437 0.003, and 0.415 0.002, in the early, acute, and
Fig. 1. (A) Infrared spectrum from heart tissue. The lipid/protein (L/P) band ratio is derived from the integrated area ratio between lipid (3000–2800 cm 1) and protein (1700–1600 cm 1). (B) Infrared spectrum of type I collagen. The second derivative of this spectrum was used as a reference spectrum for correlation comparison of collagen content in samples. The insert indicates the spectral region for correlation, i.e. the fingerprint region (1400–1000 cm 1).
220
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
Fig. 2. (A) Collagen correlation images and (B) lipid/protein ratio images of heart tissue from the ABY/SnJ mouse model at different disease stages: control, 4 days, 8 days, and 28 days post-infection. The collagen correlation was calculated using the second derivative spectrum of Type I collagen from 1400 to 1000 cm 1. Data collection conditions: pixel size = 25 mm, spectral resolution = 4 cm 1, scan number = 8 scans/pixel. The scale bar represents 1.0 mm.
chronic stages, respectively, while it is 0.329 0.017 in control tissue. Both correlation maps and the coefficient averages indicate there is no significant difference among three disease stages but a 26% increase was found in the collagen correlation for the diseased tissue relative to the control. In addition, the FTIR images show that there is a uniform distribution of collagen deposition in the diseased tissue; it is not a localized effect. This is supported by low standard deviations for the correlation coefficients (Table 1). The L/P ratio is seen in Fig. 2B. As can be seen, a high L/ P ratio exists for the control tissue (0.193 0.017), and the L/P ratio is more than three times lower in all three stages of infected tissue: 0.067 0.010, 0.078 0.001, and 0.062 0.001 for early, acute and chronic stages, respectively. No significant differences were observed between the three diseased stages. Since these data are reported as ratios to account for any sample thickness variations, these results could indicate that the diseased states possess either a higher protein content and/or a lower lipid level. However, the increased collagen content is observed in the diseased stages, strongly suggesting that protein content is increasing.
In addition, the CH2/CH3 ratio can be applied to indicate lipid content, where a higher ratio indicates longer-chained lipids and a higher lipid content [28]. In our studies, the infected tissue had comparable CH2/CH3 ratios as the control tissue (data not shown), supporting the conclusion that the lower L/P ratio in the diseased states comes from elevated protein (i.e. collagen) and not a significant reduction in lipid content. For the C57BL/6 strain, the collagen distribution and L/P ratio images are shown in Fig. 3A and B, respectively, plotted on the same color scales as in Fig. 2A and B. No differences were observed in collagen correlation between the control (0.380 0.006), 4 dpi (0.401 0.010), and 8 dpi (0.392 0.006) tissue. Similarly, the L/P ratio did not differ significantly between the control (0.078 0.010), 4 dpi (0.087 0.032), and 8 dpi (0.062 0.005) tissue. However, a dramatic reduction in collagen content (0.294 0.080) and increased L/P ratio (0.211 0.053) were observed in the 28 dpi tissue. These results suggest that C57BL/6 strain can recover from the CVB3 infection during the acute phase of disease. In addition, the original (control)
Table 1 Lipid/protein ratio (L/P) and collagen correlation (CC): mean S.E.M. Disease stage
Control
4 days
8 days
28 days
ABY/SnJ CC L/P
0.329 0.017 0.193 0.017
0.416 0.011 0.067 0.010
0.437 0.003 0.078 0.001
0.415 0.002 0.062 0.001
C57BL/6 CC L/P
0.380 0.006 0.078 0.010
0.401 0.010 0.087 0.032
0.392 0.006 0.062 0.005
0.294 0.080 0.211 0.053
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
221
Fig. 3. (A) Collagen correlation images and (B) lipid/protein band ratio maps of heart tissue from the C57BL/6 mouse models at different disease stages: control, 4 days, 8 days, and 28 days post-infection. The collagen correlation was calculated using the second derivative spectrum of type I collagen from 1400 to 1000 cm 1. Data collection conditions: pixel size = 25mm, spectral resolution = 4 cm 1, scan number = 8 scans/pixel. The scale bar represents 1.0 mm.
composition of the C57BL/6 strain differs from the ABY/ SnJ strain; the C57BL/6 strain has a higher collagen content and lower L/P ratio than the ABY/SnJ strain. These differences may influence the ability of the C57BL/6 strain to stabilize the disease and initiate recovery. For both the ABY/SnJ and C57BL/6 strains, a strong inverse correlation was observed between the collagen content and the L/P ratio (Fig. 4). This is reflected in the negative slopes for the two linear relationships: 0.70 (ABY/SnJ) and 0.73 (C57BL/6), plotted on a scale from 0 to 1, representing the lowest to highest correlation. The linear regressions, r2, of the two lines
were 0.82 (ABY/SnJ) and 0.64 (C57BL/6). Thus, as collagen concentration increases, the L/P ratio decreases, suggesting that extracellular matrix remodeling, i.e. collagen deposition, replaces normal tissue as the CVB3 infection induces myocarditis. These findings are consistent with numerous previous studies on CVB3 viral myocarditis using murine models. Specifically, ABY/ SnJ(H-2b) is classified as a permissive host, prone to sustaining chronic inflammation of myocarditis by CVB3infection; C57BL/6(H-2b) is a resistant host, which usually eliminates CVB3 during acute myocarditis, preventing the development of chronic myocarditis [34].
Fig. 4. Linear correlation between the lipid/protein ratio and the collagen correlation in the (left) ABY/SnJ and (right) C57BL/6 strains. Means and S.E.M.s were calculated using data collected at a pixel size of 6.25 mm, spectral resolution = 4 cm 1, scan number = 8 scans/pixel.
222
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
4. Conclusions and future work Collagen fibrosis has been established as a major contributor to cardiac damage as cardiomyopathy develops. Therefore, monitoring the distribution and concentration of collagen during the different stages of coxasackievirus B3-induced viral myocarditis in different immunocompetent mouse strains can help thoroughly understand the progression of disease. In this work, we demonstrated that collagen content increases and L/P ratio decreases in both the permissive (ABY/SnJ) and resistant (C57BL/6) strains of the CVB3-infected mice, indicating extracellular matrix remodeling in the disease. This effect is not localized, but broadly distributed within the tissue. However, in the chronic stages of the disease, these changes only persist in the permissive ABY/SnJ strain, consistent with the postulation that CVB3 infection induces two different immune responses in affected immunocompetent hosts; the resistant host recovers from myocarditis whereas the permissive host develops chronic disease. The FTIRI results also show that the two strains have different compositions in the uninfected state, which may also provide an insight into the mechanism for the different responses. In short, FTIRI brings a new dimension to understanding the composition of heart tissue in myocarditis, and has the potential to become a reliable tool for monitoring the disease in situ. This study represents an initial application of the technique, but further studies involving a much larger number of cases will be necessary for validation. Future studies will involve subcellular imaging of tissue composition with synchrotron infrared imaging, combined with immunohistochemistry and echocardiography studies. Acknowledgements The NSLS is supported by the United States Department of Energy under contract no. DE-AC02-98CH10886. Charite´ and Tu¨ bingen Universities are supported by the Deutsche Forschungsgemeinschaft (DFG), project no. 2005096. References [1] M. Bursch, Heart 88 (2002) 198–202. [2] W.S. Collucci, E. Braunwald, in: Braunwald, E. (Ed.), Heart Disease, A Textbook of Cardiovascular Medicine, sixth ed., W.B. Saunders Company, Philadelphia, 2001, p. 503. [3] C.A. Brown, J.B. O’Connell, Am. J. Med. 99 (3) (1995) 309–314. [4] M. Burch, S.A. Siddiqi, D.S. Celermajer, C. Scott, C. Bull, J.E. Deanfield, Br. Heart J. 72 (3) (1994) 246–250.
[5] G.W. Dec, V. Fuster, N. Engl. J. Med. 331 (23) (1994) 1564– 1575. [6] D.A. Neumann, Mayo Clin. Proc. 69 (2) (1994) 193–195. [7] T.A. Manolio, K.L. Baughman, R. Rodeheffer, T.A. Pearson, J.D. Bristow, V.V. Michels, W.H. Abelmann, W.R. Harlan, Am. J. Cardiol. 69 (17) (1992) 1458–1466. [8] M.K. Baig, J.H. Goldman, A.L. Caforio, A.S. Coonar, P.J. Keeling, W.J. McKenna, J. Am. College Cardiol. 31 (1) (1998) 195–201. [9] E. Gre`unig, J.A. Tasman, H. Ke`ucherer, W. Franz, W. Ke`ubler, H.A. Katus, J. Am. College Cardiol. 31 (1) (1998) 186–194. [10] C.J. McKenna, M.B. Codd, H.A. McCann, D.D. Sugrue, Heart (Br. Cardiac Soc.) 77 (6) (1997) 549–552. [11] K.R. Bowles, R. Gajarski, P. Porter, V. Goytia, L. Bachinski, R. Roberts, R. Pignatelli, J.A. Towbin, J. Clin. Invest. 98 (6) (1996) 1355–1360. [12] J.B. Durand, L.L. Bachinski, L.C. Bieling, G.Z. Czernuszewicz, A.B. Abchee, Q.T. Yu, T. Tapscott, R. Hill, J. Ifegwu, A.J. Marian, Circulation 92 (12) (1995) 3387–3389. [13] K. Klingel, S. Stephan, M. Sauter, R. Zell, B.M. McManus, B. Be`ultmann, R. Kandolf, J. Virol. 70 (12) (1996) 8888–8895. [14] C. Gauntt, S. Huber, Front. Biosci.: J. Virt. Libr. 8 (2003) 23–35. [15] D.R. Anderson, J.E. Wilson, C.M. Carthy, D. Yang, R. Kandolf, B.M. McManus, J. Virol. 70 (7) (1996) 4632–4645. [16] W.S. Colocci, Am. J. Cardiol. 80 (1997) 15L–25L. [17] A.M. Feldman, P.E. Ray, C.M. Silan, J.A. Mercer, W. Minobe, M.R. Bristow, Circulation 83 (6) (1991) 1866–1872. [18] I.M. Dixon, H. Ju, N.L. Reid, T. Scammell-La Fleur, J.P. Werner, G. Jasmin, J. Mol. Cell. Cardiol. 29 (7) (1997) 1837–1850. [19] D.L. Wetzel, S.M. LeVine, Science 285 (5431) (1999) 1224– 1225. [20] P. Dumas, L.M. Miller, Vibrat. Spectrosc. 32 (2003) 3–21. [21] P. Lasch, W. Haensch, D. Naumann, M. Diem, Biochim. Biophys. Acta 1688 (2) (2004) 176–186. [22] J. Kneipp, P. Lasch, E. Baldauf, M. Beekes, D. Naumann, Biochim. Biophys. Acta 1501 (2–3) (2000) 189–199. [23] R.Y. Huang, L.M. Miller, C.S. Carlson, M.R. Chance, Bone 33 (4) (2003) 514–521. [24] K. Liu, M. Jackson, M.G. Sowa, H. Ju, I.M. Dixon, H.H. Mantsch, Biochim. Biophys. Acta 1315 (2) (1996) 73–77. [25] K.Z. Liu, I.M. Dixon, H.H. Mantsch, Cardiovasc. Pathol.: Off. J. Soc. Cardiovasc. Pathol. 8 (1) (1999) 41–47. [26] R. Kandolf, P.H. Hofschneider, Proc. Natl. Acad. Sci. U.S.A. 82 (14) (1985) 4818–4822. [27] E. Gazi, J. Dwyer, N.P. Lockyer, J. Miyan, P. Gardner, C. Hart, M. Brown, N.W. Clarke, Biopolymers 77 (2005) 18–30. [28] K.M. Gough, D. Zelinski, R. Wiens, M. Rak, I.M. Dixon, Anal. Biochem. 316 (2) (2003) 232–242. [29] J.E. Strobeck, S.M. Factor, A. Bhan, M. Sole, C.C. Liew, F. Fein, E.H. Sonnenblick, Ann. N.Y. Acad. Sci. 317 (1979) 59–88. [30] G. Davison, C.S. Hall, J.G. Miller, M. Scott, S.A. Wickline, Circulation 90 (3) (1994) 1334–1342. [31] B. Stuart, in: D.J. Ando (Ed.), Biological Applications of Infrared Spectroscopy, John Wiley and Sons, Inc., Chichester, 1997, pp. 113– 153. [32] N.P. Camacho, P. West, P.A. Torzilli, R. Mendelsohn, Biopolymers 62 (1) (2001) 1–8. [33] M. Jackson, M.G. Sowa, H.H. Mantsch, Biophys. Chem. 68 (1–3) (1997) 109–125. [34] K. Klingel, J.J. Schnorr, M. Sauter, G. Szalay, R. Kandolf, Am. J. Pathol. 162 (5) (2003) 1709–1720.
Infrared imaging of compositional changes in inflammatory cardiomyopathy Qi Wang a,1, Wasiem Sanad b,1, Lisa M. Miller a,*, Antje Voigt b, Karin Klingel c, Reinhard Kandolf c, Karl Stangl b, Gert Baumann b a
National Synchrotron Light Source, Brookhaven National Laboratory, Building 725D, Upton, NY 11973-5000, USA b Charite´ Medical School, Humboldt University, Berlin, Germany c Pathology Institute, University of Tu¨bingen, Tu¨bingen, Germany Accepted 14 February 2005 Available online 28 April 2005
Abstract Myocarditis is an inflammatory disorder that affects heart muscle cells and results in their dysfunction. The group B coxsackieviruses are the most common agents of myocarditis. In most cases the disease is self-limiting; however, idiopathic dilated cardiomyopathy may in many instances represent the end stage of an immunologically mediated disease initiated by an episode of enteroviral myocarditis. Studies indicated that inflammatory dilated cardiomyopathy accounts for about 25% of the cases of heart failure in the United States. Earlier studies on a cardiomyopathic Syrian hamster model have shown that extracellular matrix remodeling plays a significant role in the cardiac dysfunction [I.M. Dixon, H. Ju, N.L. Reid, T. Scammell-La Fleur, J.P. Werner, G. Jasmin, J. Mol. Cell. Cardiol. 29 (7) (1997) 1837–1850] and previous FTIR microspectroscopy studies demonstrated elevated fibrillar collagen in the pathological myocardium [K.M. Gough, D. Zelinski, R. Wiens, M. Rak, I.M. Dixon, Anal. Biochem. 316 (2) (2003) 232–242]. The aim of this study was to investigate the chemical composition of the myocardium in an immuno-resistant (C57BL/6) and an immuno-permissive (ABY/SnJ) mouse model of cardiomyopathy, using Fourier transform infrared imaging (FTIRI). Cross-sections (8 mm thick) of heart muscle tissue from murine CVB-models were acquired at 4, 8, and 28 days post-inoculation with the virus, mounted on infrared-reflective glass slides, and FTIRI was performed (n = 2 for each stage). Agematched controls were also examined (n = 3). The composition and distribution of proteins and lipids were obtained at a pixel resolution of 6.25 mm. Specifically, the lipid/protein ratio (area ratio of 3000–2800/1700–1600 cm 1) and the collagen content (correlation analysis from 1400 to 1000 cm 1) were determined. Results from both mouse models showed that the lipid/protein ratio decreased as the collagen content increased, suggesting extracellular matrix remodeling. In the permissive model (ABY/SnJ), the lipid/protein ratio decreased and the collagen content increased continuously as the disease progressed. However, in the resistant model (C57BL/6), the lipid/protein ratio decreased and the collagen content increased in the early stages of the disease (4 and 8 days), but then reversed in the later stages (28 days), suggesting a recovery from the disease. These results demonstrate chemical differences between the inflammatory responses in these two mouse models, providing insight into why the disease can be self-limiting in some cases while fatal in others. # 2005 Elsevier B.V. All rights reserved. Keywords: Myocarditis; Cardiomyopathy; Infrared imaging; Extracellular matrix remodeling
1. Introduction
Abbreviations: IR, infrared; FTIRI, Fourier transform infrared imaging; L/P, lipid/protein ratio; CC, collagen correlation; dpi, days postinfection * Corresponding author. Tel.: +1 631 344 2091; fax: +1 631 344 3238. E-mail address: [email protected] (L.M. Miller). 1 Both authors contributed equally to this work. 0924-2031/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.vibspec.2005.02.011
Heart failure, a common clinical burden in adult medicine, is associated with debilitating limiting symptoms, even with optimal modern medical management [1]. Heart (or cardiac) failure is the pathophysiological state in which the heart is unable to pump blood at a rate commensurate with the requirements of the metabolizing
218
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
tissues (or can so only from an abnormally elevated filling pressure) [2]. About one fourth of the congestive heart failure in the United States is due to idiopathic dilated cardiomyopathy [3]. Dilated cardiomyopathy remains the principal indication for cardiac transplantation in children worldwide. The prognosis for dilated cardiomyopathy is around 60% at five years from presentation with a high attrition within 6 months of presentation [4]. The cause(s) of dilated cardiomyopathy remains unclear. Familial, genetic, viral, and immunological abnormalities have been speculated [5,6]. Familial transmission is relatively frequent with the disease being genetically heterogeneous [7–12]. Dilated cardiomyopathy is also believed by some to be mediated by an autoimmune reaction following an episode of viral myocarditis [3]. Coxsackievirus B3 (CVB3), a member of picornavirus, has been recognized to be a highly active etiologic agent in human beings and causes different acute and chronic diseases, such as myocarditis, meningoencaphalitis, hepatitis and pancreatitis [13]. It is found that murine models of viral myocarditis induced by CVB3 usually develop a similar pattern of pathological parameters as humans do at different disease phases [14]. Like in humans, CVB3 infection in the mouse depends on the immunocompetence of the host. A permissive host, in which the virus can evade immunological surveillance, develops chronic myocarditis associated with CVB3 virus persistence, whereas a resistant host eliminates the virus during acute myocarditis and completely recovers from CVB3 infection [15]. The development of myocarditis involves a complex of events at the molecular and cellular level [16,17]. Among them, extracellular matrix remodeling plays a significant role in the disease, where studies have demonstrated elevated fibrillar collagen in the pathological myocardium of cardiomyopathic hamster models [18]. Fourier transform infrared imaging (FTIRI) is an established technique for providing spatially resolved information about different chemical components in biological tissue [19, and references therein]. The structure and composition of proteins, lipids, nucleic acids and different organic/inorganic components in various normal and pathological tissues have been probed precisely and rapidly with this technique [20–23]. Specifically, chemical changes including collagen composition have been detected with FTIRI in cardiac models of myocarditis [24,25]. In this study, FTIRI has been extended to two coxsackievirus B3-induced mouse models of myocarditis: chronic (ABY/SnJ) and acute (C57BL/6) forms of the disease. The protein (specifically collagen) and lipid content of the heart tissue were examined at various time points after infection in order to characterize chemical differences as a function of immunocompetence.
2. Materials and methods 2.1. Experimental animal models and tissue preparation All animal experiments were carried out in accordance with European and German legal and ethical regulations. cDNA generated CVB3 (Nancy strain) [26] was grown and propagated in Vero cells (African green monkey kidney cells). Stock virus was prepared by three times freezing and thawing and further purified by sucrose gradient centrifugation. Nine 4–5 week-old inbred mice of the strains C57BL/6 and ABY/SnJ (both H-2b) were infected intraperitoneally (i.p.) with 1 105 PFU of purified CVB3. Animals were sacrificed at 4, 8, 28 days post-infection; non-infected animals of the two mouse strains were used as controls. Hearts were removed and frozen in liquid nitrogen. For analysis, 8 mm thick-heart tissue sections were mounted on MirrIR reflective microscope slides (Kevley Technologies, Chesterland, OH) and fixed for 20 min at 4 8C in 2% paraformaldehyde before use, which has been shown to preserve cellular components for FTIR microspectroscopy [27]. For each mouse model (ABY/SnJ and C57BL/6), nine animals were used; three were used as controls and six were infected with coxsackievirus B3. 2.2. FTIR imaging data collection IR images were collected using a Perkin Elmer Spectrum Spotlight FTIR Imaging System using the ‘‘image’’ mode of the instrument. For each tissue section, the entire area of the tissue was defined (approximately 5 mm in diameter) and an IR image was produced for this area using a liquid nitrogen cooled, 16-pixel mercury cadmium telluride (MCT-A) array detector at a 25 mm pixel resolution. Typically, 5000–10,000 spectra were collected for each tissue section. In addition, two or three smaller images were collected from each section at higher pixel resolution (6.25 mm). For the highresolution images, over 100,000 spectra were collected for each tissue section. Using a 25 mm pixel size (4 cm 1 resolution, 4 scans/pt.), a 4 mm 4 mm area of tissue took 110 min to image. Using a 6.25 mm pixel size (4 cm 1 resolution, 4 scans/pt.), imaging the same area of tissue would take approximately 29 h to collect. An absorbance spectrum was recorded for each pixel in trans-reflectance mode. Background spectra were collected from a clean region of the IR-reflective slide. All spectra were collected in the mid-infrared spectral region (4000– 720 cm 1). The spectral resolution was 4 cm 1 and eight scans were averaged for each spectrum. A visible image of the sample was also collected. A type I collagen sample (from bovine achilles tendon, 4 mg/ml in 0.02 M acetic acid [28]) was prepared on a MirrIR microscope slide and was used for a collagen reference spectrum. The spectrum was recorded using ‘‘point’’ mode of the Spectrum Spotlight with an aperture of 50 mm 50 mm, and a spectral resolution of 4 cm 1 in the mid IR spectral
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
region (4000–720 cm 1). The absorbance spectrum was acquired by co-adding 1024 scans in trans-reflectance mode. 2.3. FTIR imaging data evaluation and analysis Three FTIR spectral regions were used in the data evaluation (Fig. 1): (1) lipid content was determined by integrating the area between 3000 and 2800 cm 1 (linear baseline: 3000–2800 cm 1), which represents the C–H stretching region and is dominated by the long hydrocarbon chains of lipids; (2) protein content was determined by integrating the amide I band, 1700–1600 cm 1 (single-point linear baseline at 1800 cm 1), which is assigned to the carbonyl stretching mode of the protein amide backbone (–NH–C O); (3) collagen content was determined through correlation analysis of the region from 1400 to 1000 cm 1, which has a series of absorbance features unique to collagen [25]. The correlation analysis was performed against the reference spectrum of pure type I collagen, using the second derivative. Accordingly, two parameters were established to evaluate the FTIR images in this study: the lipid/protein (L/P) ratio and the collagen content. Images were plotted with a rainbow color scale where lowest values were blue and highest values were red. All images for each parameter (L/P ratio and collagen content) were plotted on the same scale for comparison.
3. Results and discussion Enteroviruses have been shown to induce an acute inflammation of the myocardium without cardiac dysfunction (i.e. myocarditis) or with cardiac dysfunction (i.e. inflammatory cardiomyopathy) that can transform to virus-negative cardiomyopathy. The three phases generally involved in this disease are: early, acute, and chronic stages. Extracellular
219
matrix remodeling plays a significant role in cardiac dysfunction. Lipid replacement by collagen deposition has been recognized as one of major indicators in the progression of cardiac dysfunction [29,30] and infrared imaging can be used to detect both collagen and lipid in the disease [28]. Fig. 1A shows a typical IR absorption spectrum of heart muscle tissue. The IR spectrum can be separated into three major regions related to this work. The peaks between 3000 and 2800 cm 1 are assigned to C–H stretching modes. Specifically, the asymmetric C–H stretching vibrations of the –CH3 and CH2 fall at 2959 and 2921 cm 1, respectively, while the symmetric modes occur at 2872 and 2852 cm 1, respectively [31]. This region is a sensitive indicator for lipid content. The absorbance peak centered near 1650 cm 1 originates from C O stretching vibration of the amide bond backbone in proteins [31]. The exact frequency of this mode depends on the secondary structure of the protein being examined. This peak was used to estimate the concentration of protein in tissue. Collagen content was determined by examining the fingerprint region from 1400 to 1000 cm 1. The spectrum of collagen can be seen in Fig. 1B. Collagen absorption features at 1338, 1283, 1240, 1204, 1082 and 1032 cm 1 are attributed to CH2 wagging, CH3 deformation, C–N stretching and C–OH stretching of collagen [24,25,28,32,33]. Fig. 2 shows FTIR images for the permissive mouse strain (ABY/SnJ) at different disease stages: early (4 dpi), acute (8 dpi) and chronic (28 dpi). The collagen distribution can be seen in Fig. 2A, as determined by the collagen correlation from 1400 to 1000 cm 1. All images are plotted on the same scale. Results show that the diseased tissue has a higher correlation with the pure collagen reference spectrum, indicating that collagen levels are elevated in this tissue compared to the control tissue. Specifically, the collagen correlation coefficients are 0.416 0.011, 0.437 0.003, and 0.415 0.002, in the early, acute, and
Fig. 1. (A) Infrared spectrum from heart tissue. The lipid/protein (L/P) band ratio is derived from the integrated area ratio between lipid (3000–2800 cm 1) and protein (1700–1600 cm 1). (B) Infrared spectrum of type I collagen. The second derivative of this spectrum was used as a reference spectrum for correlation comparison of collagen content in samples. The insert indicates the spectral region for correlation, i.e. the fingerprint region (1400–1000 cm 1).
220
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
Fig. 2. (A) Collagen correlation images and (B) lipid/protein ratio images of heart tissue from the ABY/SnJ mouse model at different disease stages: control, 4 days, 8 days, and 28 days post-infection. The collagen correlation was calculated using the second derivative spectrum of Type I collagen from 1400 to 1000 cm 1. Data collection conditions: pixel size = 25 mm, spectral resolution = 4 cm 1, scan number = 8 scans/pixel. The scale bar represents 1.0 mm.
chronic stages, respectively, while it is 0.329 0.017 in control tissue. Both correlation maps and the coefficient averages indicate there is no significant difference among three disease stages but a 26% increase was found in the collagen correlation for the diseased tissue relative to the control. In addition, the FTIR images show that there is a uniform distribution of collagen deposition in the diseased tissue; it is not a localized effect. This is supported by low standard deviations for the correlation coefficients (Table 1). The L/P ratio is seen in Fig. 2B. As can be seen, a high L/ P ratio exists for the control tissue (0.193 0.017), and the L/P ratio is more than three times lower in all three stages of infected tissue: 0.067 0.010, 0.078 0.001, and 0.062 0.001 for early, acute and chronic stages, respectively. No significant differences were observed between the three diseased stages. Since these data are reported as ratios to account for any sample thickness variations, these results could indicate that the diseased states possess either a higher protein content and/or a lower lipid level. However, the increased collagen content is observed in the diseased stages, strongly suggesting that protein content is increasing.
In addition, the CH2/CH3 ratio can be applied to indicate lipid content, where a higher ratio indicates longer-chained lipids and a higher lipid content [28]. In our studies, the infected tissue had comparable CH2/CH3 ratios as the control tissue (data not shown), supporting the conclusion that the lower L/P ratio in the diseased states comes from elevated protein (i.e. collagen) and not a significant reduction in lipid content. For the C57BL/6 strain, the collagen distribution and L/P ratio images are shown in Fig. 3A and B, respectively, plotted on the same color scales as in Fig. 2A and B. No differences were observed in collagen correlation between the control (0.380 0.006), 4 dpi (0.401 0.010), and 8 dpi (0.392 0.006) tissue. Similarly, the L/P ratio did not differ significantly between the control (0.078 0.010), 4 dpi (0.087 0.032), and 8 dpi (0.062 0.005) tissue. However, a dramatic reduction in collagen content (0.294 0.080) and increased L/P ratio (0.211 0.053) were observed in the 28 dpi tissue. These results suggest that C57BL/6 strain can recover from the CVB3 infection during the acute phase of disease. In addition, the original (control)
Table 1 Lipid/protein ratio (L/P) and collagen correlation (CC): mean S.E.M. Disease stage
Control
4 days
8 days
28 days
ABY/SnJ CC L/P
0.329 0.017 0.193 0.017
0.416 0.011 0.067 0.010
0.437 0.003 0.078 0.001
0.415 0.002 0.062 0.001
C57BL/6 CC L/P
0.380 0.006 0.078 0.010
0.401 0.010 0.087 0.032
0.392 0.006 0.062 0.005
0.294 0.080 0.211 0.053
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
221
Fig. 3. (A) Collagen correlation images and (B) lipid/protein band ratio maps of heart tissue from the C57BL/6 mouse models at different disease stages: control, 4 days, 8 days, and 28 days post-infection. The collagen correlation was calculated using the second derivative spectrum of type I collagen from 1400 to 1000 cm 1. Data collection conditions: pixel size = 25mm, spectral resolution = 4 cm 1, scan number = 8 scans/pixel. The scale bar represents 1.0 mm.
composition of the C57BL/6 strain differs from the ABY/ SnJ strain; the C57BL/6 strain has a higher collagen content and lower L/P ratio than the ABY/SnJ strain. These differences may influence the ability of the C57BL/6 strain to stabilize the disease and initiate recovery. For both the ABY/SnJ and C57BL/6 strains, a strong inverse correlation was observed between the collagen content and the L/P ratio (Fig. 4). This is reflected in the negative slopes for the two linear relationships: 0.70 (ABY/SnJ) and 0.73 (C57BL/6), plotted on a scale from 0 to 1, representing the lowest to highest correlation. The linear regressions, r2, of the two lines
were 0.82 (ABY/SnJ) and 0.64 (C57BL/6). Thus, as collagen concentration increases, the L/P ratio decreases, suggesting that extracellular matrix remodeling, i.e. collagen deposition, replaces normal tissue as the CVB3 infection induces myocarditis. These findings are consistent with numerous previous studies on CVB3 viral myocarditis using murine models. Specifically, ABY/ SnJ(H-2b) is classified as a permissive host, prone to sustaining chronic inflammation of myocarditis by CVB3infection; C57BL/6(H-2b) is a resistant host, which usually eliminates CVB3 during acute myocarditis, preventing the development of chronic myocarditis [34].
Fig. 4. Linear correlation between the lipid/protein ratio and the collagen correlation in the (left) ABY/SnJ and (right) C57BL/6 strains. Means and S.E.M.s were calculated using data collected at a pixel size of 6.25 mm, spectral resolution = 4 cm 1, scan number = 8 scans/pixel.
222
Q. Wang et al. / Vibrational Spectroscopy 38 (2005) 217–222
4. Conclusions and future work Collagen fibrosis has been established as a major contributor to cardiac damage as cardiomyopathy develops. Therefore, monitoring the distribution and concentration of collagen during the different stages of coxasackievirus B3-induced viral myocarditis in different immunocompetent mouse strains can help thoroughly understand the progression of disease. In this work, we demonstrated that collagen content increases and L/P ratio decreases in both the permissive (ABY/SnJ) and resistant (C57BL/6) strains of the CVB3-infected mice, indicating extracellular matrix remodeling in the disease. This effect is not localized, but broadly distributed within the tissue. However, in the chronic stages of the disease, these changes only persist in the permissive ABY/SnJ strain, consistent with the postulation that CVB3 infection induces two different immune responses in affected immunocompetent hosts; the resistant host recovers from myocarditis whereas the permissive host develops chronic disease. The FTIRI results also show that the two strains have different compositions in the uninfected state, which may also provide an insight into the mechanism for the different responses. In short, FTIRI brings a new dimension to understanding the composition of heart tissue in myocarditis, and has the potential to become a reliable tool for monitoring the disease in situ. This study represents an initial application of the technique, but further studies involving a much larger number of cases will be necessary for validation. Future studies will involve subcellular imaging of tissue composition with synchrotron infrared imaging, combined with immunohistochemistry and echocardiography studies. Acknowledgements The NSLS is supported by the United States Department of Energy under contract no. DE-AC02-98CH10886. Charite´ and Tu¨ bingen Universities are supported by the Deutsche Forschungsgemeinschaft (DFG), project no. 2005096. References [1] M. Bursch, Heart 88 (2002) 198–202. [2] W.S. Collucci, E. Braunwald, in: Braunwald, E. (Ed.), Heart Disease, A Textbook of Cardiovascular Medicine, sixth ed., W.B. Saunders Company, Philadelphia, 2001, p. 503. [3] C.A. Brown, J.B. O’Connell, Am. J. Med. 99 (3) (1995) 309–314. [4] M. Burch, S.A. Siddiqi, D.S. Celermajer, C. Scott, C. Bull, J.E. Deanfield, Br. Heart J. 72 (3) (1994) 246–250.
[5] G.W. Dec, V. Fuster, N. Engl. J. Med. 331 (23) (1994) 1564– 1575. [6] D.A. Neumann, Mayo Clin. Proc. 69 (2) (1994) 193–195. [7] T.A. Manolio, K.L. Baughman, R. Rodeheffer, T.A. Pearson, J.D. Bristow, V.V. Michels, W.H. Abelmann, W.R. Harlan, Am. J. Cardiol. 69 (17) (1992) 1458–1466. [8] M.K. Baig, J.H. Goldman, A.L. Caforio, A.S. Coonar, P.J. Keeling, W.J. McKenna, J. Am. College Cardiol. 31 (1) (1998) 195–201. [9] E. Gre`unig, J.A. Tasman, H. Ke`ucherer, W. Franz, W. Ke`ubler, H.A. Katus, J. Am. College Cardiol. 31 (1) (1998) 186–194. [10] C.J. McKenna, M.B. Codd, H.A. McCann, D.D. Sugrue, Heart (Br. Cardiac Soc.) 77 (6) (1997) 549–552. [11] K.R. Bowles, R. Gajarski, P. Porter, V. Goytia, L. Bachinski, R. Roberts, R. Pignatelli, J.A. Towbin, J. Clin. Invest. 98 (6) (1996) 1355–1360. [12] J.B. Durand, L.L. Bachinski, L.C. Bieling, G.Z. Czernuszewicz, A.B. Abchee, Q.T. Yu, T. Tapscott, R. Hill, J. Ifegwu, A.J. Marian, Circulation 92 (12) (1995) 3387–3389. [13] K. Klingel, S. Stephan, M. Sauter, R. Zell, B.M. McManus, B. Be`ultmann, R. Kandolf, J. Virol. 70 (12) (1996) 8888–8895. [14] C. Gauntt, S. Huber, Front. Biosci.: J. Virt. Libr. 8 (2003) 23–35. [15] D.R. Anderson, J.E. Wilson, C.M. Carthy, D. Yang, R. Kandolf, B.M. McManus, J. Virol. 70 (7) (1996) 4632–4645. [16] W.S. Colocci, Am. J. Cardiol. 80 (1997) 15L–25L. [17] A.M. Feldman, P.E. Ray, C.M. Silan, J.A. Mercer, W. Minobe, M.R. Bristow, Circulation 83 (6) (1991) 1866–1872. [18] I.M. Dixon, H. Ju, N.L. Reid, T. Scammell-La Fleur, J.P. Werner, G. Jasmin, J. Mol. Cell. Cardiol. 29 (7) (1997) 1837–1850. [19] D.L. Wetzel, S.M. LeVine, Science 285 (5431) (1999) 1224– 1225. [20] P. Dumas, L.M. Miller, Vibrat. Spectrosc. 32 (2003) 3–21. [21] P. Lasch, W. Haensch, D. Naumann, M. Diem, Biochim. Biophys. Acta 1688 (2) (2004) 176–186. [22] J. Kneipp, P. Lasch, E. Baldauf, M. Beekes, D. Naumann, Biochim. Biophys. Acta 1501 (2–3) (2000) 189–199. [23] R.Y. Huang, L.M. Miller, C.S. Carlson, M.R. Chance, Bone 33 (4) (2003) 514–521. [24] K. Liu, M. Jackson, M.G. Sowa, H. Ju, I.M. Dixon, H.H. Mantsch, Biochim. Biophys. Acta 1315 (2) (1996) 73–77. [25] K.Z. Liu, I.M. Dixon, H.H. Mantsch, Cardiovasc. Pathol.: Off. J. Soc. Cardiovasc. Pathol. 8 (1) (1999) 41–47. [26] R. Kandolf, P.H. Hofschneider, Proc. Natl. Acad. Sci. U.S.A. 82 (14) (1985) 4818–4822. [27] E. Gazi, J. Dwyer, N.P. Lockyer, J. Miyan, P. Gardner, C. Hart, M. Brown, N.W. Clarke, Biopolymers 77 (2005) 18–30. [28] K.M. Gough, D. Zelinski, R. Wiens, M. Rak, I.M. Dixon, Anal. Biochem. 316 (2) (2003) 232–242. [29] J.E. Strobeck, S.M. Factor, A. Bhan, M. Sole, C.C. Liew, F. Fein, E.H. Sonnenblick, Ann. N.Y. Acad. Sci. 317 (1979) 59–88. [30] G. Davison, C.S. Hall, J.G. Miller, M. Scott, S.A. Wickline, Circulation 90 (3) (1994) 1334–1342. [31] B. Stuart, in: D.J. Ando (Ed.), Biological Applications of Infrared Spectroscopy, John Wiley and Sons, Inc., Chichester, 1997, pp. 113– 153. [32] N.P. Camacho, P. West, P.A. Torzilli, R. Mendelsohn, Biopolymers 62 (1) (2001) 1–8. [33] M. Jackson, M.G. Sowa, H.H. Mantsch, Biophys. Chem. 68 (1–3) (1997) 109–125. [34] K. Klingel, J.J. Schnorr, M. Sauter, G. Szalay, R. Kandolf, Am. J. Pathol. 162 (5) (2003) 1709–1720.