This document describes a study that used microspectrofluorometry to analyze and compare the natural fluorescence of normal and neoplastic (cancerous) human colon tissue. The study found:
1) Normal and cancerous colon tissues exhibited different patterns of fluorescence intensity and spectral shape, related to their distinct histological organizations.
2) The most evident spectral differences involved the stromal compartment and were likely due to different fluorochromes, related to the host response to tumors.
3) The nature and extent of autofluorescence modifications between normal and cancerous tissues helped explain previous "in vivo" analysis findings and highlighted the importance of excitation parameters for exploiting autofluorescence in diagnosis.
Natural fluorescence of normal and neoplastic human colon
1. Lasers in Surgery and Medicine 1648-60 (1995)
Natural Fluorescence of Normal and
Neoplastic Human Colon: A
Comprehensive “Ex Vivo” Study
Giovanni Bottiroli, PhD, Anna C. Croce, PhD, Donata Locatelli, BSC,
Renato Marchesini, PhD, Emanuele Pignoli, PhD, Stefan0 Tomatis, PhD,
Carolina Cuzzoni, MD, Silvana Di Palma, MD, Marco Dalfante, MD, and
Pasquale Spinelli, MD
Center for Histochemisrry, CNR (G.B., A.C.C., D.L.) and Department of Surgery,
University of Pavia (C.C.), Pavia and Division of Health Physics (R.M., E.P., S.T.) and
Divisions of Pathology and Cytology (S.D.f?)and Endoscopy (M. P.S.), National
D.,
Cancer Institute, Milan, Italy
Background and Objective: A microspectrofluorometric analysis
on “ex vivo” samples from normal tissue and adenocarcinoma of
the human colon has been performed to characterize the histo-
logical, biochemical, and biophysical bases of the autofluores-
cence.
Study DesignlMaterials and Methods: Differences between nor-
mal and tumor tissues are found that concern both the intensity
distribution and spectral shape of the autofluorescence emission.
The different pattern of the fluorescence intensity can be related
to the histological organization of the tissue, and involves mainly
the arrangement of the submucosa, the most fluorescent layer.
Results: The most evident differences in the spectral shape found
in the 480-580 nm range involve the stromal compartment, seem
to be due to the presence of different fluorochromes, and are
possibly related to the host response to the tumor.
Conclusion: The nature and the extent of the autofluorescence
modification between normal and tumor tissue in sections ex-
plain at least partly the evidence of the “in vivo” analysis and
highlight the importance of excitation for full exploitation of the
potentials of autofluorescence in diagnosis. Q 1995 Wiley-Liss, Inc.
Key words: microspectrofluorometry, fluorescence imaging, endogenous
fluorophors, adenocarcinoma, tissue organization
INTRODUCTION changes and the identification of the stage of the
“adenoma-carcinoma” sequence are of potential
Gastrointestinal neoplasia is a pathology clinical importance in the screening and treat-
that receives particular attention owing t o its ment of patients at high risk for cancer. Such a
high and increasing incidence in Western indus- goal is difficult to attain through a colonoscopic
trialized countries. Gastrointestinal malignant analysis because it is based on gross architectural
cancer is supposed t o arise in a premalignant le- and morphological changes of the tissue. More
sion, since various stages of tumor progression specific techniques, exploiting cytological and bio-
can be found in the same individual. A sequence
of events is proposed, based on genetic and envi-
ronmental factors as initiating and promoting Accepted for publication March 7 , 1994.
agents of the malignant progression, causing a
Address reprint requests to Dr. Giovanni Bottiroli, Centro di
field defect of colonic mucosa that can evolve from Studio per l’Istochimica, CNR, Dipartirnento di Biologia
flat dysplasia to adenomatous polyp and finally to Anirnale, Universita di Pavia, Piazza Botta, 10-27100 Pavia,
carcinoma [1,21. The detection of early dysplastic Italy.
0 1995 Wiley-Liss, Inc.
2. Microspectrofluorometry of Human Colon 49
chemical alterations, are time-consuming and number of excitation and emission bands [91. Dif-
cannot be accomplished directly “in vivo,” be- ferences between the fluorescence pattern ob-
cause they require biopsy. For some years there served in the overall emission of normal and in
has been a growing interest in detection of tumors that of neoplastic tissue confirmed the validity of
through laser-induced fluorescence analysis, a the spectroscopic approach and allowed diagnostic
non-invasive technique in which fluorescent algorithms to be proposed. The importance of the
drugs, mainly porphyrin derivatives, are used, be- morphological in addition t o the biochemical fea-
cause of their tumor-localizing properties [31. tures of the tissues in determining the discrimi-
More recently, the intrinsic photophysical nating characteristics has been suggested by
properties of biological tissues have been consid- Schomacker et al. [6], who attributed the rela-
ered as a parameter to discriminate diseased from tively low fluorescence signal of neoplastic tis-
normal tissues, In fact, when excited at a suitable sues t o a reduced involvement of the collagenous
wavelength, most of the biological components, layers. The complexity, however, of the biological
either related to the tissue organization or in- substrate makes it difficult to ascertain the rela-
volved in metabolic and functional processes, give tionship between the spectroscopic evidence and
rise t o a fluorescence emission over a wide spec- the biochemical and histological features of the
tral range in the UV-visible region. Since the flu- tissues that could guide the choice of the experi-
orescence emission is strictly dependent on the mental parameters of fluorescence spectroscopy
biochemical and physicochemical properties of suitable for an optimal diagnostic scheme.
the substrate, tissues with different morphofunc- The purpose of this study is to define the in-
tional conditions are in principle discriminated on trinsic properties of the colon tissues and to iden-
the basis of their spectral properties. tify the biochemical and histological aspects that
The literature reports evidence of different result in spectroscopic characteristics sufficiently
fluorescence properties of normal and pathologi- different for malignant lesions to be discrimi-
cal conditions of human colon, concerning adeno- nated from the corresponding normal tissue. To
carcinoma, adenoma, hyperplastic, and normal this end, a microspectrofluorometric study has
tissue. Cothren et al. [41 showed that, under exci- been performed on “ex vivo” samples from normal
tation at 370 nm, the “in vivo” fluorescence in- tissues and adenocarcinoma, and the intrinsic
tensity in the visible range of adenomas is signif- photophysical properties of the biological compo-
icantly lower than that of non-neoplastic tissues. nents have been related to the morphofunctional
Results obtained by Kapadia et al. [5] in a study characteristics of the tissues.
on specimens removed at colonoscopy indicated
that adenomatous polyps can be distinguished
from hyperplastic polyps and normal tissue by a MATERIALS AND METHODS
quantitative analysis of the autofluorescence ex- Tissue Processing and Handling
cited at 325 nm. Schomacker et al. [61 observed Samples of non-neoplastic human colon and
that, under excitation at 337 nm, the fluores- human colonic tumors were obtained from seven
cences of the hyperplastic and adenomatous tis- patients who underwent oncological surgery. Nor-
sues differ from that of the normal tissue, primar- mal controls were obtained from uninvolved areas
ily in the 450 nm spectral component, attributed in resection specimens. Tumors were histologi-
to NADH. The authors showed that at endoscopy, cally diagnosed as moderately differentiated ade-
bnder excitation at 405 nm, neoplasias differ from nocarcinoma. Soon after surgical resection, the bi-
normal tissues in fluorescence intensity, and that opsy was divided into two parts: One of them was
lesions with different degrees of malignancy are immediately frozen, paying attention to the ori-
distinguished by their spectral shape [71. entation of the sample. Colon tissue longitudinal
“Ex vivo” studies confirmed that the spectro- sections of 20 p,m thickness were then obtained at
scopic properties of the overall fluorescence emis- cryostat and immediately submitted to microspec-
sion of malignant, premalignant, and non-neo- trofluorometric analysis. The remaining part of
plastic tissue are sufficiently different t o ensure a the biopsy was homogenized in a buffer solution
reliable differentiation in oncological diagnosis at moderate ionic strength, where most proteins
181. In addition, spectrof luorometric analysis are maximally soluble [lo]. Phosphate buffer so-
based on methods using fluorescence excitation- lution (pH 7.4; 0.05 M) containing anti-oxidant
emission matrices evidenced the heterogeneous (dl-dithiothreitol; 20 mM), proteolytic inhibitor
nature of the autofluorescence, resulting in a (phenylmethane sulphonyl fluoride; 1 mM), che-
3. 50 Bottiroli et al.
lating agent (EDTA; 5 mM), and detergent for Microspectrofluorometric analysis on tissue
protein solubilization (Triton X-100; 0.01% w/v) sections was performed by means of a Leitz mi-
was used. Pieces of tissue, always kept wet with crospectrograph (Wetzlar, Germany), equipped
buffer solution, were disrupted into small frag- with an optical multichannel analyzer (EG & G,
ments by means of a surgical blade and then Princeton Applied Research, Princeton NJ). Flu-
ground three times for 15 s each time with an orescence emission was spectrally dispersed along
Ultra Turrax T25 tissue disintegrator (Janke & the horizontal axis of the exit plane of the poly-
Kunkel-IKA Labotechnik, Staufen, Germany), chromator and imaged into 512-element intensi-
keeping the tube in an ice bath. The homogenate fied linear diode array detector (mod. 1420/512),
was centrifuged for 15 min at 5,OOOg.The super- digitalized, and processed by the Optical Multi-
natant fraction was collected, diluted 10 times channel Analyzer, under computer control.
with the phosphate buffer solution, and centri- A high-pressure 100 W Mercury lamp ( 0 s -
fuged again until a clear solution was obtained, to ram, Berlin, Germany) was used as an excitation
be submitted to fluorometric analysis. source, in combination with KG1 and BG38 an-
The possible interferences of extraction me- tithermic filters, and UG1 band filter. Either 366
dium were evaluated by measuring the fluores- nm (HBw = 10 nm; T = 30%)or 405 nm (HBw =
cence of the pure solutions of single components. 10 nm; T = 35%)interference filters were used to
Triton X-100 only turned out to be fluorescent, in select the excitation wavelengths, corresponding
the spectral range 400-500 nm, under excitation to the 366 and 405 nm lines of the lamp. Excita-
at 310 nm. At, however, the dilution reached in tion was performed under epi-illumination condi-
the final extraction solution, its contribution to tions, using a TK405 dichroic mirror. The trans-
the tissue extract fluorescence did not exceed 3% mittance curve of the dichroic mirror (T400 =
at 440 nm, and was quite negligible at 540 nm. lo%, T410 50%, T,,, = 90%) induces some dis-
=
The pellet was submitted for further extrac- tortions in the short-wavelength region (410-430
tion of lipid fluorophores, using a mixture of chlo- nm) of the spectra, which are constant in all mea-
roform, methanol, and water (2:l:O.E by vol.) ac- surements.
cording to Armstrong et al. [ll]. Leitz objectives, 25x (N.A. 0.50) and 40x
(N.A. 0.651, were employed for all the measure-
Chemicals ments.
Fluorescence analysis was performed on
pure solution of the following compounds: bovine Fluorescence of Tissue
albumin, elastin, collagen type I11 (concentration Microspectrofluorometric scan-analysis was
0.01 mg/ml), pyridoxine, pyridoxal-5-phosphate, performed on longitudinal sections of normal and
4-pyridoxic acid, flavin-mononucleotide, flavin- tumor tissues, from the region exposed to the lu-
dinucleotide, and protoporphyrin IX disodium men toward the inside, with steps of 43 pm. A
salt (concentration of M). All the compounds fixed diaphragm, with a 300 x 30 pm2 area on
were directly dissolved in phosphate buffer solu- the objective plane, was used to delimit the tissue
tion (pH 7.4,0.05 M), except elastin and collagen, region under measurement.
dissolved in acetic acid (0.1 M), and then neutral- A variable iris diaphragm, with an area
ized in phosphate buffer. All the compounds were ranging from a few t o about 1,000 pm2 on the
purchased from Sigma (St. Louis, MO). objective plane, was used to perform microspec-
Lipopigment solutions were obtained from trofluorometric analysis on selected histological
neurons and non-neuronal cells of rat brain by structures.
extraction according to Armstrong et al. [ll].
Experimental Apparatus RESULTS
Fluorescence analysis in solution was per- Fluorescence Scan-Analysis of Longitudinal
formed by means of a spectrofluorometer (Applied Tissue Sections
Photophysics, London, UK; mod SP-2), equipped A typical fluorescence pattern of the longi-
with a Xenon 300 W lamp (O.R.C., Azusa, CA), a tudinal section of normal human colon is shown
photomultiplier tube RCA 8850, and a photon in Figure la. Differences in fluorescence inten-
counting system (EG & G, Ortec, Princeton, NJ). sity make the histological organization of the tis-
Excitation and emission light was selected by sue visible: the superficial epithelium, which in-
means of monochromators. vaginates to form mucosal glands extending into
4. Microspectrofluorometry of Human Colon 51
Fig. 1. Fluorescence microphotographs of colon tissue sec- carcinoma tissue showing the neoplastic cells of a degener-
ltions obtained under excitation a t 366 nm. a Non-neoplastic
: ated gland. f: A detail of the stroma: Yellow-fluoresceing li-
tissue. b: A detail of a showing glands and lamina propria. c: popigment granules associated ta infiltrating cells are visible.
b detail of a showing the dense collagenous connective tissue a, d, bar = 90 pm; b,c,e,f, bar = 22.5 pn.
of submucosa. d adenocarcinoma tissue. e: A detail of adeno-
U e lamina propria, the muscularis mucosa, and
h illustrating portions of mucosa with glands and
U e strongly fluorescent submucosa. Details at
h lamina propria, and of submucosa, respectively.
higher magnification are shown in Figure lb,c, An example of fluorescence patterns of lon-
5. 52 Bottiroli et al.
-
170n
Fig. 2. Patterns of microspectrofluorometric scan-analysis performed along the axis of lon-
gitudinal tissue sections. Each curve corresponds t o the emission spectrum recorded on a
diaphragmed area of 300 x 30 pm2 on the tissue section. Spectra shown are measured every
170 km. Excitation wavelength: 366 nm. a Non-neoplastic tissue. b: Adenocarcinoma le-
:
sion.
gitudinal sections of moderately differentiated considered and were characterized by a strict de-
human colon adenocarcinomas is shown in Figure pendence on the tissue depth. From the lumen
Id. The hyperplastic condition of the neoplasia inward, the fluorescence intensity at first showed
modifies drastically the organization of the tissue. a slight decrease, followed by a marked increase.
The tumor has infiltrated deeply through the sub- The average curve of the fluorescence intensity as
mucosa and muscularis, so that these two layers a function of the tissue depth (Fig. 3) indicated
are no longer observed. Figure le,f shows, at that the slight decrease occurred at a depth less
higher magnification, an area of neoplastic cells, than 50 p,m, passing from superficial epithelium
and a portion of stroma, respectively. to mucosa, while the marked increase appeared at
Typical examples of the fluorescence pattern about 450 Fm, corresponding to the depth where
obtained through microspectrofluorometric scan- the muscolaris mucosa and the fibrous and dense
analysis performed on longitudinal sections of connective tissue, submucosa, are known to be lo-
both non-neoplastic and neoplastic colon, under cated.
excitation at 366 nm, are shown in Figure 2a,b, In tumor tissue, the fluorescence pattern
respectively. In order to avoid the reabsorption was found t o vary among samples. In each sample
effect due t o the presence of heme-based mole- the values of the fluorescence intensity, although
cules 1121, only section regions not affected by the somewhat variable, were comparable to those of
presence of blood, arising from tissue trauma dur- the mucosa layer of normal tissue over almost all
ing resection, were selected. the scanned region. Highly fluorescent regions,
The patterns of the tumors differed from corresponding t o submucosa, were observed only
those of the non-neoplastic tissues in both the in- in tumors with a low degree of invasiveness, at a
tensity and the spectral shape of the fluorescence depth greater than 800 pm. The intensity values
emission. The non-neoplastic tissues exhibited were comparable to those of submucosa of normal
patterns that were very consistent in all samples tissue. Examples of curves of the fluorescence in-
6. Microspectrofluorometry of Human Colon 53
25.000
20.000
h
$
Y
i
E
6
v)
15.000
I-
z
W
0
Z
% 10.000
8
U
3
!
-I
LL
5.000
0 '-
0 200 400 600 800 1.ooo 1.200 1.400
TISSUE DEPTH @rn)
Fig. 3. Curves of fluorescence intensity at 440 2 10nm vs. ments performed on different regions of sections from four
tissue depth measured in sections of non-neoplastic and ade- samples of non-neoplastic tissue. (*) Adenocarcinoma tissue.
nocarcinoma tissues. Excitation wavelength 366 nm. (A) The curves shown were recorded on different regions of sec-
Non-neoplastic tissue. The curve is the average of measure- tions from three samples of adenocarcinoma.
tensity as a function of the tumor depth are shown consisted of a poorly structured emission band in
in Figure 3. the 440-580 nm region, with significant modifi-
As well as the signal amplitude, the fluores- cations depending on the histological component
cence emission measured on the longitudinal sec- considered. The less structured profile was ob-
tions of normal and tumor tissue differed in the served in the case of submucosa, which exhibited
spectral shape. In particular, a broadness of the a narrow emission band centered at about 440
emission band in the green-yellow region was re- nm. A red-shift of the peak position, along with a
corded in the spectra of the tumor lesion, along broadening of the emission band toward longer
with the occasional appearance of a narrow band wavelengths, was measured in the superficial ep-
at about 630 nm. ithelium and in the mucosa. In this latter case,
some degree of variability in the spectral shape
MicrospectrofluorometricAnalysis of was observed when the measurements were per-
Histological Structures formed on the lamina propria. A shape closely re-
To investigate the possibility of a correlation sembling that of submucosa was recorded in the
between the spectral properties of the tissues and region characterized by the presence of a fluores-
their morphofunctional characteristics, a spectral cent net, possibly attributable to fibres of collagen
analysis was performed on selected and dia- in the loose connective tissue. A shoulder at about
phragmed regions corresponding to the histologi- 520 nm was measured in relation to the presence
cal components of both normal and tumor tissues. of green-yellowish granules. These granules,
Typical visible-emission spectra of normal tissue, which were observed in relatively small numbers
excited at 366 nm and filtered through a 405 nm in all samples considered and were unevenly dis-
dichroic mirror, are shown in Figure 4a. They tributed in the tissue, appeared associated with
7. 54 Bottiroli et al.
m
10 -
440 460 480 500 520 540 560 580 600 620 640 660 680 700 440 460 480 500 520 590 560 580 600 620 640 660 680 700
WAVELENGTH lnml WAVELENGTH lnml
Fig. 4. Emission spectra recorded on regions of tissue sec- ponent (n = 10 for each of the seven patients). a Non-neo-
:
tions selected by means of a diaphragm with a dimension plastic tissue: (--.--.-) superficial epithelium, FI = 160 5
variable from a few t o about 1,000 pm2, depending on the 6.4; (-- - - -) lamina propria, FI = 132 2 16.5; - ( 1
histological component to be measured. Excitation wave- glands, FI = 119 k 14.2; (. . . .) submucosa, FI = 1,430 f
length = 366 nm. The curves are normalized to the peak 480. b: Adenocarcinoma tissue: - ( ) neoplastic cells, FI
values. Average values of fluorescence intensity a t the emis- = 105 f 13.2; (- - - - 1 stroma, FI = 139 +- 21.7; (. . . .)
sion peak (FI, arbitrary units, f SD), normalized to the area submucosa, FI = 1,250 I (--.--.-)
371; yellow granules FI
under measurement, arc reported for each histological com- = 721 2 247.
cells infiltrating the connective tissue and resem- glands. As to the other histological components,
bled closely, in both fluorescence and morphol- the typical superficial epithelium is hardly recog-
ogy, particles attributed by some authors t o lipo- nizable in the tumor lesion. Microspectrofluoro-
pigment-loaded lysosomes [13,141. metric analysis was performed on the surface
As to the emission intensity, the signal am- area exposed to the lumen: The spectra, although
plitudes of the histological components, normal- somewhat affected by a certain variability, exhib-
ized t o the area under measurement, are reported ited a relative amplitude at longer wavelengths
in the legend of Figure 4a. When the fluorescence higher than in non-neoplastic epithelium. The
intensity of submucosa is compared to that of mu- spectra recorded on tumor stroma were character-
cosa, a higher relative amplitude is found than ized by the presence of an emission band at about
that observed in the microspectrofluorometric 510-520 nm that appeared more evident than
scan-analysis. This can be understood taking into that of normal lamina propria and that was pref-
account that, owing to the dimension of the mea- erentially observed in the regions close to the ar-
suring diaphragm, the scan-analysis possibly in- eas exposed t o the lumen. The microscopical ob-
volves non-f luoresceing regions produced by the servation of these regions did not allow a well
action of the knife on the fibrous tissue. defined attribution to any histological component,
In the neoplastic tissue the emission spectra although staining with the lipophilic dye Nile
recorded on the submucosa layer, when present, Red [151 showed the presence of lipid microdrop-
exhibited shapes quite similar to those of the cor- lets dispersed into the tissue. Shoulders at about
responding layer of normal tissue, as spectra of 520-560 nm were also measured in some regions
neoplastic cells resembled those of the normal where an enrichment was found in the green-yel-
8. Microspectrofluorometryof Human Colon 55
100
50.
I I
300 350 300 350 400 450
WAVELENGTH (nm)
Fig. 5. Average excitation spectra of buffer extracts from ad- (FI, arbitrary units) normalized to the tissue weight are
enocarcinoma (A), and non-neoplastic (B) tissue. Emission reported. Non-neoplastic tissue: FI(exc.280 nm, em. 440 nm) - -
wavelengths: 440 nm (a) and 540 nm (b).The curves are 5,000, FI(exc, 330 nm, em, 540 nm) = 3,500; tumor tissue:
normalized to the peak values. The fluorescence intensities ‘‘kx~. 280 om, em. 440 nml = 7,500; F1(exc. 330 nm, em. 540) = 2,800’
lowish granules already described in normal tis- mucosa in either normal tissue or tumor lesion. A
sue. Limited areas of some sections exhibited a slight increase in the relative amplitude of emis-
narrow emission band at 630 nm that can be at- sion in the green-yellowish region was observed
tributed to the presence of endogenous porphy- in the spectra recorded on glands and lamina pro-
rins, possibly related to either an altered metab- pria. The modifications appeared more evident in
olism of the cancer tissue [161 or a microbial the case of tumor lesion, mainly for the stroma,
synthesis [171. where the appearance of the 510-520 nm shoul-
The emission intensities, normalized t o the der was favored (data not shown).
area, are reported in the legend of Figure 4b. The
fluorescence intensity of submucosa of neoplastic Spectrofluorometric Analysis of Tissue Extracts
tissue is comparable to that of the corresponding In an attempt to characterize the excitation
layer of normal tissue, as was that of neoplastic properties of the endogenous fluorophores, a spec-
cells to that of the normal glands. In this latter trofluorometric study has been performed on ex-
case, it must be noted that necrosis, when present, tract solutions of both normal and tumor tissues.
resulted in an appreciable increase in the fluores- Data have been normalized t o both the tissue
cence intensity. As t o the stroma, normal tissues weight, in order t o estimate the total amount of
exhibited fluorescence intensities slightly higher the solved fluorophores, and to the signal ampli-
than tumor, when the measurements were per- tude, in order to evaluate the differences in the
formed in regions without yellow granules. The relative concentration of the f luorophores be-
presence of the granules, which are characterized tween normal and tumor tissues.
by a high fluorescence intensity, resulted in a sig- Peak amplitude-normalized excitation spec-
nal variability that appeared particularly marked tra of extracts obtained from both the lesion and
in the tumor. its normal surrounding tissue are shown in Fig-
When the histological structures were ex- ure 5. The measurement wavelength ranges cho-
cited at 405 nm, a remarkable decrease of the sen were 440 ? 5 and 540 ? 5 nm, approximately
emission amplitude was observed in all the cases corresponding to the main emission band and to
with respect to the excitation at 366 nm that can the shoulders of the emission spectra recorded on
be only partially ascribed to the reduced excita- tissue sections, respectively.
tion intensity in the experimental conditions em- The spectra of Figure 5a indicate that the
ployed. As t o the spectral shape, no significant fluorescence emission at 440 nm on both normal
modification was detected in the case of the sub- and tumor tissues is mainly favoured upon exci-
9. 56 Bottiroli et al.
250 300 350 400 450 500
WAVELENGTH (nm)
Fig. 6. Excitation (a)and emission (b) spectra of pure solu- taining protein, albumin FI = 26. C: Elastin FI = 64. D:
tion of endogenous fluorophores, possibly involved in the Pyridoxal-5-phosphate FI = 22. E: Pyridoxine FI = 590. F:
autofluorescence. Spectral shapes are shown for the best rel- 4-pyridoxic acid FI = 7,960. G: NADH FI = 950. H: Lipo-
ative excitationiemission conditions. Fluorescence emission pigments obtained by chloroformimethanol extraction from
intensity values (FI, arbitrary units) are reported for each rat brain tissue FI = 523. I: IX protoporphyrin FI = 25,400.
compound, normalized to the weight of .01 mgiml, exc. 366 L: flavin-mononucleotide FI = 17,272 or flavin-adenin-dinu-
nm, emission wavelength corresponding to 380 nm for colla- cleotide FI = 1,960. All curves are normalized t o the peak
gen, albumin, and elastin, and at the emission peak for all the values.
other compounds. A: collagen FI = 840.B: Tryptophan-con-
tation at about 280 nm and in the 325-350 nm found that appeared like a fibrous material. The
range. Apart from the relative amplitudes of the microspectrofluorometric analysis performed on
two bands, the spectra are quite similar, thus sug- pellet smears confirms the connective nature of
gesting that the fluorescence emission at 440 nm the unsoluble residuals.
can be ascribed to the same fluorophores in both When the fluorescence intensity values were
normal and tumor tissues. When compared to the normalized to the tissue weight, the normal tis-
excitation spectra of pure solutions (Fig. 6a) and sues gave a signal about 35%less than that of the
to the data reported in the literature on endoge- lesions under excitation at 280 nm, thus indicat-
nous fluorophores, the 280 nm excitation band ing a larger amount of easily soluble proteins in
can be attributed to the aromatic aminoacids of the lesions, in agreement with the fact that the
proteins, namely tryptophan and tyrosine. The unsolved pellet is significantly higher in normal
excitation band at 325-350 nm can be ascribed to than in tumor tissue. On the contrary, when ex-
fluorophores such as pyridoxine derivatives and cited at 330-350 nm, the fluorescence signal was
NAD(P)H, along with the solved fraction of colla- about 20-25% greater in normal than in tumor
gen, that fluoresces at longer wavelengths than tissues. This difference is to be attributed to a
proteins, because of the presence of cross links larger amount of NAD(P)H, as indicated by the
involving tyrosine residues [181. In both normal emission spectra performed under excitation at
and tumor tissues the excitation band at 280 nm 340 nm, that exhibited an emission band at 460
exhibited a greater amplitude than that at 330- nm larger in normal than in tumor tissues (data
350 nm. This can be explained taking into ac- not shown). The larger amount of connective com-
count that collagen, one of the fluorophores con- ponents in normal than in tumor tissues accounts
tributing to the 330-350 nm excitation band, is only partly for the difference, because of its low
poorly soluble. In fact, it must be noted that after solubility.
extraction an amount of unsolved pellet was The excitation spectra measured at 540 nm
10. Microspectrofluorometry of Human Colon 57
were well structured in the 300-450 nm range. In
particular, they appeared as a convolution of a
number of excitation bands centered at about 330,
360, 380, and 440 nm. A minimum was observed
B t about 410 nm that can be partially ascribed to 5
z
W
the reabsorption effect resulting from the pres- z
ence of blood, as confirmed by evaluating the 0
W
amount of hemoglobin in the absorption spectra. 0
Y
I
The excitation spectra indicated that the 0
emission at 540 nm could reasonably be ascribed 3
t o the simultaneous contributions of different flu-
orophores, in agreement with the variability of
the emission spectrum shape observed in tissue D 350 400 450 500 550 600
sections (Fig. 4a,b). After normalization to the tis- WAVELENGTH (nm)
sue weight, the fluorescence intensity under ex-
Fig. 7. Excitation (observed wavelength: 510 nm) and emis-
citation in the 330-350 nm range was found to be sion (excitation wavelength: 370 nmj spectra of a solution
greater in normal than in tumor tissue by about obtained by extraction with ch1oroform:methanol:water solu-
15 -25%. tion (2:l:O.E vol.) of the adenocarcinomapellet, previously
by
The comparison with the spectra of pure submitted to extraction by buffer solution. The curves are
solutions indicated that pyridine derivatives, normalized to the peak values.
NAD(P)H and flavins, which are substances
mainly related to metabolic processes, are respon-
sible for the autofluorescence at 540 nm, along The autofluorescence, excited at 366 nm and
with the collagen. As expected, owing to their analyzed in the 430-700 nm spectral range, ap-
physicochemical nature, lipopigments, which are pears characterized by a main emission band
present t o a greater extent in tumor tissue, were peaking at about 440-460 nm along with the oc-
extracted in a negligible quantity in a polar sol- casional presence of shoulders in the 500-580 nm
vent, so that they were not significantly detect- region. The less structured spectral shape is found
able in the excitation spectra shown in Figure in submucosa that is the collagenous connective
5a,b. The presence of lipopigments was evidenced layer acting as a base of the mucosa in the colon
by performing the extraction with organic sol- organization. As expected, because of its non-tu-
vents. Figure 7 shows the excitation and emission moral nature, submucosa did not exhibit any ap-
spectra recorded on chloroform-methanol extracts preciable difference between normal and tumor
obtained from tumor tissues. These spectra ap- tissues. Only a narrow emission band centered at
peared somewhat different from those obtained about 440 nm was observed, that, owing to the
from purified endogenous fluorophores (Fig. 6). It layer composition, can be ascribed to the contri-
must, however, be noted that lipopigments are a bution of constituent proteins such as collagen
highly heterogeneous group of substances that and elastin, With respect t o the spectral shape
are found in several tissues, mainly in brain, and recorded in submucosa, a slight broadening of the
that are supposed t o be derived from lysosomal band toward longer wavelengths was observed in
material through both oxidation and polymeriza- gland cells of mucosa, both normal and neoplastic.
tion processes. They differ in basic structure, in A shoulder at about 460-470 nm was detected,
degree of polymerization, and in the nature and that can be attributed to the contribution of the
amount of non-lipid material included in the po- coenzyme NAD(P)H, as suggested by the meta-
lymerizing mass [ 191. bolic activity exherted by the cells and supported
by the spectral evidence obtained in both pure
DISCUSSION solutions and tissue extracts. Actually, the con-
tribution of NAD(P)H appears markedly less ev-
The microspectrofluorometric analysis per- ident than that of the proteins. In this regard, it
formed on histological structures and along the must be considered that the tissue processing re-
longitudinal section of “ex vivo” samples indi- quired by the extraction procedure possibly re-
cated that tumor differs from normal tissue both sults in oxidative processes leading t o an under-
in the spectral shape and in the intensity distri- estimation of the NAD(P)H contribution [6].
bution of the autofluorescence. Shoulders in the 480-580 nm region were
11. 58 Bottiroli et al.
observed in the emission spectra recorded in the ganization of the tissue could provide information
stroma, their relative amplitudes varying be- useful for interpreting the results previously ob-
tween tumor and normal tissues. In particular, tained in an “in vivo” study [71. In that study,
these shoulders appeared more evident and the adenocarcinoma was found to differ from the non-
spectral profile was better structured in the tu- neoplastic surrounding tissue in terms of both
mor tissue. The presence of fluorescing granules fluorescence intensity and spectral shape. The
and lipid microdroplets in greater number than in difference of fluorescence intensity concerns the
normal tissue could at least partly explain the emission band at 440-460 nm, which exhibited a
differences observed in that spectral region, as is higher amplitude in normal than in tumor tis-
suggested by the spectra of organic solvent ex- sues. This band, which is the most important com-
tract. ponent of the overall autofluorescence of colonic
Both the complexity of the stromal tissue tissue, was shown to be the major spectral com-
and the processes occurring in this compartment ponent of the emission of all the histological
as a consequence of the host-tumor response [20- structures, and in particular of the submucosa, in
221 could account for these differences. Areas of relation to the density of the strongly fluorescent
inflammation and granulation can be found, with collagen. Taking into account the tissue penetra-
accumulation of interstitial fluid and migrating bility of the light in the range 300-410 nm [6,261,
cells, especially granulocytes containing large it is likely that the difference of location of sub-
numbers of granules and lipid bodies when en- mucosa between normal and tumor tissue results
gaged in an inflammatory response [231. An ex- in a different involvement in the fluorescence ex-
cess of lipid storage can also be found in tumor citation-emission process, thus contributing to
tissue, possibly related to an injury of mitochon- the differences in the intensity of the overall flu-
drial membranes consequent upon a hypoxic con- orescence emission of the two tissues. This is in
dition [241. agreement with Schomacker et al. [6] who found
Other fluorophores, such as flavins, that are that under excitation at 337 nm, a condition par-
known to fluoresce at about 500-530 nm, could ticularly favorable for collagen, the major change
contribute to the modifications observed, owing to in autof luorescence in malignant progression is a
the unbalance of the natural redox equilibrium relative decrease in 390 nm fluorescence, as a
1251. consequence of the screening of fluorescence from
The presence of the shoulders becomes more the collagenous submucosal layer by the thick-
noticeable when the autofluorescence is excited ened mucosa present in polyps. A contribution to
at a longer wavelength, that is, 405 nm instead of fluorescence enhancement in normal tissue is
366 nm. This can be ascribed to a reduction of the also provided by the superficial epithelium, which
relative amplitude of the main band at 440-460 was found more fluorescent than the surface area
nm, rather than t o an actual enhancement of exposed t o the lumen in the tumors. Although the
emission at longer wavelength. In fact, the emis- differences found in the fluorescence intensity are
sion band at 440-460 nm, attributable to the con- rather low and concern a very short thickness, it
tribution of proteins, collagen and NAD(P)H, is must be considered that this layer is deeply in-
particularly favored by exciting in the range 280- volved in the excitation-emission process. A fur-
360 nm, as is indicated by the excitation spectra ther contribution could be provided by the mu-
measured in both pure solution and tissue ex- cosa, which appears somewhat more fluorescent
tracts. in normal than in tumor stroma, because of the
The differences in the fluorescence intensity presence of a fluorescent net not observed in the
pattern are related to the histological organiza- tumor stroma that is attributable to collagen fi-
tion of the tissue and can be attributed mainly to bers. This additional fluorescence, however,
the distribution of submucosa. This layer, which should be balanced by the presence of a larger
is about ten times more fluorescent than the re- amount of green-yellowish highly fluorescent
maining part, in normal tissue is usually located granules in neoplastic than in normal stroma.
at a constant depth not exceeding 450-500 pm, The fluorescence intensity ratio up to 5
while in neoplastic lesions, when present, it starts found between normal and tumor tissues in the
at a depth greater than 800 pm, depending on the “in vivo” measurements, however, suggests that
tumor invasiveness. aspects other than the tissue organization might
The dependence of the autofluorescence be involved. A different relative concentration of
properties on the histological composition and or- fluorophores, and, in particular, a change of
12. Microspectrofluorometry of Human Colon 59
NAD(P)H fluorescence, according to the data ob- modifications between the pre-malignant lesions
tained from tissue extracts, could be hypothesized and the surrounding normal tissue. At present we
in the two tissues. Lesser NAD(P)H fluorescence cannot rule out that this different behavior may
intensity in transformed than in normal cultured be attributed to a remarkable degeneration of the
cells has already been observed in relation to the stroma in lesions passing from the pre-malignant
absolute coenzyme concentration [27]. Moreover, to the malignant condition.
a reduction of binding sites for NAD(P)H in can-
cerous tissue has been observed 1281, resulting in CONCLUSIONS
a lower proportion of the coenzyme form with
higher fluorescence efficiency 1291. The results presented concerning the spec-
This hypothesis is not fully supported by the trof luorometric analysis of “ex vivo” samples sup-
data obtained in tissue sections, which showed port the view that adenocarcinoma lesions can be
only a 10% higher fluorescence in the normal distinguished from normal colonic tissue by both
than in tumor gland cell compartment. It must be, autofluorescence intensity and emission spectral
however, considered that an underestimation of shape. The differences observed on tissue sections
the NAD(P)H contribution occurs in measure- can be related to the presence of various fluoro-
ments in tissue sections. Richards-Korthum et al. phores in terms of both their relative concentra-
[91, by applying a method of investigation based tion and distribution, in connection with the ar-
on fluorescence excitation-emission matrices to rangement and, possibly, the composition of the
“ex vivo” samples, evidenced a peak attributed to histological components. The nature and the ex-
NAD(P)H twice as intense in normal as in ade- tent of the differences, which are consistent with
nomatous tissue. This result is consistent with the data previously obtained “in vivo,” highlight
the findings obtained in patients affected by ade- the role of the excitation wavelength in defining
nomas, by Cothren et al. [41 and by some of the an optimal diagnostic scheme exploiting various
present authors [71, who observed a greater fluo- properties of tissue autof luorescence. In particu-
rescence intensity in surrounding normal tissue lar, the results obtained suggest that excitation at
than in neoplastic lesion, under excitation at 370 short wavelengths (<350 nm), favoring the emis-
and 405 nm, respectively. sion at 440 nm typical of the biological tissues,
The shape differences found in the 500-580 result in differences of fluorescence intensity de-
nm range between the emission spectra of stroma pending on the organization of the colonic tissue
in tumor and those of normal tissue in sections (collagen) and, possibly, on the cell metabolic ac-
could account for the spectral modifications mea- tivity (NAD(P)H).Excitation at long wavelengths
sured “in vivo,” the spectral region involved being (>350 nm), involving fluorophores in some way
the same. No evidence of spectral modification in related to degeneration processes, could enhance
this range is reported in the literature on either the differences in the spectral shape arising from
“ex vivo” or “in vivo,” under excitation in the the tumor host response.
range 320-370 nm. In this context Shomacker et An important factor in the validity of the
al. [61 suggested that by excitation at short wave- excitation wavelength choice would be the mor-
lengths, where the contribution of highly fluores- phological organization of the neoplastic tissue, a
cent collagen from extracellular tissue is favored, characteristic that is among the criteria for the
differences in spectral shape between dysplastic histogenetic classification of tumors. The propor-
and normal cells could be masked. Actually, this tion of tumor cells to stroma, for instance, could
is supported by our measurements on tissue sec- influence the relative contribution of the tissue
tions where a greater emission of the fluoro- structures to the overall autofluorescence. In the
phores related to the longer wavelengths is ob- perspectives of the foregoing considerations, great
tained under 405 nm excitation than at 366 nm. attention must be paid to the excitation-emission
Moreover, the excitation spectra obtained on tis- delivery system, which, depending on the probe
sue extracts showed that the emission at 540 nm geometry, can influence the sensitivity of the re-
can be particularly influenced by the excitation sponse toward the biochemical and morphological
wavelength. properties of the tissue.
It must, however, be noted that measure- The use of a Monte Carlo model and re-
ments performed by the authors “in vivo” on ad- gression analysis, considering the histological or-
enomas in the same excitation conditions used for ganization and photophysical properties of the
adenocarcinomas evidenced only slight spectral biological substrate, will allow us to verify the
13. 80 Bottiroli et al.
experimental conditions that enhance the modifi- 14. Goldfisher S, Villaverde U, Forschirm R. The demonstra-
bation of autofluorescence characteristics, thus tion of acid hydrolase, thermostable reduced diphos-
phopyridine nucleotide tetrazolium reductase and perox-
approaching an “in vivo” histochemical analysis. idase activities in human lipofuscin granules.
Histochem Cytochem 1966; 14:641-652.
15. Greenspan P Mayer E, Fowler SD. Nile red: a selective
ACKNOWLEDGMENTS fluorescent stain for intracellular lipid droplets. J Cell
This work was supported by the CNR Special Biol 1985; 100:975-973.
16. Leibovici L, Schoenfeld N, Yehoshua HA, Mamet R, Ra-
Project “Tecnologie Elettro-Ottiche”. kowsky E, Shindel Asher, Atsmon A. Activity of porpho-
bilinogen deaminase in peripheal blood mononuelear
cells of patients with metastatic cancer. Cancer 1988; 62:
REFERENCES 2297-2300.
1. Muto T, Bussey HJR, Morson BC. The evolution of cancer 17. Harris DM, Weekhaven J. Endogenous porphyrin fluo-
of the colon and rectum. Cancer 1975; 36:2251-2270. rescence in tumors. Lasers Surg Med 1987; 7~467-472.
2. Morson BC, Bussey HJR. Magnitude of risk for cancer in 18. Underfriend S. Proteins and peptides. In: Horecker B,
patients with colorectal adenomas. Br J Surg 1985; 72: Marmur J, Kaplan N, Scheraga HA, eds. “Fluorescence
23-28. Assay in Biology and Medicine, Vol. 11, Molecular Biol-
3. Profio AE. Review of fluorescence diagnosis using por- ogy.” London: Academic Press, 1969, pp 248-283.
phyrins. SPIE 1988; 905:150-156. 19. Wolman M. Lipid pigments (chromolipids): their origin,
4. Cothren RM, Richards-Kortum R, Sivak MV, Fitzmau- nature and significance. Pathobiol Annu 1980; 10:253-
rice M,Rava RP.Gastrointestinal tissue diagnosis by la- 267.
ser-induced fluorescence spectroscopy at endoscopy. En- 20. Goldberg B, Rabonovitch M, Connective tissue. In: Weiss
doscopy 1990; 36:105-111. L, ed. “Cell and Tissue Biology.” Munchen: Urban &
5. Kapadia CR, Cutruzzola WF, O’Brien KM, Stetz ML, En- Schwarzenberg Inc., 1988, pp 157-188.
riquez R, Deckelbaum LI. Laser induced fluorescence 21. Barnes D, Aggarwal S, Thomsen S, Fitzmaurice M,
spectroscopy of human colonic mucosa. Gastroenterology Richards Kortum R. A characterization of the fluorescent
1990; 99~150-157. properties of circulating human eosinophils. Photochem
6. Schomacker KT, Frisoli JK, Compton CC, Flotte TJ, Photobiol 1993; 58297-303.
Richter JM, Nisioka NS, Deutsch TF. Ultraviolet laser- 22. Kaiser HE. Stroma, generally a non-neoplastic structure
induced fluorescence of colonic tissue: basic biology and of the tumor. In: Liotta LA, ed. Influence of Tumor De-
diagnostic potential. Lasers Surg Med 1992; 12:63-78. velopment on the Host. London: Kluwer Academic Pub-
7. Bottiroli G, Marchesini R. Croce AC, Dal Fante M, Cuz- lishers, 1989, pp 1-8.
zoni C , Di Palma S, Spinelli P. Autofluorescence of nor- 23. Weller PF, Ackerman SJ, Nicholson-Weller A, Dvorak
mal and tumor mucosa of human colon. A comprehensive AM. Cytoplasmic lipid bodies of human neutrophilic leu-
analysis. SPIE Proc 1993; 1887:205-212. kocytes. Am J Path01 1989; 135:947-959.
8. Marchesini R, Brambilla M, Pignoli E, Bottiroli G, Croce 24. Freitas I. Lipid accumulation: the common feature to
AC, Dal Fante M, Spinelli P, Di Palma S. Light-induced photosensitizer-retaining normal and malignant tissues.
fluorescence spectroscopy of adenomas, adenocarcinomas J Photochem Photobiol B 1990; 7:359-365.
and non-neoplastic mucosa in human colon. I: In vitro 25. Thorell B. Flow-cytometric monitoring of intracellular
measurements. J Photochem Photobiol B 1992; 14219- flavins simultaneously with NAD(P)H levels. Cytometry
230. 1983; 4~61-65.
9. Richard-Kortum R, Rava RP, Petras RE, Fitzmaurice M, 26. Marchesini R, Pignoli E, Tomatis S, Fumagalli S, Sichi-
Sivak M, Feld MS. Spectroscopic diagnosis of colonic dys- rollo AE, Di Palma S, Croce AC, Bottiroli G. “In vitro”
plasia. Photochem Photobiol 1991; 53:777-786. optical properties of human colon tissues. Lasers Surg
10. Ersson B, Ryden L, Janson JC. Introduction to protein Med (in press).
purification. In: Janson C, Ryden L, eds. “Protein Purifi- 27. Schwartz IP, Possoneau JV, Johnson S, Pasta I. The effect
cation.” New York: VCH Publishers Ine., 1989, pp 3-32. of growth conditions on NAD’ and NADH concentration
11. Armstrong D, Wilhelm J , Smid F, Elleder M. Chroma- and the NAD+:NADH ratio in normal and transformed
tography and spectrofluorometry of brain fluorophors in fibroblasts. J Biol Chem 1974; 249:4138-4143.
neuronaI ceroid lipofuscinosis (NCL). Mech Ageing Dev 28. Galeotti T, Van Rossum GD, Mayer DH, Chance B. On
1992; 64:293-302. the fuorescence of NAD(P)H in whole-cell preparations of
12. Liu CH, Tang GC, Pradhan A, Sha WL, Alfano RR. Ef- tumors and normal tissues. Eur J Biochem 1970; 17:485-
fects of self-absorption by hemoglobins on the fluores- 496.
cence spectra from normal and cancerous tissues. Lasers 29. Salmon JM, Kohen E, Viallet P, Hirschberg JG, Wouters
Life Sci 1989; 3:167-176. AW, Kohen C, Thorell B. Microspectrofluorometric ap-
13. Koenig H. The autofluorescence of lysosomes. Its value proach t o the study offreeibound NAD(P)H ratio as met-
for the identification of lysosomal constituents. J His- abolic indicator in various cell types. Photochem Photo-
tochern Cytochem 1963; 11:556-557. biol 1982: 36:585-593.