0 1992 Wiley-Liss, Inc.
Cytometry 13:329-338 (1992)
Time-Resolved Fluorescence Imaging of Europium Chelate Label in Immunohistochemistry and In Situ Hybridization Lahja Seveus, Mikko Vaisala, Stina Syrjanen, Minna Sandberg, Ari Kuusisto, Raimo Harju, Juha Salo, Ilkka Hemmila, Hannu Kojola, and Erkki Soini' Department of Anatomy, BMC, University of Uppsala, Uppsala, Sweden (L.S.); Wallac Oy, Turku, Finland (M.V., R.H., J.S., I.H., H.K.); Department of Pathology, University of Kuopio, Kuopio, Finland (S.S.);Institute of Biomedicin, University of Turku, Turku, Finland (M.S.); Centre for Biotechnology, Turku, Finland (A.K., E.S.) Received for publication June 4, 1991; accepted October 15, 1991
Fluorescent lanthanide chelates with long decay times allow the suppression of the fast decaying autofluorescence in biological specimens. This property makes lanthanide chelates attractive as labels for fluorescence microscopy. As a consequence of the suppression of the background fluorescence the sensitivity can be increased. We modified a standard epifluorescence microscope for time-resolved fluorescence imaging by adding a pulsed light source and a chopper in the narrow aperture plane. A cooled CCD-camera was used for detection and the images were digitally processed. A fluorescent europium chelate was conjugated to antisera and to streptavidin. These conjugates were used for the localization of tumor associated antigen C242 in the malignant mucosa of human
The availability of bioreagents such a s monoclonal antibodies and nucleic acid probes for biomedical purposes has opened up new possibilities for the localization of proteins and nucleic acid sequences in tissues, cells, and chromosomes. However, the use of these advances in molecular biology places special demands in particular on the sensitivity of microscopic detection of the probe. In microscopy the methods of choice today involve enzymatic, radioactive, and fluorescent labels. Autoradiography provides a high sensitivity, but also has many drawbacks such as long exposure times, inaccurate localization, and environmental and health hazards. Fluorescent and enzyme labels, as nonisotopic methods, on the other hand, offer several advantages over autoradiography but the sensitivity is sometimes
colon, for the localization of type I1 collagen mRNA in developing human cartilaginary growth plates, and for the detection of HPV type specific gene sequences in the squamous epithelium of human cer-
vix. The specific slowly decaying fluorescence of the europium label could be effectively separated from the fast decaying background fluorescence. It was possible to use the europium label at the cell and tissue level and the autofluorescence was effectively suppressed in in situ hybridization and immunohistochemical reactions in both frozen and formaldehydefixed, wax-embedded specimens. 0 1992 Wiley-Liss, Inc.
Key terms: Autofluorescence, lanthanide chelates, CCD-camera, digital imaging, image processing
limited by the background signal emitted by the organic compounds of the biological specimen itself. Autofluorescence of a biological specimen has a short decay time (10 ps) can be observed without interference from autofluorescence. Lanthanide chelates are fluorescent compounds with a decay time in the order of
'Address reprint requests to Dr. Erkki Soini, Centre for Biotechnology, P.O. Box 123, SF-20521, Turku, Finland.
330
SEVEUS ET AL.
S=C=N
/- coo-
N
l
sized from the corresponding amino-compound (12) by a standard procedure (5). Secondary antibodies and streptavidin were labeled in 0.05 M carbonate buffer (pH 9.0) at room temperature for 18 h using a 70100-fold molar excess of the chelate. Labeled proteins were fractionated by Sephadex G-50 chromatography to remove free chelates.
cooFIG.1. Fluorescent europium chelate labeling reagent.
10-1,000 ps. Therefore, the specific signal of lanthanide chelates can be effectively separated from the fast decaying autof luorescence. Time-resolved f luoroimmunoassays employing lanthanide chelates as labels have been successfuly applied to in vitro diagnostics and the technology has been reviewed by Soini and Lovgren (8).Time-resolved f luoroimmunoassays have been found to be comparable to or even more sensitive than radioisotopic assays. Europium chelates have been used in immunohistochemistry and the fluorescence signal has been found comparable with the signal of the conventional fluorochromes in conventional fluorescence microscopy (3,9). Tanke modified a fluorescence microscope to be used in the application of time-resolved imaging for cytological studies (13). He also introduced crystalline phosphors activated by europium. These crystals have the advantage of not fading under excitation. Such crystals have been used for the localization of cell surface antigens (l),but so far no results have been presented concerning their use a s labels in tissue sections. The lanthanides have properties which make them very suited for potential labels for time-resolved fluorescence imaging. The chemistry of the labeling of proteins and nucleic acids with lanthanide chelates has been described previously but it still needs improvements (12). Lanthanide chelates with different properties can be prepared allowing optimization for different purposes. The small size of the label suggests that they can be used as labels for cell surface proteins a s well as for intracellular proteins and nucleic acid sequences. We investigated the use of europium chelates as labels for biospecific probes in time-resolved fluorescence imaging and constructed a work station for this purpose.
MATERIALS AND METHODS To study the relevance of time-resolved fluorescence imaging, we chose applications and specimens which have previously been studied using other labels. The Europium Chelate The stable fluorescent europium chelate of 4-(4-isothio -cyanatophenylethynyl)- 2,6-bis[N,N - bis(carboxymethyl)aminomethyl]-pyridine (Fig. 1) was synthe-
Immunohistochemical Application of Colon Cancer Antigen C242 The specimen. Sections taken from the same tissue block were prepared for comparative study with fluorescein labeling, peroxidase labeling, alkaline phosphatase labeling, and europium labeling. For this purpose sections, 6 pm thick, were cut from fixed (normal buffered formalin), conventionally processed, and waxembedded surgical biopsies from malignant mucosa of human colon. The sections were collected on polylysine coated glass slides and dried. The specimens were dewaxed in xylene for 15 min and then rapidly rehydrated. After rinsing in phosphate-buffered saline (PBS, pH 7.4), the background reaction was blocked with normal rabbit serum (Pharmacia Diagnostics Ab, Uppsala, Sweden) diluted to 1:20 with PBS (30 min). The sections were incubated for 30 min at room temperature with Mab C242, a n antibody to a tumor-associated antigen (Pharmacia Canag, Gothenburg, Sweden) diluted to 2 pgiml with 0.2% BSA in PBS. Finally the sections were rinsed with PBS (3 x 30 m i d . Detection with fluorescein. The sections were incubated with fluorescein-conjugated rabbit anti-mouse antibody (1:40) for 30 min a t room temperature. After rinsing with PBS and distilled water the sections were mounted in Aquamount (BDH Ltd., Poole, England) for immediate observation with the microscope in prompt fluorescence mode. Detection with peroxidase. Sections were incubated with the biotinylated and affinity purified rabbit anti-mouse antibody (Dakopats, Cat. No. E354) a t a concentration of 5 pgiml. The peroxidase conjugated biotin-avidin complex (Dakopats K355) was applied to the sections for 30 min a t room temperature and successively developed with freshly prepared substrate for peroxidase (3-amino-9-ethylcarbazole).After rinsing in water the sections were mounted in glycerol-gelatin for observation with the microscope in absorption mode. Detection with alkaline phosphatase. Sections were incubated with the biotinylated and affinity purified rabbit anti-mouse antibody a s above. The alkaline phosphatase conjugated biotin-avidin complex (Biogenex, USA) was added with the incubation time of 10 min at room temperature. Sections were mounted in Aquamount for microscopic observation with the microscope in absorption mode. Detection with europium. The sections were incubated with europium-labeled rabbit anti-mouse IgG antibody for 30 min at room temperature. After rinsing with PBS and distilled water the sections were
TIME-RESOLVED FLUORESCENCE IMAGING
mounted in Aquamount for immediate observation with the microscope in time resolved fluorescence mode. Negative controls. Firstly, brain tissue sections negative to Mab 242 were treated in the same immunostaining series with all alternative labels. Secondly, positive colon tissue sections was treated in the same immunostaining series with all alternative labels omitting the primary specific antibody 242. The images of negative controls of brain tissue displayed no indication of positive reaction. Epithelial cells of mucosa of colon showed no indication of positive reaction whereas some cells in the connective tissue showed positive reaction with peroxidase staining when the endogenous peroxidase was not removed. In prompt fluorescence images the autof luorescence of the connective tissue was strong. The time-resolved fluorescence images of europium labelled control sections were totally empty. Positive controls. In every staining batch sections were stained by using the peroxidase label which was known to be regularly positive. The tissue block used for this study was originally prepared for a large cross reactivity study for the monoclonal antibody 242 intended for use in human therapy.
331
sections were incubated with europium-labeled streptavidin for 30 min a t room temperature. Finally after rinsing in PBS and distilled water the sections were mounted with Aquamount for immediate observation with the microscope in time-resolved fluorescence mode.
In Situ Hybridization of Cervix Cancer Papillomavirus DNA The specimen. The specimens were prepared for a
comparative study with alkaline phosphatase labeling and europium labeling. For this purpose two consecutive sections 5 p,m thick were cut from each three different formalin-fixed and wax-embedded cervical biopsies. Same blocks were known to be positive for human papillomavirus 11DNA (HPV11). In situ hybridization was performed a s earlier described Syrjanen ( 1 O ) l l ) . Briefly, the method is a s follows: The sections were mounted on slides pretreated with 2% organosilane, dewaxed in xylene, rehydrated through graded ethanols, and deproteinized with proteinase K solution (0.3 mg/ml) in PBS (Boehringer Mannheim, Germany) at 37°C for 15 min. The hybridization cocktail contained 0.5 pm!ml of biotinylated DNA probe (nick translated, whole genomic H P V l l DNA probe, size 150-200 bp) In Situ Hybridization of Type I1 for human papilloma virus 11 DNA in a hybridization Collagen mRNA buffer containing 2 x SSC (0.3 M NaCl and 0.3 M soThe specimen. The specimens were prepared for dium citrate, pH 7.2), 10% dextran sulfate, 0.4 mgiml comparison between radioactively labeled and eu- salmon sperm DNA, and 50% formamide. The HPV ropium labeled probe in in situ hybridizations. For this DNA plasmid was kindly provided by Prof. zur Hausen purpose tissue samples of developing human fingers a t (Deutsche Krebsforschungscentrum, Heidelberg, Ger16 gestational weeks were fixed with formalin and em- many). The tissue DNA and HPV DNA probe cocktail bedded in paraffin for sectioning. The 5 pm sections were simultaneously denatured by heating at 100°C for were taken from the same block for in situ hybridiza- 6 min. The hybridization was carried out at 42°C overtion. Sections were pretreated with proteinase K and night. Following the hybridization the slides were HC1 and acetylated. A 400-bp DraI-EcoRI fragment of washed with 2 x SSC for 20 min a t room temperature, the clone pHCAR3 corresponding to the untranslated 0.2 x SSC for a t 50°C for 20 min, and finally with 2 x region of the alfal(I1) procollagen mRNA (2) was used SSC a t room temperature for 5 min. for hybridization. The hybridizations were carried out Detection with alkaline phosphatase. Half of the at 42°C for 24 h using either probes labeled with 35S- slides were incubated with streptavidin-alkaline phosdeoxy(thio)ATP or biotinylated dUTP, followd by phatase complex for 20 min at 37°C and successively washing. A detailed description of the in situ hybrid- developed with nitroblue tetrazolium and bromoization protocol has been published by Sandberg and chloro-indolyl phosphate for 1 h. The slides were Vuorio (6). washed and mounted in Aquamount for observation Detection with sulfur-35. The probe was labeled with the microscope in absorption mode. with 35S-deoxy(thio) ATP by the random priming Detection with europium. The other half of the method (4). The detection was performed by autoradiog- slides was incubated with europium-labeled streptaviraphy at 4°C for 3 days and the images were produced din for 30 min a t room temperature. After rinsing in with the microscope in dark field mode (6). PBS and distilled water the sections were mounted Detection with europium. The probe was labeled with Aquamount for observation with the microscope with biotinylated dUTP by the random priming in time resolved fluorescence mode. method. The sections were rehydrated in PBS (1:20) for Work Station Concept and Design 30 min for background blocking. Then the sections were incubated with goat anti-biotin IgG antibody (1: A standard epifluorescence microscope (Leitz ARIS1,000) for 30 rnin a t room temperature and rinsed in TOPLAN) was equipped for time-resolved f luorescene PBS (3 x 10 min). After rinsing in PBS (3 x 30 min) imaging. The conventional fluorescence and absorpthe sections were incubated with biotinylated rabbit tion modes were retained so that comparisons with anti-goat IgG antibody (1:300) for 30 min at room tem- other labels could be carried out. We also mounted a n perature and rinsed in PBS (3 x 10 min). Then the epipolarization filter to allow a comparison of eu-
332
SEVEUS ET AL.
Peltler cooled CCD -camera Camera lens
I
a
Computer Image Anaiysis
chopper plate r L
Xe -flash lamp Tungsten lamp
- -k - -
2
= optical
pathway
-
= electrical connection
=>:
= data bus
FIG.2. Schematic presentation of the time-resolved fluorescence microscope.
ropium with gold/silver labeling. Figure 2 shows a path has been opened. The frequency of the flash is 70 Hz . schematic presentation of the work station design. Fluorescent images were detected with a cooled For pulsed excitation, a xenon flash lamp type FX249U with power supply PS-450 (EG&G, Electro-Op- charge coupled photo-optical device (CCD) manufactics Division, Salem, MA) was added as light source tured by Wright Instruments Ltd., UK. The camera (see Fig. 2). This type of lamp was chosen because its has a matrix size of 385 x 578 pixels. The pixel size is spectral and temporal characteristics were suitable for 23 pm, dark counts less than 0.1 electronsipixelis at the detection of europium. It provides 14 percent of the 200"K,and the quantum efficiency of the CCD is about light output between 300-400 nm and the pulse length 30% a t the emission wavelength of europium. The dynamic range of the camera is over 4 decades, which below 10 ps. To separate the europium emission from the signal of makes i t possible to analyze samples containing both the background or other fluorochromes, the detection very weak and intense parts. The camera was conof the europium signal has to be delayed. The delay trolled by a personal computer equipped with a n 80386 time after excitation should be longer than the decay CPU. The work station had a n optical output to a n additime of the autofluorescent compounds and that of the optical parts (>50 ~ s but ) significanly shorter (not tional film or video camera which could be used to more than half) than the decay time of the fluorescent record colour images in the fluorescence or absorption label. Since the decay time of europium is long (700 ps) microscopy mode. The excitation band of europium is between 320 and the delay can be carried out with a mechanical chopper. A rotating chopper plate is placed in the emission light 380 nm with a n absorption maximum a t 340 nm. The path at the ocular exit pupil of the microscope (also excitation light path of the conventional fluorescence called Ramsden disc) and to block the emission light microscope had to be optimized a t these wavelengths. during the excitation pulse and during the time delay These modifications improved the light transmission (200 ps). After the delay the light path is opened (for by one order of magnitude. The details will be pub2,000 ps) for image recording. The prompt fluorescence lished later. Measuring conditions. The samples were anacan be visualized in real time through oculars but not lyzed under the following conditions: the time-resolved fluorescence. To obtain a prompt fluorescence signal of europium using the flash lamp, a n electronic triggering device Excitation wavelength: 320-400 nm by using a coloured glass filter U G l l (SCHOTT) and a cutoff filter was introduced to synchronize the excitation flash and WG 320 (SCHOTT); the rotating disc so that the flash occurs when the light
TIME-RESOLVED FLIJORESCENCE IMAGING
* l o 3 ictsi
10-
x 1 ~ 3ictsi
333
4
13-
J
21-
FIG.3. Rhodamine salt and europium-containing yttrium oxisulfide particles. A: Prompt fluorescence mode. B: Time-resolved fluorescence mode. The prompt fluorescence from the rhodamine salt (see arrow R1 in Fig. 3A and C ) was effectively suppressed in the timeresolved mode (see arrow K2 in Fig. 3B and D). Intensity counts (cts)
per camera picture element (pixels) are shown as profiles along horizontal lines in C for prompt mode and D for time-resolved mode (profiles L l and L2, respectively). Small arrows indicate signal values from single europium crystals. Bars on the lower right hand corners mark 100 Lm.
Excitatior pulse length: 10 ps; Pulse energy: 0.4 Jipulse; Delay time after excitation: 200 ps; Emission filter: bandpass filter 615/10/80 (Ferroperm); Exposure time: typically 30 s.
cused either with phase contrast or with bright field mode using a halogen light source and a red band pass filter. All measurements were made in a dark laboratory room. Image processing was performed with the following computer programs: SAO Image (Smithsonian Astrophysical Observatory, USA), AT 1 (Wright Instruments Ltd., England), and Lineprof 5 (Wallac Oy, Finland, not commercially available). For each image, 500 kBytes were needed. The original image data from the CCD camera was read out in digital form and stored in the computer memory. The images shown in this paper are true images without “video enhancing” but pseudo colors and con-
In the prompt fluorescence mode the delay time after excitation was zero, and the excitation wavelength and the excitation pulse length were as given above for the time-resolved imaging mode. The excitation light source was a xenon flash lamp but a mercury lamp with FITC-optimized filter block (Leitz 13)was used for FITClabeled samples. The exposure times were from 1to 30 s. Before time-resolved imaging, the specimen was fo-
SEVEUS ET AL.
334
4c
100
150
200
250
HORIZONTAL
300
3kO [p’xdsl
FIG 4. Localization of C242 antigen in the epithelial cells of malignant mucosa of human colon. A: Europium labeling with prompt mode. B: Europium labeling with time-resolved mode. Intensity counts (cts) vs. camera picture element (pixels) are shown as profiles along horizontal lines in C for prompt mode and D for time-resolved mode (profiles M1 and M2, respectively). The arrows indicate exam-
4D
100
150
200 250 HORIZONTAL
300
350 ‘p’xds’
ples of areas where the europium fluorescence is most intense (see arrow S1 in Fig. 4A and C and arrow S3 in Fig 4B and D) and other areas where autofluorescence is usually observed (see arrow S2 in Fig. 4A and C and arrow S4 in Fig. 4B and D).Bars on the lower right hand corners mark 25 bm.
colon with europium labels. Sections were cut from a n aldehyde-fixed and wax-embedded surgical biopsy. The digital image recorded and shown in Figure 4A is RESULTS taken in the prompt mode and reveals a strong tissue Technical Performance autofluorescence. Figure 4B shows the same area in To verify the performance of the microscope with re- the time-resolved mode. The arrows S2 and S4 indicate gard to suppression of prompt fluorescence, salts of examples of areas where the autofluorescence was suprhodamine and europium-activated yttrium oxisulfide pressed. Sections from similar specimens are shown with percrystals were analyzed. Figure 3A-D shows the effect of suppression of the prompt fluorescence from the rho- oxidase labeling in Figure 5A, with alkaline phosdamine salt. On the basis of corresponding digital data phatase labeling in Figure 5B, with fluorescein labelthe suppression of the prompt fluorescence in the time- ing in Figure 5C, with europium labeling with prompt resolved mode was more than three orders of magni- mode in Figure 5D, and with europium labeling with time-resolved mode in Figure 5E. Arrows indicate the tude. areas where strong autofluorescence of the aldehydeBiological Applications fixed tissue can be easily observed. Localization of type I1 collagen mRNA in cartiTime-resolved fluorescence imaging of intraand extracellular antigens in tissue specimens laginous growth plate by using europium-labeled with europium-labeled and enzyme-labeled cDNA probe. In situ hybridization using either 35Sprobes. Figure 4A and B shows, localization of C242 labeling and autoradiography (Fig. 6A) or biotinylated in the epithelial cells of malignant mucosa of human probes and time-resolved fluorescence imaging of eu-
trast enhancement have been applied using a computer.
TIME-RESOLVED FLUORESCENCE IMAGING
FIG.5. Localization of C242 antigen in the epithelial cells of malignant mucosa of human colon. A: Peroxidase labeling. B: Alkaline phosphatase labeling. C: Fluorescein labeling. D Europium labeling with prompt mode. E. Europium labeling with time-resolved mode. The arrows indicate examples of areas where the autofluorescence
335
was suppressed (see arrows in Fig. 5D and E). Color bar shown in Figure 5E indicates intensity values in linear scale in artificial colours in Figure 5C-E (in arbitrary units). Bars on the lower right hand corners mark 25 pm.
336
SEVEUS ET AI,.
FIG 6. In situ hybridization of type I1 collagen mRNAs in cartilaginous growth plate by radioactively labeled probe detected by autoradiography and dark field microscopy (A! and by biotinylated probe detected by fluorescent europium chelate and time-resolved microscopy (B).The different zones of cartilage from left to right are hypertrophic chondrocytes, proliferative chondrocytes, and reserve chondrocytes.
ropium (Fig. 6B) for the detection of type I1 collagen mRNA in the developing human growth plate gave similar distributions of the positive cells, the signal accentuating in the lower proliferative and upper hypertrophic chondrocytes by both detection methods. Localization of viral genomes in tissue specimens by using europium-labeled gene sequences. The possibility to use europium chelates a s labels for in situ hybridization in conventionally processed tissue specimens was investigated in a set of experiments with biotinylated H P V l l DNA in surgical biopsies from human cervix known to be positive for H P V l l DNA. Figure 7 shows positive results with the in situ hybridization method. Figure 7A shows a section where the biotinylated hybrids were detected by the substrate reaction of alkaline phosphatase conjugated t o streptavidin. Figure 7B shows a subsequent section with europium-labeled streptavidin where the image was recorded in the time-resolved mode. Both methods gave similar frequency and distribution of the positive cells.
DISCUSSION Fluorescence methods are widely used in microscopy. The sensitivity of fluorescence microscopy is, however, serviously limited by the autofluorescence of the bio-
logical object. The present study shows that lanthanide chelates are suitable markers for time-resolved fluorescence microscopy and that time-resolved fluorescence imaging offers a unique method to suppress the background fluorescence. Lanthanide chelates have many advantages. Unlike the chelates used for certain in vitro assays, the europium chelate tested in this study is directly fluorescent. The fluorescence of this chelate is a result of excitation absorption of the organic part of the molecule (phenyl-ethyl-pyridine), followed by efficient energy transfer from the excited ligand triplet state to complexed europium, which produces the emission typical of metal ions. Pyridine nitrogen and four carboxylic acids in the chelate form the required thermodynamic stability and protect the emitting ion from the quenching effect of water molecules. Therefore, there is no need to dissociate the label from the antigen-antibody complex after immunoreaction. Hence, this chelate can be used as a label for localization of antigens, mRNAs, and gene sequences a t the cell and tissue levels. We have verified earlier that time resolution provides a very high separation efficiency between two signals with substantially different decay times (7). This experiment was carried out using a time-resolved fluorometer for test cuvettes which was optimized for
TIME-RESOLVED FLUORESCENCE IMAGING
337
FIG.7. Localization of H P V l l DNA sequences in formalin-fixed and wax-embedded cervical biopsy using europium-labeled DNA probe. A: Alkaline phosphatase labeling with absorption mode. B: Europium labeling with time-resolved mode. Bars on the lower right hand corners mark 25 pm.
europium using a delay time of 400 p s and a window time of 500 ps, and an emission filter optimized for the europium emission peak. The detection sensitivity of europium using these settings was better than 0.1 pmoliliter. Fluorescein and rhodamine samples (decay time below 10 ns) in increasing concentrations were measured with the same settings and cuvette. It was found that the signal from fluorescein and rhodamine with concentrations up to 1 mmoliliter and 0.01 mmoli liter, respectively, was zero. It was concluded that europium can be measured in the presence of a millionfold excess of a short decay time fluorescence substance which can be either a second label or background substance. The time-resolved mode allows multiparameter analysis in several ways. Not only could different lanthanide chelates (terbium or samarium) with different wavelengths be used but also markers with the same wavelength but substantially different decay times (e.g., conventional fluorochromes and fluorescent stains). The combination of an organic fluorescent stain and europium label would allow the simultaneous analysis of DNA content and an antibody, which
can be of interest in tumor pathology. Since the long decay fluorescence of europium can be detected in the presence of strong autofluorescence we could apply the lanthanide chelates both with aldehyde-fixed wax-embedded specimens and with frozen specimens. When using the conventional organic f luorochromes in aldehyde-fixed and wax-embedded tissue the sensitivity is often hampered by the background autofluorescence. In contrast to the crystalline labels, the size of the lanthanide chelates is small. They can therefore be used for the localization of both cell surface and intracellular antigens. Moreover, unlike the crystalline labels, the chelates are not sticky, which facilitates the labeling of sections or cells on the glass slide. The lanthanides are present in the biological specimen in negligible amounts. Therefore, the problems inherent to the endogenous presence of the label, common to enzyme labels, are avoided. The labeling procedure is a one or two step procedure comparable to other fluorescence methods. The cell and tissue structures are well preserved also in frozen sections, since in contrast to methods involving enzyme labels, no aggressive compounds are applied to lanthanide chelate
338
SEVEUS ET AL.
labeling. In addition, none of the reagents in the europium labeling procedure causes health or environmental hazards. The problems associated with the lanthanide chelate label are its instability under excitation and in connection with a different reagents for counterstaining (e.g., hematoxylinieosin) used in microscopy. The chemical characteristics of the chelates, however, allow changes in their chemical structure to be made to optimize them for different purposes. It is not unlikely that the structure of the present chelate can be changed to impove its stability. Furthermore, the europium chelate used in this study has a low overall fluorescence intensity. Our results show, however, that despite this, objects which are known to be present in a low concentration in the cell can be visualized in the time-resolved mode equally well as with other fluorescent labels. We have verified that the time-resolved fluorescence imaging is avery fast methodwhencomparedwith theuseofautoradiography. In addition autofluorescence and background fluorescence can be effectively eliminated, but the time-resolved fluorescence microscopy does not yet provide significantly better sensitivity than microscopy with conventional fluorescent labels. Further improvements in the chemistry of the chelates and image intensifying methods are needed and progress in those fields will make this method even more powerful in biology and pathology. Digital image processing is a potential possibility for automation of the microscopic studies in diagnostic pathology and various research fields. The successful use of digital image processing in routine studies requires, however, images with good contrast and easily distinguishable objects because it is difficult for image processing algorithms to descriminate between specific points and artefacts. Elimination of background and artefacts by time-resolved imaging makes the use of digital image processing more useful and should improve the reliability of the results.
ACKNOWLEDGMENTS The authors are grateful to Prof. Mikko Niemi for the possibility of using the facilities of the D e p a r t m e n t of Anatomy, University of Turku, for these studies.
LITERATURE CITED 1. Beverloo HB, van Schadewijk A, van Gelderen-Boele S, Tanke HJ: Inorganic phosphors as new luminescent labels for irnmunocytochemistry and time-resolved microscopy. Cytometry 11 :784792, 1990. 2. Elima K, Vuorio T, Vuorio E: Determination of the single polyadenylation site of the human type I1 collagen gene. Nucleic Acids Res 15:9499-9504, 1987. 3. Elster AD, Jackels SC, Allen NS, Marrache RC: EuropiumDTPA: A gadolinium analogue traceable by fluorescence microscopy. AJNR 10:1138-1144, 1989. 4. Feinberg A, Vogelstein B: A technique for radio-labeling DNA restriction endonuclease fragments to high specific activity. Anal Biochem 132:6-13, 1983. 5. Mukkala V-M, Mikola H, Hemmila I: The synthesis and use of activated N-benzyl derivatives of diethylenetriaminetetraacetic acids: Alternative reagents for labeling of antibodies with metal ions. Anal Biochem 176:319-325, 1989. 6. Sandberg M, Vuorio E: Localization of type I, I1 and I11 collagen mRNAs in developing human skeletal tissue by in situ hybridization. J Cell Biol 104:1077-1084, 1987. 7. Soini EJ, Hemrnila IA, Dahlen P: Time-resolved fluorescence for immunoassays and DNA-probes. In: Automation and New Technology in the Clinical Laboratory, Okuda K (ed). Blackwell Scientific Publications, Oxford, 1990, pp 63-67. 8. Soini E J , Lovgren T Time-resolved fluorescence of lanthanide probes and applications in biotechnology. CRC Crit Rev Anal Chem 18:105-154, 1987. 9. Soini EJ, Pelliniemi U,Hemmila IA, Mukkala V-M, Kankare JJ, Frojdman K: Lanthanide chelates as new fluorochrome labels for cytochemistry. J Histochem Cytochem 36:1449-1451, 1988. 10. Syrjiinen SM: Basic concepts and practical applications of recombinant DNA techniques in detection of human papillomavirus (HPV) infections. APMIS 98:95-110, 1990. 11. Syrjanen S, Partanen P, Mantyjarvi R, Syrjanen K: Sensitivity of in situ hybridization techniques using biotin- and 35S-labeied human papillomavirus (HPV) DNA probes. J Virol Methods 19:225238, 1988. 12. Takalo H: Synthesis of complexing compounds containing a substituted 4-ethylpyridine subunit. Thesis, University of Turku, Turku, Finland, 1988. 13. Tanke HJ: Does light microscopy have a future? J Microsc 155: 405-418, 1989.
Cytometry 13:329-338 (1992)
Time-Resolved Fluorescence Imaging of Europium Chelate Label in Immunohistochemistry and In Situ Hybridization Lahja Seveus, Mikko Vaisala, Stina Syrjanen, Minna Sandberg, Ari Kuusisto, Raimo Harju, Juha Salo, Ilkka Hemmila, Hannu Kojola, and Erkki Soini' Department of Anatomy, BMC, University of Uppsala, Uppsala, Sweden (L.S.); Wallac Oy, Turku, Finland (M.V., R.H., J.S., I.H., H.K.); Department of Pathology, University of Kuopio, Kuopio, Finland (S.S.);Institute of Biomedicin, University of Turku, Turku, Finland (M.S.); Centre for Biotechnology, Turku, Finland (A.K., E.S.) Received for publication June 4, 1991; accepted October 15, 1991
Fluorescent lanthanide chelates with long decay times allow the suppression of the fast decaying autofluorescence in biological specimens. This property makes lanthanide chelates attractive as labels for fluorescence microscopy. As a consequence of the suppression of the background fluorescence the sensitivity can be increased. We modified a standard epifluorescence microscope for time-resolved fluorescence imaging by adding a pulsed light source and a chopper in the narrow aperture plane. A cooled CCD-camera was used for detection and the images were digitally processed. A fluorescent europium chelate was conjugated to antisera and to streptavidin. These conjugates were used for the localization of tumor associated antigen C242 in the malignant mucosa of human
The availability of bioreagents such a s monoclonal antibodies and nucleic acid probes for biomedical purposes has opened up new possibilities for the localization of proteins and nucleic acid sequences in tissues, cells, and chromosomes. However, the use of these advances in molecular biology places special demands in particular on the sensitivity of microscopic detection of the probe. In microscopy the methods of choice today involve enzymatic, radioactive, and fluorescent labels. Autoradiography provides a high sensitivity, but also has many drawbacks such as long exposure times, inaccurate localization, and environmental and health hazards. Fluorescent and enzyme labels, as nonisotopic methods, on the other hand, offer several advantages over autoradiography but the sensitivity is sometimes
colon, for the localization of type I1 collagen mRNA in developing human cartilaginary growth plates, and for the detection of HPV type specific gene sequences in the squamous epithelium of human cer-
vix. The specific slowly decaying fluorescence of the europium label could be effectively separated from the fast decaying background fluorescence. It was possible to use the europium label at the cell and tissue level and the autofluorescence was effectively suppressed in in situ hybridization and immunohistochemical reactions in both frozen and formaldehydefixed, wax-embedded specimens. 0 1992 Wiley-Liss, Inc.
Key terms: Autofluorescence, lanthanide chelates, CCD-camera, digital imaging, image processing
limited by the background signal emitted by the organic compounds of the biological specimen itself. Autofluorescence of a biological specimen has a short decay time (10 ps) can be observed without interference from autofluorescence. Lanthanide chelates are fluorescent compounds with a decay time in the order of
'Address reprint requests to Dr. Erkki Soini, Centre for Biotechnology, P.O. Box 123, SF-20521, Turku, Finland.
330
SEVEUS ET AL.
S=C=N
/- coo-
N
l
sized from the corresponding amino-compound (12) by a standard procedure (5). Secondary antibodies and streptavidin were labeled in 0.05 M carbonate buffer (pH 9.0) at room temperature for 18 h using a 70100-fold molar excess of the chelate. Labeled proteins were fractionated by Sephadex G-50 chromatography to remove free chelates.
cooFIG.1. Fluorescent europium chelate labeling reagent.
10-1,000 ps. Therefore, the specific signal of lanthanide chelates can be effectively separated from the fast decaying autof luorescence. Time-resolved f luoroimmunoassays employing lanthanide chelates as labels have been successfuly applied to in vitro diagnostics and the technology has been reviewed by Soini and Lovgren (8).Time-resolved f luoroimmunoassays have been found to be comparable to or even more sensitive than radioisotopic assays. Europium chelates have been used in immunohistochemistry and the fluorescence signal has been found comparable with the signal of the conventional fluorochromes in conventional fluorescence microscopy (3,9). Tanke modified a fluorescence microscope to be used in the application of time-resolved imaging for cytological studies (13). He also introduced crystalline phosphors activated by europium. These crystals have the advantage of not fading under excitation. Such crystals have been used for the localization of cell surface antigens (l),but so far no results have been presented concerning their use a s labels in tissue sections. The lanthanides have properties which make them very suited for potential labels for time-resolved fluorescence imaging. The chemistry of the labeling of proteins and nucleic acids with lanthanide chelates has been described previously but it still needs improvements (12). Lanthanide chelates with different properties can be prepared allowing optimization for different purposes. The small size of the label suggests that they can be used as labels for cell surface proteins a s well as for intracellular proteins and nucleic acid sequences. We investigated the use of europium chelates as labels for biospecific probes in time-resolved fluorescence imaging and constructed a work station for this purpose.
MATERIALS AND METHODS To study the relevance of time-resolved fluorescence imaging, we chose applications and specimens which have previously been studied using other labels. The Europium Chelate The stable fluorescent europium chelate of 4-(4-isothio -cyanatophenylethynyl)- 2,6-bis[N,N - bis(carboxymethyl)aminomethyl]-pyridine (Fig. 1) was synthe-
Immunohistochemical Application of Colon Cancer Antigen C242 The specimen. Sections taken from the same tissue block were prepared for comparative study with fluorescein labeling, peroxidase labeling, alkaline phosphatase labeling, and europium labeling. For this purpose sections, 6 pm thick, were cut from fixed (normal buffered formalin), conventionally processed, and waxembedded surgical biopsies from malignant mucosa of human colon. The sections were collected on polylysine coated glass slides and dried. The specimens were dewaxed in xylene for 15 min and then rapidly rehydrated. After rinsing in phosphate-buffered saline (PBS, pH 7.4), the background reaction was blocked with normal rabbit serum (Pharmacia Diagnostics Ab, Uppsala, Sweden) diluted to 1:20 with PBS (30 min). The sections were incubated for 30 min at room temperature with Mab C242, a n antibody to a tumor-associated antigen (Pharmacia Canag, Gothenburg, Sweden) diluted to 2 pgiml with 0.2% BSA in PBS. Finally the sections were rinsed with PBS (3 x 30 m i d . Detection with fluorescein. The sections were incubated with fluorescein-conjugated rabbit anti-mouse antibody (1:40) for 30 min a t room temperature. After rinsing with PBS and distilled water the sections were mounted in Aquamount (BDH Ltd., Poole, England) for immediate observation with the microscope in prompt fluorescence mode. Detection with peroxidase. Sections were incubated with the biotinylated and affinity purified rabbit anti-mouse antibody (Dakopats, Cat. No. E354) a t a concentration of 5 pgiml. The peroxidase conjugated biotin-avidin complex (Dakopats K355) was applied to the sections for 30 min a t room temperature and successively developed with freshly prepared substrate for peroxidase (3-amino-9-ethylcarbazole).After rinsing in water the sections were mounted in glycerol-gelatin for observation with the microscope in absorption mode. Detection with alkaline phosphatase. Sections were incubated with the biotinylated and affinity purified rabbit anti-mouse antibody a s above. The alkaline phosphatase conjugated biotin-avidin complex (Biogenex, USA) was added with the incubation time of 10 min at room temperature. Sections were mounted in Aquamount for microscopic observation with the microscope in absorption mode. Detection with europium. The sections were incubated with europium-labeled rabbit anti-mouse IgG antibody for 30 min at room temperature. After rinsing with PBS and distilled water the sections were
TIME-RESOLVED FLUORESCENCE IMAGING
mounted in Aquamount for immediate observation with the microscope in time resolved fluorescence mode. Negative controls. Firstly, brain tissue sections negative to Mab 242 were treated in the same immunostaining series with all alternative labels. Secondly, positive colon tissue sections was treated in the same immunostaining series with all alternative labels omitting the primary specific antibody 242. The images of negative controls of brain tissue displayed no indication of positive reaction. Epithelial cells of mucosa of colon showed no indication of positive reaction whereas some cells in the connective tissue showed positive reaction with peroxidase staining when the endogenous peroxidase was not removed. In prompt fluorescence images the autof luorescence of the connective tissue was strong. The time-resolved fluorescence images of europium labelled control sections were totally empty. Positive controls. In every staining batch sections were stained by using the peroxidase label which was known to be regularly positive. The tissue block used for this study was originally prepared for a large cross reactivity study for the monoclonal antibody 242 intended for use in human therapy.
331
sections were incubated with europium-labeled streptavidin for 30 min a t room temperature. Finally after rinsing in PBS and distilled water the sections were mounted with Aquamount for immediate observation with the microscope in time-resolved fluorescence mode.
In Situ Hybridization of Cervix Cancer Papillomavirus DNA The specimen. The specimens were prepared for a
comparative study with alkaline phosphatase labeling and europium labeling. For this purpose two consecutive sections 5 p,m thick were cut from each three different formalin-fixed and wax-embedded cervical biopsies. Same blocks were known to be positive for human papillomavirus 11DNA (HPV11). In situ hybridization was performed a s earlier described Syrjanen ( 1 O ) l l ) . Briefly, the method is a s follows: The sections were mounted on slides pretreated with 2% organosilane, dewaxed in xylene, rehydrated through graded ethanols, and deproteinized with proteinase K solution (0.3 mg/ml) in PBS (Boehringer Mannheim, Germany) at 37°C for 15 min. The hybridization cocktail contained 0.5 pm!ml of biotinylated DNA probe (nick translated, whole genomic H P V l l DNA probe, size 150-200 bp) In Situ Hybridization of Type I1 for human papilloma virus 11 DNA in a hybridization Collagen mRNA buffer containing 2 x SSC (0.3 M NaCl and 0.3 M soThe specimen. The specimens were prepared for dium citrate, pH 7.2), 10% dextran sulfate, 0.4 mgiml comparison between radioactively labeled and eu- salmon sperm DNA, and 50% formamide. The HPV ropium labeled probe in in situ hybridizations. For this DNA plasmid was kindly provided by Prof. zur Hausen purpose tissue samples of developing human fingers a t (Deutsche Krebsforschungscentrum, Heidelberg, Ger16 gestational weeks were fixed with formalin and em- many). The tissue DNA and HPV DNA probe cocktail bedded in paraffin for sectioning. The 5 pm sections were simultaneously denatured by heating at 100°C for were taken from the same block for in situ hybridiza- 6 min. The hybridization was carried out at 42°C overtion. Sections were pretreated with proteinase K and night. Following the hybridization the slides were HC1 and acetylated. A 400-bp DraI-EcoRI fragment of washed with 2 x SSC for 20 min a t room temperature, the clone pHCAR3 corresponding to the untranslated 0.2 x SSC for a t 50°C for 20 min, and finally with 2 x region of the alfal(I1) procollagen mRNA (2) was used SSC a t room temperature for 5 min. for hybridization. The hybridizations were carried out Detection with alkaline phosphatase. Half of the at 42°C for 24 h using either probes labeled with 35S- slides were incubated with streptavidin-alkaline phosdeoxy(thio)ATP or biotinylated dUTP, followd by phatase complex for 20 min at 37°C and successively washing. A detailed description of the in situ hybrid- developed with nitroblue tetrazolium and bromoization protocol has been published by Sandberg and chloro-indolyl phosphate for 1 h. The slides were Vuorio (6). washed and mounted in Aquamount for observation Detection with sulfur-35. The probe was labeled with the microscope in absorption mode. with 35S-deoxy(thio) ATP by the random priming Detection with europium. The other half of the method (4). The detection was performed by autoradiog- slides was incubated with europium-labeled streptaviraphy at 4°C for 3 days and the images were produced din for 30 min a t room temperature. After rinsing in with the microscope in dark field mode (6). PBS and distilled water the sections were mounted Detection with europium. The probe was labeled with Aquamount for observation with the microscope with biotinylated dUTP by the random priming in time resolved fluorescence mode. method. The sections were rehydrated in PBS (1:20) for Work Station Concept and Design 30 min for background blocking. Then the sections were incubated with goat anti-biotin IgG antibody (1: A standard epifluorescence microscope (Leitz ARIS1,000) for 30 rnin a t room temperature and rinsed in TOPLAN) was equipped for time-resolved f luorescene PBS (3 x 10 min). After rinsing in PBS (3 x 30 min) imaging. The conventional fluorescence and absorpthe sections were incubated with biotinylated rabbit tion modes were retained so that comparisons with anti-goat IgG antibody (1:300) for 30 min at room tem- other labels could be carried out. We also mounted a n perature and rinsed in PBS (3 x 10 min). Then the epipolarization filter to allow a comparison of eu-
332
SEVEUS ET AL.
Peltler cooled CCD -camera Camera lens
I
a
Computer Image Anaiysis
chopper plate r L
Xe -flash lamp Tungsten lamp
- -k - -
2
= optical
pathway
-
= electrical connection
=>:
= data bus
FIG.2. Schematic presentation of the time-resolved fluorescence microscope.
ropium with gold/silver labeling. Figure 2 shows a path has been opened. The frequency of the flash is 70 Hz . schematic presentation of the work station design. Fluorescent images were detected with a cooled For pulsed excitation, a xenon flash lamp type FX249U with power supply PS-450 (EG&G, Electro-Op- charge coupled photo-optical device (CCD) manufactics Division, Salem, MA) was added as light source tured by Wright Instruments Ltd., UK. The camera (see Fig. 2). This type of lamp was chosen because its has a matrix size of 385 x 578 pixels. The pixel size is spectral and temporal characteristics were suitable for 23 pm, dark counts less than 0.1 electronsipixelis at the detection of europium. It provides 14 percent of the 200"K,and the quantum efficiency of the CCD is about light output between 300-400 nm and the pulse length 30% a t the emission wavelength of europium. The dynamic range of the camera is over 4 decades, which below 10 ps. To separate the europium emission from the signal of makes i t possible to analyze samples containing both the background or other fluorochromes, the detection very weak and intense parts. The camera was conof the europium signal has to be delayed. The delay trolled by a personal computer equipped with a n 80386 time after excitation should be longer than the decay CPU. The work station had a n optical output to a n additime of the autofluorescent compounds and that of the optical parts (>50 ~ s but ) significanly shorter (not tional film or video camera which could be used to more than half) than the decay time of the fluorescent record colour images in the fluorescence or absorption label. Since the decay time of europium is long (700 ps) microscopy mode. The excitation band of europium is between 320 and the delay can be carried out with a mechanical chopper. A rotating chopper plate is placed in the emission light 380 nm with a n absorption maximum a t 340 nm. The path at the ocular exit pupil of the microscope (also excitation light path of the conventional fluorescence called Ramsden disc) and to block the emission light microscope had to be optimized a t these wavelengths. during the excitation pulse and during the time delay These modifications improved the light transmission (200 ps). After the delay the light path is opened (for by one order of magnitude. The details will be pub2,000 ps) for image recording. The prompt fluorescence lished later. Measuring conditions. The samples were anacan be visualized in real time through oculars but not lyzed under the following conditions: the time-resolved fluorescence. To obtain a prompt fluorescence signal of europium using the flash lamp, a n electronic triggering device Excitation wavelength: 320-400 nm by using a coloured glass filter U G l l (SCHOTT) and a cutoff filter was introduced to synchronize the excitation flash and WG 320 (SCHOTT); the rotating disc so that the flash occurs when the light
TIME-RESOLVED FLIJORESCENCE IMAGING
* l o 3 ictsi
10-
x 1 ~ 3ictsi
333
4
13-
J
21-
FIG.3. Rhodamine salt and europium-containing yttrium oxisulfide particles. A: Prompt fluorescence mode. B: Time-resolved fluorescence mode. The prompt fluorescence from the rhodamine salt (see arrow R1 in Fig. 3A and C ) was effectively suppressed in the timeresolved mode (see arrow K2 in Fig. 3B and D). Intensity counts (cts)
per camera picture element (pixels) are shown as profiles along horizontal lines in C for prompt mode and D for time-resolved mode (profiles L l and L2, respectively). Small arrows indicate signal values from single europium crystals. Bars on the lower right hand corners mark 100 Lm.
Excitatior pulse length: 10 ps; Pulse energy: 0.4 Jipulse; Delay time after excitation: 200 ps; Emission filter: bandpass filter 615/10/80 (Ferroperm); Exposure time: typically 30 s.
cused either with phase contrast or with bright field mode using a halogen light source and a red band pass filter. All measurements were made in a dark laboratory room. Image processing was performed with the following computer programs: SAO Image (Smithsonian Astrophysical Observatory, USA), AT 1 (Wright Instruments Ltd., England), and Lineprof 5 (Wallac Oy, Finland, not commercially available). For each image, 500 kBytes were needed. The original image data from the CCD camera was read out in digital form and stored in the computer memory. The images shown in this paper are true images without “video enhancing” but pseudo colors and con-
In the prompt fluorescence mode the delay time after excitation was zero, and the excitation wavelength and the excitation pulse length were as given above for the time-resolved imaging mode. The excitation light source was a xenon flash lamp but a mercury lamp with FITC-optimized filter block (Leitz 13)was used for FITClabeled samples. The exposure times were from 1to 30 s. Before time-resolved imaging, the specimen was fo-
SEVEUS ET AL.
334
4c
100
150
200
250
HORIZONTAL
300
3kO [p’xdsl
FIG 4. Localization of C242 antigen in the epithelial cells of malignant mucosa of human colon. A: Europium labeling with prompt mode. B: Europium labeling with time-resolved mode. Intensity counts (cts) vs. camera picture element (pixels) are shown as profiles along horizontal lines in C for prompt mode and D for time-resolved mode (profiles M1 and M2, respectively). The arrows indicate exam-
4D
100
150
200 250 HORIZONTAL
300
350 ‘p’xds’
ples of areas where the europium fluorescence is most intense (see arrow S1 in Fig. 4A and C and arrow S3 in Fig 4B and D) and other areas where autofluorescence is usually observed (see arrow S2 in Fig. 4A and C and arrow S4 in Fig. 4B and D).Bars on the lower right hand corners mark 25 bm.
colon with europium labels. Sections were cut from a n aldehyde-fixed and wax-embedded surgical biopsy. The digital image recorded and shown in Figure 4A is RESULTS taken in the prompt mode and reveals a strong tissue Technical Performance autofluorescence. Figure 4B shows the same area in To verify the performance of the microscope with re- the time-resolved mode. The arrows S2 and S4 indicate gard to suppression of prompt fluorescence, salts of examples of areas where the autofluorescence was suprhodamine and europium-activated yttrium oxisulfide pressed. Sections from similar specimens are shown with percrystals were analyzed. Figure 3A-D shows the effect of suppression of the prompt fluorescence from the rho- oxidase labeling in Figure 5A, with alkaline phosdamine salt. On the basis of corresponding digital data phatase labeling in Figure 5B, with fluorescein labelthe suppression of the prompt fluorescence in the time- ing in Figure 5C, with europium labeling with prompt resolved mode was more than three orders of magni- mode in Figure 5D, and with europium labeling with time-resolved mode in Figure 5E. Arrows indicate the tude. areas where strong autofluorescence of the aldehydeBiological Applications fixed tissue can be easily observed. Localization of type I1 collagen mRNA in cartiTime-resolved fluorescence imaging of intraand extracellular antigens in tissue specimens laginous growth plate by using europium-labeled with europium-labeled and enzyme-labeled cDNA probe. In situ hybridization using either 35Sprobes. Figure 4A and B shows, localization of C242 labeling and autoradiography (Fig. 6A) or biotinylated in the epithelial cells of malignant mucosa of human probes and time-resolved fluorescence imaging of eu-
trast enhancement have been applied using a computer.
TIME-RESOLVED FLUORESCENCE IMAGING
FIG.5. Localization of C242 antigen in the epithelial cells of malignant mucosa of human colon. A: Peroxidase labeling. B: Alkaline phosphatase labeling. C: Fluorescein labeling. D Europium labeling with prompt mode. E. Europium labeling with time-resolved mode. The arrows indicate examples of areas where the autofluorescence
335
was suppressed (see arrows in Fig. 5D and E). Color bar shown in Figure 5E indicates intensity values in linear scale in artificial colours in Figure 5C-E (in arbitrary units). Bars on the lower right hand corners mark 25 pm.
336
SEVEUS ET AI,.
FIG 6. In situ hybridization of type I1 collagen mRNAs in cartilaginous growth plate by radioactively labeled probe detected by autoradiography and dark field microscopy (A! and by biotinylated probe detected by fluorescent europium chelate and time-resolved microscopy (B).The different zones of cartilage from left to right are hypertrophic chondrocytes, proliferative chondrocytes, and reserve chondrocytes.
ropium (Fig. 6B) for the detection of type I1 collagen mRNA in the developing human growth plate gave similar distributions of the positive cells, the signal accentuating in the lower proliferative and upper hypertrophic chondrocytes by both detection methods. Localization of viral genomes in tissue specimens by using europium-labeled gene sequences. The possibility to use europium chelates a s labels for in situ hybridization in conventionally processed tissue specimens was investigated in a set of experiments with biotinylated H P V l l DNA in surgical biopsies from human cervix known to be positive for H P V l l DNA. Figure 7 shows positive results with the in situ hybridization method. Figure 7A shows a section where the biotinylated hybrids were detected by the substrate reaction of alkaline phosphatase conjugated t o streptavidin. Figure 7B shows a subsequent section with europium-labeled streptavidin where the image was recorded in the time-resolved mode. Both methods gave similar frequency and distribution of the positive cells.
DISCUSSION Fluorescence methods are widely used in microscopy. The sensitivity of fluorescence microscopy is, however, serviously limited by the autofluorescence of the bio-
logical object. The present study shows that lanthanide chelates are suitable markers for time-resolved fluorescence microscopy and that time-resolved fluorescence imaging offers a unique method to suppress the background fluorescence. Lanthanide chelates have many advantages. Unlike the chelates used for certain in vitro assays, the europium chelate tested in this study is directly fluorescent. The fluorescence of this chelate is a result of excitation absorption of the organic part of the molecule (phenyl-ethyl-pyridine), followed by efficient energy transfer from the excited ligand triplet state to complexed europium, which produces the emission typical of metal ions. Pyridine nitrogen and four carboxylic acids in the chelate form the required thermodynamic stability and protect the emitting ion from the quenching effect of water molecules. Therefore, there is no need to dissociate the label from the antigen-antibody complex after immunoreaction. Hence, this chelate can be used as a label for localization of antigens, mRNAs, and gene sequences a t the cell and tissue levels. We have verified earlier that time resolution provides a very high separation efficiency between two signals with substantially different decay times (7). This experiment was carried out using a time-resolved fluorometer for test cuvettes which was optimized for
TIME-RESOLVED FLUORESCENCE IMAGING
337
FIG.7. Localization of H P V l l DNA sequences in formalin-fixed and wax-embedded cervical biopsy using europium-labeled DNA probe. A: Alkaline phosphatase labeling with absorption mode. B: Europium labeling with time-resolved mode. Bars on the lower right hand corners mark 25 pm.
europium using a delay time of 400 p s and a window time of 500 ps, and an emission filter optimized for the europium emission peak. The detection sensitivity of europium using these settings was better than 0.1 pmoliliter. Fluorescein and rhodamine samples (decay time below 10 ns) in increasing concentrations were measured with the same settings and cuvette. It was found that the signal from fluorescein and rhodamine with concentrations up to 1 mmoliliter and 0.01 mmoli liter, respectively, was zero. It was concluded that europium can be measured in the presence of a millionfold excess of a short decay time fluorescence substance which can be either a second label or background substance. The time-resolved mode allows multiparameter analysis in several ways. Not only could different lanthanide chelates (terbium or samarium) with different wavelengths be used but also markers with the same wavelength but substantially different decay times (e.g., conventional fluorochromes and fluorescent stains). The combination of an organic fluorescent stain and europium label would allow the simultaneous analysis of DNA content and an antibody, which
can be of interest in tumor pathology. Since the long decay fluorescence of europium can be detected in the presence of strong autofluorescence we could apply the lanthanide chelates both with aldehyde-fixed wax-embedded specimens and with frozen specimens. When using the conventional organic f luorochromes in aldehyde-fixed and wax-embedded tissue the sensitivity is often hampered by the background autofluorescence. In contrast to the crystalline labels, the size of the lanthanide chelates is small. They can therefore be used for the localization of both cell surface and intracellular antigens. Moreover, unlike the crystalline labels, the chelates are not sticky, which facilitates the labeling of sections or cells on the glass slide. The lanthanides are present in the biological specimen in negligible amounts. Therefore, the problems inherent to the endogenous presence of the label, common to enzyme labels, are avoided. The labeling procedure is a one or two step procedure comparable to other fluorescence methods. The cell and tissue structures are well preserved also in frozen sections, since in contrast to methods involving enzyme labels, no aggressive compounds are applied to lanthanide chelate
338
SEVEUS ET AL.
labeling. In addition, none of the reagents in the europium labeling procedure causes health or environmental hazards. The problems associated with the lanthanide chelate label are its instability under excitation and in connection with a different reagents for counterstaining (e.g., hematoxylinieosin) used in microscopy. The chemical characteristics of the chelates, however, allow changes in their chemical structure to be made to optimize them for different purposes. It is not unlikely that the structure of the present chelate can be changed to impove its stability. Furthermore, the europium chelate used in this study has a low overall fluorescence intensity. Our results show, however, that despite this, objects which are known to be present in a low concentration in the cell can be visualized in the time-resolved mode equally well as with other fluorescent labels. We have verified that the time-resolved fluorescence imaging is avery fast methodwhencomparedwith theuseofautoradiography. In addition autofluorescence and background fluorescence can be effectively eliminated, but the time-resolved fluorescence microscopy does not yet provide significantly better sensitivity than microscopy with conventional fluorescent labels. Further improvements in the chemistry of the chelates and image intensifying methods are needed and progress in those fields will make this method even more powerful in biology and pathology. Digital image processing is a potential possibility for automation of the microscopic studies in diagnostic pathology and various research fields. The successful use of digital image processing in routine studies requires, however, images with good contrast and easily distinguishable objects because it is difficult for image processing algorithms to descriminate between specific points and artefacts. Elimination of background and artefacts by time-resolved imaging makes the use of digital image processing more useful and should improve the reliability of the results.
ACKNOWLEDGMENTS The authors are grateful to Prof. Mikko Niemi for the possibility of using the facilities of the D e p a r t m e n t of Anatomy, University of Turku, for these studies.
LITERATURE CITED 1. Beverloo HB, van Schadewijk A, van Gelderen-Boele S, Tanke HJ: Inorganic phosphors as new luminescent labels for irnmunocytochemistry and time-resolved microscopy. Cytometry 11 :784792, 1990. 2. Elima K, Vuorio T, Vuorio E: Determination of the single polyadenylation site of the human type I1 collagen gene. Nucleic Acids Res 15:9499-9504, 1987. 3. Elster AD, Jackels SC, Allen NS, Marrache RC: EuropiumDTPA: A gadolinium analogue traceable by fluorescence microscopy. AJNR 10:1138-1144, 1989. 4. Feinberg A, Vogelstein B: A technique for radio-labeling DNA restriction endonuclease fragments to high specific activity. Anal Biochem 132:6-13, 1983. 5. Mukkala V-M, Mikola H, Hemmila I: The synthesis and use of activated N-benzyl derivatives of diethylenetriaminetetraacetic acids: Alternative reagents for labeling of antibodies with metal ions. Anal Biochem 176:319-325, 1989. 6. Sandberg M, Vuorio E: Localization of type I, I1 and I11 collagen mRNAs in developing human skeletal tissue by in situ hybridization. J Cell Biol 104:1077-1084, 1987. 7. Soini EJ, Hemrnila IA, Dahlen P: Time-resolved fluorescence for immunoassays and DNA-probes. In: Automation and New Technology in the Clinical Laboratory, Okuda K (ed). Blackwell Scientific Publications, Oxford, 1990, pp 63-67. 8. Soini E J , Lovgren T Time-resolved fluorescence of lanthanide probes and applications in biotechnology. CRC Crit Rev Anal Chem 18:105-154, 1987. 9. Soini EJ, Pelliniemi U,Hemmila IA, Mukkala V-M, Kankare JJ, Frojdman K: Lanthanide chelates as new fluorochrome labels for cytochemistry. J Histochem Cytochem 36:1449-1451, 1988. 10. Syrjiinen SM: Basic concepts and practical applications of recombinant DNA techniques in detection of human papillomavirus (HPV) infections. APMIS 98:95-110, 1990. 11. Syrjanen S, Partanen P, Mantyjarvi R, Syrjanen K: Sensitivity of in situ hybridization techniques using biotin- and 35S-labeied human papillomavirus (HPV) DNA probes. J Virol Methods 19:225238, 1988. 12. Takalo H: Synthesis of complexing compounds containing a substituted 4-ethylpyridine subunit. Thesis, University of Turku, Turku, Finland, 1988. 13. Tanke HJ: Does light microscopy have a future? J Microsc 155: 405-418, 1989.
