Human colorectal cancers display abnormal Fourier-transform infrared spectra

Transcription

1 Proc Nati Acad Sci USA Vol 87, pp , October 199 Medical Sciences Hman colorectal cancers display abnormal Forier-transform infrared spectra (colon cancer/infrared spectroscopy/high-pressre spectroscopy/tmor markers) BASIL RIGASt, SUSAN MORGLLOt, IRA S GOLDMAN, AND PATRICK T T WONG Departments of Medicine and tpathology, Cornell University Medical College, New York, NY 121; Department of Medicine, North Shore University Hospital-Comell University Medical College, Manhasset, NY 3; and 1Division of Chemistry, National Research Concil of Canada, ON KlA OR6, Canada Commnicated by Lewis Thomas, November 27, 1989 (received for review September 13, 1989); revision received Jly 1, 199 ABSTRACT Forier-transform infrared spectroscopy (FT-IR) was applied to the stdy of tisse sections of hman colorectal cancer Pairs of tisse samples from colorectal cancer and histologically normal mcosa 5-1 cm away from the tmor were obtained from patients who nderwent partial colectomy All cancer specimens displayed abnormal spectra compared with the corresponding normal tisses These changes involved the phosphate and C-O stretching bands, the CH stretch region, and the pressre dependence of the CH2 bending and C=-O stretching modes Or findings indicate that in colonic malignant tisse, there are changes in the degree of hydrogen-bonding of (i) oxygen atoms of the backbone of ncleic acids (increased); (ii)oh grops of serine, tyrosine, and threonine resides (any or all of them) of cell proteins (decreased); and (iii) the C=O grops of the acyl chains of membrane lipids (increased) In addition, they indicate changes in the strctre of proteins and membrane lipids (as jdged by the changes in their ratio of methyl to methylene grops) and in the packing and the conformational strctre of the methylene chains of membrane lipids The cell(s) of the malignant colon tisses responsible for these spectral abnormalities is nknown Cltred colon adenocarcinoma cell lines displayed similarly abnormal FT-IR spectra The diagnostic potential of the observed changes is discssed In recent years spectral methods have been sed in the evalation of malignancy, and some attempts have been made to tilize them as a diagnostic tool Most of the work involves NMR spectroscopy (1-3); however, the initial enthsiasm for the diagnostic seflness of NMR has sbsided considerably (4) Infrared spectroscopy is becoming an increasingly powerfl tool for the stdy of the composition and strctre of celllar components within intact tisses (5-7) Crrently, methodological and technological advances are greatly enhancing the sensitivity of infrared spectroscopy In particlar, appreciation of the role of pressre on spectral parameters sch as freqency, intensity, band shape, and band splitting and the conseqent advances in high-pressre instrmentation have frther facilitated the analysis of vibrational spectra (8-1) In this paper we report or findings from the analysis of hman colon cancer tisses by Forier transform infrared (FT-IR) spectroscopy that was also combined with the se of high-pressre (pressre-tning) infrared spectroscopy, when appropriate Or data, showing clear-ct spectral differences between normal and malignant colonic tisse, sggest the potential applicability of this approach to a host of biological problems The pblication costs of this article were defrayed in part by page charge payment This article mst therefore be hereby marked "advertisement" in accordance with 18 USC 1734 solely to indicate this fact PATINTS AND MTHODS Patients The patients whose tisse samples were stdied nderwent partial large-bowel resection for colorectal cancer at North Shore University Hospital-Cornell University Medical College, Manhasset, NY Five were women and 6 were men; their average age was 75 years (range, 59-84) Staging of tmors sing the modified Dkes' classification () showed 1 patient with stage A tmor, 3 with stage B1, 5 with stage B2, and 2 with stage C Tisse samples were obtained immediately after bowel resection From each patient, samples were obtained from both the tmor itself and the normal-appearing mcosa 5-1 cm away from the tmor, frozen in liqid nitrogen, and stored at -18 C ntil sed ach tisse sample was ct into several pairs of small segments (thickness, -1 mm; srface area, =z-2 mm2) One member of each pair was sed for spectroscopic stdies From the other member of the pair, 5-,mthick microtome sections were ct from the side facing the companion tisse segment sed in spectroscopy These sections were fixed in formaldehyde, stained with hematoxylin and eosin, and examined histologically by two pathologists The composition of each tisse section was scored blindly as a percentage of malignant and normal segments In this way we were able to monitor as closely as possible the histological composition of each tisse section examined spectroscopically Technical aspects of tisse sample collection for these stdies are described elsewhere (12) FT-IR Spectroscopy Small amonts (typically 1 mg) of tisse samples were placed at room temperatre, together with powdered a-qartz, in a 37-mm-diameter hole in a 23-mm-thick stainless steel gasket monted on a diamond anvil cell Pressres at the sample were determined from the 695-cm-1 infrared absorption band of a-qartz as described (13) FT-IR spectra at varios pressres in kbars (1 bar = 1 kpa) were measred with a Digilab FTS-6 Forier transform spectrometer sing a liqid nitrogen-cooled mercrycadmim-tellride detector For each spectrm, 512 scans were coadded at a spectral resoltion of 4 cm-' Data redction was performed with software developed in or Ottawa laboratory Cell Lines Hman colon cell lines from the American Type Cltre Collection were maintained in cltre following standard protocols and the instrctions of American Type Cltre Collection They inclded LoVo, HCT15, HT29, SW16, SW43, SW48 (all colon adenocarcinomas) and CCD-18Co (normal hman colon fibroblast) Cells were harvested from conflent plates either by trypsinization or gentle scraping of the tisse cltre plates Control experiments revealed that the infrared spectra of these cells were Abbreviation: FT-IR, Forier-transform infrared spectroscopy tto whom reprint reqests shold be addressed at: Division of Digestive Diseases F-231, New York Hospital-Cornell Medical Center, 525 ast 68th Street, New York, NY

2 Medical Sciences: Rigas et al Proc Natl Acad Sci USA 87 (199) 8141 not affected by the method of cell harvesting Briefly, cells were washed with phosphate-bffered saline (PBS) and incbated with 3 ml of trypsin/dta (GIBCO and BRL) for 1 min at room temperatre; after trypsin was aspirated, the cells were incbated at 37C for 6 min The detached cells were resspended in 1 ml of PBS and washed once with PBS and aliqots were pelleted by centrifgation-all at 4C The pellet was stored in liqid nitrogen ntil analyzed by spectroscopy Scraping of the cells was done in 1 ml of PBS after their washing with PBS, and the same procedre was followed as for the trypsinized cells RSULTS Changes in Phosphate Bands Stdy of the freqency region cm-' revealed significant differences in the infrared spectra between normal and malignant colonic tisse Fig 1, showing the sperimposed infrared spectra of a pair of normal and malignant tisse samples from one patient, illstrates these differences The strongest bands are at 1241 cm-1 and 1824 cm-1 (freqencies in normal tisse) They are de to the asymmetric (vaspo2j) and symmetric (v5poj) phosphate stretching modes, respectively (14) In all cases, when compared with normal tisses, the malignant tisses displayed decreased intensity of the vaspo2 band, increased intensity of the vpoj band, and changes in the shape of these bands, as shown in Fig 1 In addition, the freqencies of these bands were shifted in malignant tisses The vaspo2 band was shifted from 1241 cm-1 in the normal tisse to cm-1, while the vspoi band was shifted from 1824 cm-' to 1851 cm-1 The intensity of these peaks varied from patient to patient and also within the same tmor sample The ratio of the peak intensity of these two bands varied as well The variation in the intensity and band shape in this freqency region may reflect the variation in the proportion of nonmalignant and nonepithelial cells present in each tisse section These bands originate mainly in the phosphodiester backbone of celllar ncleic acids (14) The contribtion of these bands by phosphate resides present in membrane lipids is negligible, as indicated by the relative intensities of other vibrational modes of membrane lipids For example, the peak intensity ratio between the vc=o band and the vaspo2 band of phospholipids is In the spectra of hman colon tisses, this ratio is This finding indicates that phospholipids do not contribte appreciably to the intensity of the vaspo2 band in the infrared spectra of hman colon tisses Since the intensity of the vspoi band is slightly FIG 2 Infrared spectra of a pair of normal and malignant colon tisses (freqency region, cm-1; ato-scale plotting) (A) Original spectra (B) Spectra after band narrowing with Forier self-deconvoltion as described in text weaker than that of PasPO2 in phospholipids, phospholipids also do not contribte to the intensity of the vspoj band in the infrared spectra of hman colon tisses The enlarged spectra of the vaspo- band of a pair of normal and malignant colonic tisses are plotted together in Fig 2 Fig 2A shows the original spectra, whereas Fig 2B shows the corresponding vaspo2 spectra after band narrowing sing Forier self-deconvoltion with an enhancement factor of 145 and band width of 2 cm-' (15) In malignant tisse this band splits into two at and cm-1; sch splitting is barely discernible in normal tisse The freqency of the vaspoj mode is at cm-' when it is completely non-hydrogen-bonded and at abot 122 cm-' when it is flly hydrogen-bonded (16) Therefore, in these malignant colonic tisses, the hydrogen-bonding of the oxygen atoms of the phosphate backbone of ncleic acids is increased as indicated by the increase in the intensity of the cm-' band This is in contrast to the sitation in the corresponding normal colonic tisse Fig 3 shows the pressre dependences of the freqencies of the two vaspoi component bands of malignant colonic tisse The freqencies of these bands were calclated from the third-power derivative spectra with a breakpoint of 3 (15) The freqency of the band at cm-' decreases with increasing pressre This provides frther evidence that this 'I) c) -Q U) : cj ) [ 1 23 [ [ w p p FIG 1 Infrared spectra of tisse sections from colon adenocarcinoma and histologically normal mcosa 1 cm away from the tmor (freqency region, cm-1; ato-scale plotting configration, in which peak intensities are atomatically normalized to the highest peak) FIG 3 Pressre dependences of the asymmetric phosphate stretching freqencies of malignant colon tisse

3 8142 Medical Sciences: Rigas et al Proc Natl Acad Sci USA 87 (199) 8 8) :3 LL- 75 [ 7 [ ~O~O~ a- cp n B FIG 4 Infrared spectra of a pair of normal and malignant colon tisses (freqency region, cm' ; ato-scale plotting) (A) Original spectra (B) Spectra after band narrowing with Forier self-deconvoltion as described in text band is de to the vaspo2 mode of the hydrogen-bonded phosphate backbone of ncleic acids (17) The freqency of the cm-1 component band increases with increasing pressre, which is the reslt of compression of the chemical bonds in non-hydrogen-bonded PO- grops (17, 18) Changes in the C-O Band The relatively weak band at 642 cm-1 (this is its freqency in normal tisse) is de to the stretching mode (vc-o) of C-O grops present in cell proteins (14) The vc-o band of the acyl chains of membrane lipids appears in this freqency region as well However, the contribtion to this band by membrane lipids is negligible, as indicated by the relative intensities of the vc-o and vc=o modes of membrane lipids The peak intensity ratio between the vc=o band and the vc-o band of lipids is in the range , whereas in the spectra of hman colon tisses this ratio is 6-8 Conseqently, the vc-o band in the spectra of hman colon tisses is essentially from the vc-o mode of cell proteins In malignant tisse this band changes in both shape and peak maximm The peak maximm shifts from 642 cm1 in normal tisses to 731 cm-1 in malignant tisses Fig 4 shows the enlarged spectra of this band Fig 4A shows the original spectra, whereas Fig 4B shows the corresponding spectra after band narrowing sing Forier self-deconvoltion with an enhancement factor of 14 and a band width of 2 cm-' (15) a- a) Lo 82 malignant 8 o normal OOcy<bO &)O O FIG 5 Pressre dependences of the center of gravity of the vc-o bands of a pair of normal and malignant colon tisses FIG 6 Pressre dependences of the freqencies of the vc-o component bands of normal colon tisse Stdy of the pressre dependence of spectral parameters of these tisses reveals changes in both the position of the center of gravity and the peak maximm of these bands When pressre is applied to the tisse, the center of gravity of this band moves towards lower freqencies in normal tisse In contrast, in malignant tisses the center of gravity shifts towards higher freqencies (Fig 5) The freqencies in Fig 5 were calclated from the third-power derivative spectra with a breakpoint of 2 (15) The deconvolted spectra (Fig 4B) show that the--vc-o band consists of two overlapping bands in both normal and malignant tisses The intensity of the low-freqency component band is dramatically decreased in malignant tisses The pressre dependence of the freqencies of these two component bands of normal colonic tisse, calclated from the third-power derivative spectra with a breakpoint of 3 (15), is shown in Fig 6 The freqency of the high-freqency component band increases with increasing pressre, while that of the low-freqency component band decreases with increasing pressre In general, the vibrational freqency of a C-O fnctional grop in a molecle is increased by external pressre becase of the compression of the chemical bond of the fnctional grop and the pressre-enhanced intermoleclar interactions On the other hand, if the fnctional grop is hydrogen-bonded, its vibrational freqency is decreased With increasing pressre, the vibrational freqency of the fnctional grop is decreased frther (17) The reslts shown in Figs 4 and 6 indicate that the low-freqency component band is de to the vc-o mode of the hydrogenbonded C-O grops, whereas the high-freqency band is de to the v'c-o mode of the non-hydrogen-bonded C-O o cancer 2% cancer 8% cancer ~n /\ FIG 7 Infrared spectra of the vc-o band of three colonic tisse samples differing in the proportions of normal and malignant components These samples are from the same patient

4 Medical Sciences: Rigas et al Proc Natl Acad Sci USA 87 (199) o normal malignant C-) U) U C) C3 r (D 1 74 % oo ~oco I Freqency Cm FIG 8 Infrared spectra of a pair of normal and malignant colon tisses (freqency region, cm-l; ato-scale plotting) grops The dramatic decrease in the intensity of the lowfreqency component band in malignant tisses sggests that in colon cancer most of the hydrogen bonds in the C-O grops have disappeared The integrated intensity of the lower freqency component band decreases as the proportion of malignant cells in a tisse sample increases (Fig 7) Changes in the CH Stretch Region Fig 8 shows the infrared spectra of a pair of normal and malignant colonic tisses in the CH stretching region The band at cm-1 is de to the symmetric CH2 stretching mode (vpch2) of the methylene chains in membrane lipids The band at cm-1 is de to the asymmetric stretching mode of the methyl grops (PasCH3) The vasch3 modes of both the end-methyl grops of membrane lipids and the methyl grops of proteins are at abot the same freqency (14, 17) Therefore, the vasch3 mode of methyl grops of both membrane lipids and proteins contribte to the intensity of the band at cm-' In malignant colonic tisses, the intensity of the vasch3 band is decreased while that of the vsch2 band is increased, when compared with the corresponding bands in normal tisse (Fig 8) This indicates that the ratio of the nmber of methyl grops to that of methylene grops is decreased in malignant tisse, compared with normal colonic tisse While this ratio is decreased in the malignant tisse of all pairs of normal and malignant tisses that we stdied, its absolte vale varies from patient to patient Pressre Dependence of the CH2 Bending Mode The pressre dependence of this mode has been widely sed to C) LL [ 1 47 [ [ o norma ma lignant d FIG 9 Pressre dependences of the CH bending freqencies of methylene chains of membrane lipids of a pair of normal and malignant colon tisses p U kbor Pressre, FIG 1 Pressre dependences of the C=O stretching freqencies of methylene chains of membrane lipids of a pair of normal and malignant colon tisses investigate the packing characteristics of methylene chains in membrane lipids (19, 2) Fig 9 shows the pressre dependence of the CH2 bending mode of the methylene chains in membrane lipids in a pair of normal and malignant colonic tisse The freqencies were obtained from the third-power derivative spectra with a break point of 4 (15) At atmospheric pressre the freqency of this band is higher in the malignant tisse than in its normal conterpart; this relationship is reversed as pressre increases The range of relative change of the freqency of this mode over a wide range of pressres is mch more restricted in the malignant tisses compared with the normal controls (in this example, in the normal tisse the increase in freqency of this mode in going from atmospheric pressre to 2 kbar is 4 times greater than that noted in malignant tisse) These findings indicate that in malignant tisses at atmospheric pressre, the conformational strctre of the methylene chains of membrane lipids is more disordered than in normal tisses As pressre increases the interchain interactions are enhanced as shown by the increase of this freqency The pressre-enhanced interchain interactions is larger in the more conformationally ordered normal tisse (19, 2) Pressre Dependence of the C=O Stretching Mode These freqencies were obtained from the third-power derivative spectra with a breakpoint of 3 The intensity of the C=O stretching mode (vc=o) in the spectra of tisses arises from the vibration of the C=O grops of the acyl chains in membrane lipids (17, 18) Sometimes, however, the vc=o mode of tisse triglycerides presenting as microscopic oil drops, contribtes to the intensity of this band The data of the vc=o band shown in Fig 1 were obtained from a pair (I) c) U U) , 85 FIG Infrared spectra of the vc-o bands of cltred colon fibroblasts and the LoVo colon adenocarcinoma cell line

5 8144 Medical Sciences: Rigas et al of normal and malignant tisses from a single patient after the oil droplets were removed by careflly washing the tisse samples with 2H2 The intensity of this band varied from patient to patient In both normal and malignant tisses, the freqency of this band decreased with increasing pressre; however, this decrease was more prononced in malignant tisses The latter finding indicates that the hydrogenbonding of the acyl C=O grops is stronger in malignant tisses compared with normal tisses (17, 18) IR Spectra of Colon Cell Lines To frther evalate these findings, an infrared spectroscopic stdy of a series of cltred hman colon cell lines was ndertaken Details will be given elsewhere Many of the infrared spectroscopic featres are common to both the cltred cell lines and the colonic tisses that we stdied For instance, the vc-o bands of cltred colon fibroblasts and the LoVo cell line are compared in Fig These spectra are similar to those of colonic tisse shown in Fig 4A DISCUSSION Or data demonstrate that colorectal cancers display abnormal FT-IR spectra compared with normal control tisse from the same individal The entire spectrm of the evoltion of colon cancer-ie, the varios forms of adenoma -have not been examined yet to know whether they also manifest these spectral abnormalities The spectral changes that we have observed reflect alterations in the strctre of important informational and strctral molecles in the malignant tisse They involve the degree of hydrogen-bonding of (i) oxygen atoms of the backbone of ncleic acids (they are the two oxygen atoms bond to P that do not participate in the phosphodiester bond); (ii) C-O grops in cell proteins; and (iii) the C=O grops of the acyl chains of membrane lipids There are also additional changes in the strctre of proteins and membrane lipids, as jdged by the changes in their ratio of methyl to methylene grops Finally, the packing and the conformational strctre of the methylene chains of membrane lipids are changed in colonic malignant tisse Althogh or findings do not allow a more specific assignment of these changes, nevertheless they indicate extensive chemical and physical changes in the malignant colonic tisse, and, in a way, attest to the complexity of the malignant phenotype It is interesting to note that the severe redction in hydrogen bonding of C-O grops of proteins in malignant colon tisse can involve only three amino acids-namely, tyrosine, serine, and threonine The OH grops of these three amino acids, which form the C-O bonds monitored here, are phosphorylated by several oncoproteins, an event considered important in carcinogenesis (21) We do not know the cell type responsible for the changes in the FT-IR spectra of malignant colonic tisses There are two reasons to sspect that these changes may originate in the malignant colonocyte per se: (i) the integrated intensity of the C-O band in malignant tisse changes in tandem with the proportion of malignant cells present in a given tisse section, and (ii) the spectral characteristics of cltred colon cancer cell lines are, in general, similar to those observed in tisse Proc Natl Acad Sci USA 87 (199) sections Regarding the latter point, however, it shold be noted that, besides the general problem of extrapolating tisse cltre findings into intact tisses, or colon cancer cell lines were compared with normal colonic fibroblasts (for lack of a reliable cltre colonic epithelial cell) Ongoing work is attempting to assess directly the celllar origin of the observed spectral abnormalities Whatever the exact natre of the physical and chemical alterations in the malignant colonic tisses, or stdy has two practical implications First, it demonstrates that it is feasible to apply prodctively pressre-tning FT-IR spectroscopy to the stdy of hman tisses This approach provides the opportnity to monitor simltaneosly several chemical and physical parameters of a tisse with very small samples Second, it sggests that determination of these spectra may be of some se in rapid evalation of the malignant phenotype in colonic tisse and perhaps other tisses We thank Drs C J McDogall and S-S Ngoi for their assistance dring an early stage of this work; Dr H Lafer for evalating tisse sections; Linda Messina, Lilly Stiel, and Christopher Lovelace for technical assistance; and Mrray Cohen for assistance in the preparation of this manscript 1 Fossel, T, Carr, J M & McDonagh, J N (1986) N ngl J Med 315, Peeling, J, Stherland, G, Marat, K, Tomchk, & Bock, (1988) J Nerosrg 68, Smith, I C P, Sanders, J K, Barr, J R, Williams, P A & Montford, C (1986) Biophys J 49, Sppl, 393 (abstr) 4 Shlman, R (199) N ngl J Med 322, Takahashi, H, French, S W & Wong, P T T (1989) Hepatology 1, 75 (abstr) 6 Takahashi, H, French, S W & Wong, P T T (199) Alcohol Clin xp Res, in press 7 Ager, M, Jarrell, H C, Smith, l C P, Wong, P T T, Siminovitch, D J & Mantsch, H H (1987) Biochemistry 26, Wong, P T T (1987) in High Pressre Chemistry and Biochemistry, eds von ldik, R & Jonas, J (Reidel, Dordrecht, The Netherlands), NATO ASI Ser C, Vol 197, pp Wong, P T T (1987) in Crrent Perspective in High Pressre Biology, ed Marglis, R (Academic, London), pp McDonal, R S (1986) Anal Chem 58, Astler, V B & Coller, F A (1954) Ann Srg 139, Wong, P T T & Rigas, B (199) Appl Spectrosc, in press 13 Wong, P T T, Moffatt, D J & Badais, F L (1985) Appl Spectrosc 41, Parker, F S (1971) Application of Infrared Spectroscopy in Biochemistry, Biology and Medicine (Plenm, New York) 15 Cameron, D G & Moffatt, D J (1984) J Testing val 12, Swai, H (1969) in Strctre and Stability of Biological Macromolecles, eds Timasheff, S N & Fasman, G D (Dekker, New York), pp Wong, P T T (1987) in Vibrational Spectra and Strctre, ed Drig, J R (lsevier, Amsterdam), Vol 16, pp Wong, P T T & Mantsch, H H (1988) Chem Phys Lipids 46, Wong, P T T & Mantsch, H H (1988) Biophys J 54, Wong, P T T & Hang, C (1989) Biochemistry 28, Bishop, M (1985) Cell 42, 23-38