JOURNAL OF PATHOLOGY, VOL.
164: 5 1-58 (1 99 1)
INVESTIGATION OF A QUANTITATIVE POST-HYBRIDIZATION SIGNAL AMPLIFICATION SYSTEM FOR mRNA-OLIGODEOXY RIBONUCLEOTIDE IN SITU HYBRIDIZATION JUDITH A. HOYLAND AND ANTHONY J. FREEMONT
Departments of Internal Medicine and Pathological Sciences, University of Manchester Medical School, Stopford Building, Oxford Road, Manchester M 1 3 9 P T , U.K. Received 13 June 1990 Accepted 26 November I990
SUMMARY This study was undertaken to assess a post-hybridization amplification system for increasing the sensitivity of in-situ hybridization with oligodeoxyribonucleotide (oligonucleotide) probes. The technique employs a multilayer streptavidin-biotinylated link protein amplification technique, similar to that used in the ABV histochemical amplification system, to attach increasing numbers of radiolabelled streptavidin molecules to a biotinylated ologonucleotide probe. In tissue sections the amplification technique increases the signal-to-noise ratio dramatically, increasing the sensitivity of in situ hybridization and reducing the autoradiographic exposure time. On an artificial medium the amplification technique has been shown to increase the hybridization signal 100-fold when compared with a directly radiolabelled oligonucleotide probe. The technique could be used to compare relative amounts of target molecule, there being a linear relationship between the detectable signal and the log of the concentration of the target molecule. KEY
WORDS-In situ hybridization, oligonucleotide probes, amplification, quantification.
INTRODUCTION
ZH situ hybridization (ISH) for cytoplasmic and nuclear messenger R N A (mRNA) in tissue sections is a histopathological technique for studying cell function in normal and diseased tissues and complements existing tissue probing techniques. In theory, it should permit any cell to be examined for expression of the m R N A specific for exported o r retained proteins including intracellular enzymes. m R N A ISH is limited by the availability of probes against specific m R N A molecules. Until a wide range of conventional c D N A o r c R N A (riboprobe) probes becomes generally available, the histopathologist, without access to a dedicated Addressee for correspondence: A. J. Freemont, Department of Pathological Sciences, University of Manchester Medical School, Stopford Building, Oxford Road, Manchester MI 3 9PT, U.K.
0022-341 7/9 1 jOl005 1-08 $05.00 0 1991 by John Wiley & Sons, Ltd
molecular biology laboratory, cannot take full advantage of the potential of the technique to explore alterations in cell function in disease states. One solution that avoids the need for complex molecular biology is to use short synthetic probes to specific sequences of the m R N A under investigation (oligonucleotide probes). With access to suitable computational facilities it is relatively straightforward to design oligonucleotide probes between 25 and 30 bases long, complementary to segments of the m R N A being studied.' Careful cross-checking against all known m R N A sequences enables oligonucleotide probes t o be designed that have minimal homology with any base sequence other than that of the specific m R N A . Control of the conditions of in situ hybridization further ensures specificity of the hybridization reaction. F o r the histopathologist these short probes have the advantage'of good penetration into sections of
52
J. A . HOYLAND AND A. J. FREEMONT
tissue fixed using cross-linking fixatives. The major problem encountered with oligonucleotide probes is in their detection. Large probes labelled with a series of radioisotope or chemically tagged bases can be disclosed using autoradiography or a suitable chemical system.’ Although oligonucleotide probes can be labelled with similar molecules, the nature of the labelling technique, in practical terms, means that significantly less labelled bases are added than wiih larger probes, thus reducing the relative sensitivity of oligonucleotide ISH. As in immunohistochemistry, methods are available for amplifying the signal from non-radiolabelled probes during the disclosing reaction, producing a coloured final reaction p r ~ d u c tUsing . ~ this type of system, quantification of the reaction product is difficult and, as a result, comparison of the amounts of cellular mRNA becomes unreliable. It would be useful to know if the amount of cellular mRNA is altered in disease states as this may indicate changes in cell activity that have a significant bearing on understanding disease mechanisms. If quantitation of in sifu product were possible, it would therefore be a bonus. Radiolabelled probes, detected by autoradiography, lend themselves to quantification by some form of silver grain counting technique, but with oligonucleotide probes the signal is often too low to make this practicable. The signal from the radiolabelled base can be amplified by the use of high energy labels, e.g., 32P, but with a resultant decrease in signal localization, or by increasing the exposure time to several weeks, which is often an unexceptably long period between initiating an experiment and data analysis. Ideally what is required is a technique that is sensitive, that allows quantification, and that can be completed in a reasonable length of time. In the study described here we explore a quantitative method for increasing the sensitivity of oligonucleotide in situ hybridization by posthybridization signal amplification.
MATERIALS AND METHODS In general terms, the practical aspects of in .ritu hybridization involve a series of steps which can be summarized as: specimen preparation: (2) probe formulation; (3) probe labelling; (4) hybridization: (1)
(5) disclosure of bound probe; and (6) quantification of probe signal. I . Specimen prepurution Two types of specimen were employed in this study: (a) randomized, serial sections from conventionally fixed and processed tissue from active Dupuytren’s nodules. Dupuytren’s tissue was selected because the cells are actively biosynthetic and there is a reasonable degree of cell separation which facilitates autoradiographic grain localization (see Fig. 1). The tissue was fixed for 24 h in neutral buffered formalin, processed overnight into paraffin wax, and 5 p m serial tissue sections were mounted onto aminopropyltriethoxysilane-coated glass slide^.^ Sections were randomized and arranged into groups of five. All studies were performed on sets of five sections and the data generated in each part of each study are a mean of the five. (b) An artificial medium onto which had been baked known concentrations of a synthetic homologue of poly A + RNA (i.e., the sequence common to all molecules of mRNA) between 25 and 30 bases long. 2. Probe formulation The oligonucleotide probes were purchased from Pharmacia Ltd (Pharmacia LKB Biotechnology Division, Milton Keynes, Bucks.). One consisted of between 25 and 30 thymidine bases, and was designed to react with the repetitive sequence of adenine bases that form the ‘poly A tail’ present on all mRNA molecules. The second probe consisted of 25-30 adenine bases, a complementary base sequence to the first probe. It was used both labelled as a hybridization control (Section 7.2) and unlabelled as an artificial reaction medium for assessing the sensitivity of post-hybridization amplification systems. 3. Labelling of probes
Probes were labelled at the 3‘ end using terminal deoxynucleotidyl transferase (TdT) by a standard method;’ the poly T probe with deoxythmidine 5’- [a trip ho s p ha t e ([ S]- dTTP) (Amersham International, Aylesbury, Bucks.; SJ307) or with biotinylated deoxyuridine triphosphate (b-dUTP) (Boehringer Mannheim U.K., Lewes, East Sussex; 1093-070); and the poly-A probe with biotinylated
’’
AMPLIFICATION OF IN m
u HYBRIDIZATION
53
Fig. I-Comparison of the directly radiolabelled probe (A) and the two-layer amplification system (B). Autoradiograph counterstained with H&E
deoxyadenine triphosphate (b-dATP) (Boehringer Mannheim 1093468) or with deoxyadenine 5'-[a''SJtriphosphate (t3'S]-dATP) (Amersham SJ304). Radiolabelling with [35S]-dTTP gave a specific activity of 4.8 x lo6 dpmlpg. 4. Hybridization Tissue sections were dewaxed and rehydrated through xylene, alcohol, and water, with a final wash in diethyl pyrocarbonate (DEPC)-treated twice-distilled water. Hybridization was similar to that described by Pringle et al.' Slides were covered with hybridization buffer at 37°C containing oligonucleotide probe at a concentration of 160ng/ml and 1 0 m ~ dithiothreitol at 37°C for 18h, then washed with SSC at 37"C, dehydrated, and air-dried. 5. Disclosure of boundprobe 5.1. General description of disclosure systems-
Throughout this study the presence of probe was detected autoradiographically. The probes were either radiolabelled directly, requiring only that autoradiography be applied to the section, or labelled with a biotinylated base which was then detected using radiotagged chemicals. These methods are detailed in Section 7.2.
5.2. Autoradiography-Slides were dipped in a photographic emulsion consisting of 50 per cent Ilford K5 emulsion in water, then placed in lighttight boxes containing desiccant for 2-8 days, removed, developed, stained with H&E, dehydrated, and mounted. 6. Quantijication ofprobe signal
The signal was measured by assessing the density of silver grains over individual cells using a Sight Systems Image Manager image analysis system (Sight Systems, Newbury, Berkshire). Because the emission from "S (the radiolabel used throughout the study) has a maximum penetration of 10 pm in emulsion, the area studied included the cell plus a zone 10pm wide outside the periphery of the cell. Sections were examined by epi-illumination using a x 50 oil immersion objective. The study zone of 100 cells was delineated in transmitted light and stored in the computer. The lighting was then changed to epi-illumination and the computer was used to measure the grain density over each cell by establishing the fractional area of the study zone occupied by silver grains. The silver grain count over 100 cells was measured for each slide. Zeroing measurements and measurements of background
54
J. A. HOYLAND AND A. J. FREEMONT
‘noise’, defined as the grain density over acellular areas, were also performed. The results are expressed as ratios of signal to noise. 7. Development and assessment of the signal amplijication system
7.1. General principles-Signal amplification systems (such as immunoalkaline phosphatase and avidin-biotin complex techniques) are widely used in immunohistochemistry to improve epitope detection. They all exploit the principle that increased sensitivity can be obtained by linking several histochemically active molecules to one bound antibody molecule. The disclosing system is usually an enzyme which gives a coloured reaction product when the section is incubated in a suitable medium. The disclosing steps employed following in situ hybridization with non-radioactively labelled probes are similar to those used for recognizing bound antibody. In the experiments described here, we exploited the same amplification systems used to improve the visualization of antibody binding, to attach radioisotopes to the biotin tagged base of single hybridized oligonucleotide molecules in tissue sections and we assessed whether or not the results are reproducible. In addition, we used a similar system to relate the amount of autoradiographic reaction product to the quantity of a homologue of mRNA in an artificial medium treated as if it were a tissue section. The disclosing system used is based on a multistep streptavidin-biotin binding technique. Simply, the initial step in the disclosing system is the binding of a molecule of radiolabelled streptavidin to the biotin molecule on the hybridized probe. To this can be bound a link molecule (because of availability we used alkaline phosphatase) containing more than one biotin molecule. Further radiolabelled streptavidin molecules can be layered on to bind with the free biotinylated regions on the link molecule. The link protein-streptavidin steps can then be repeated. In this way, a large network of radiolabelled streptavidin molecules can be bound to the single biotin molecule on the probe.
7.2. Assessment of the ampllfication system on tissue sections-In this study a comparison was made between the density of autoradiographic final reaction product using sections in which mRNA is:
(a) hybridized with j5S-labelled poly T probe; (b) hybridized with biotinylated poly T probe reacted with (i) one streptavidin molecule (one-layer technique), (ii) two layers of streptavidin-biotinylated alkaline phosphatase complexes (SBAPC) (two-layer technique), or (iii) three layers of SBAPC (three-layer technique). The reaction product was also examined at different exposure times of 2, 5 , and 8 days. At each time interval, 15 sections (three sets of five sections (see Section la) were treated with each of the four disclosing systems (a, bi, bii, and biii above). For the directly labelled probe the hybridization method detailed above was followed directly by autoradiography. For the biotinylated probe, following hybridization each of the 45 sections was incubated with a 0.5 p~ solution of ”S-labelled streptavidin (Amersham SJ436) for 30 min at room temperature. Sections were then washed in 0.125 M Tris-buffered saline (TBS) for 5 rnin. The labelling was either terminated at this point and the autoradiographic steps were undertaken (1 5 sections), or the sections were treated with biotinylated link protein. The link protein used was biotinylated alkaline phosphatase (from the Dakopatts ‘strep AB complex AP’ kit K391; Dako Ltd., High Wycombe, Bucks.) at the product concentration (1 in 1000 dilution of the contents of the biotinylated alkaline phosphatase vial) recommended by Dako, for 30 min at room temperature. The sections were washed in 0.125 M TBS for 5 min and the streptavidin reaction was repeated. Fifteen sections were processed for autoradiography and the last 15 treated with the biotinylated link protein-streptavidin procedure once more before autoradiography. 7.3. Optimal streptavidin concentration-The major disadvantage of using the techniques described in Section 7.2 is that radiolabelled streptavidin is expensive. An experiment was therefore designed to assess whether smaller quantities of streptavidin could be used. Five-pm-thick tissue sections of Dupuytren’s tissue were hybridized as described above. A singlelayer disclosing procedure was used, autoradiography was undertaken, and the reaction product quantified.
55
AMPLIFICATION OF IN SITU HYBRIDIZATION
The technique was repeated with ten-fold dilutions of streptavidin until a fall of more than 10 per cent in the autoradiographic grain density was noted (1 0 per cent was chosen as being a limit outside which the fall was unlikely to be due to chance). The concentration was then increased by doubling until the grain count returned to the original amount. This concentration was then determined to be the optimal streptavidin concentration. The procedure was repeated for the two- and three-layer disclosing techniques. 7.4. Relationship between reaction product and amount of message-In this experiment the effect of varying the numbers of bound streptavidin molecules on the sensitivity of the detection system was tested. A diluent (9 parts 20 x SSC and 1 part 10 mg/ml of salmon sperm DNA) was used to produce varying concentrations of the poly d(A) probe in the range of 500 ng/,d to 100 pglpl. The solutions were boiled for 10 min and then chilled rapidly on ice. One p1 aliquots of each dilution were applied to nine nitrocellulose strips 8 cm x 1 cm where each formed a round plaque 2.0mm in diameter. The strips were then placed between paper and baked at 80°C for 2 h, after which they were immersed in 4 x SSC, removed, and placed in a plastic bag with 4ml of pre-hybridization fluid for 1 h. After this time, the pre-hybridization fluid was removed and replaced with hybridization fluid containing biotinylated poly d(T) probe at a concentration of 1000 ng/ml. Hybridization was allowed to proceed for 18 h at 37°C. The strips were then treated as if they were slides with the exception that all the reactions were carried out within sealed plastic bags. Disclosure of bound probe was undertaken using the single-layer technique on three strips, the two-layer technique on a further three strips and the three-layer technique on the remaining three. An identical experiment was performed using the radiolabelled poly d(T) probe for comparison. At the end of the hybridization all 12 nitrocellulose strips were wrapped in cling film and exposed to X-ray film for 4 days at - 70°C. The reaction product was assessed microdensitometrically by grey level analysis of autoradiographic grain density over each plaque, and the minimum concentration at which product was detected was noted.
RESULTS Assessment of the amplijkation on tissue sections (7.2.) The results of this experiment are summarized in Table I. All controls, with the exception of those treated with DNase, were negative. The results for the DNase-treated tissue were similar to those of the test systems. Table I-A comparison of the signal, noise, and signal-tonoise ratio for each of the four labelling techniques at different exposure times
2
No. of days exposure 5 8
Direct label
Signal Noise SIN
3.3 f 1.4 2.1 f0.6 1.6f0.2
5.1 f 1.9 3.0 f0.7 1.7 f 0 . 1
16.7 f5.9 5.4 f2.1 3.1 f 0 . 3
One layer
Signal Noise SIN
12.5f3.6 6.1 f 1.0 2.1 f0.4
36.4f8.1 8.4k4.2 4.3 f0.4
74.3 f9.6 10.6 f2.1 7.0 f0.5
Signal Noise SIN
33.5 f8.4 8.0 f3.1 4.2 f0.6
67.9 f6.8 10.4k2.2 6.5 f2.4
95.4 f7.3 9.4 f5.3 10.1 f 3 . 4
Signal Noise SIN
50.2 f 17.3 96.5 f8.6 8.4 f3.4 10.7 & 3.0 9.0f 1.1 5.3 f2.7
98.9 f 5.3 8.8 f2.6 11.2f2.3
Two layer
Three layer
As a general rule, the detected signal increased with the exposure time and the number of streptavidin molecules used in the disclosing system. Even after 2 days exposure all the disclosing systems registered grain counts over cells in excess of background, (i.e. values for S/N 3 l), indicating hybridization to have occurred. The background grain counts were similar for all the streptavidin-biotin techniques and did not vary significantly with the number of SBAPC layers or the exposure time. The background was, however, between two and three times higher than that obtained using the directly labelled probe, presumably because of non-specific binding of streptavidin to the tissue. The increase in signal was much greater than the increase in noise, making the signal-tonoise ratio much higher. Unfortunately, with the two- and three-layer techniques the silver grain densities were very
56
J. A. HOYLAND AND A. J. FREEMONT
100
-
50
-
043
t3
o
n
CONC OI:
ARTIFICIAL “mRNA“
-
ndpl
10
W
5-
2.5-
DIRECT
ONE
LABEL
LAYER
Two L A Y FR
.,.
1IIREE LAYER
DISCLOSING TECHNIOUP. Fig. 2-Graph comparing the lowest concentration of artificial mRNA detected in each of the autoradiographic strips examined with the four disclosing techniques: (0)direct labelling; (0) one layer; ( 0 )two layer; and (H) three layer
high at long exposures and the cellular detail was obscured by overlapping grains. This made grain counting inaccurate and probably explains the smaller difference between the 5- and 8-day signalto-noise ratios and the corresponding 2- and 5-day ratios. The best definition was obtained with the three-layer technique at 2 days and with the two-layer technique at 5 days. Optimal streptavidin concentration ( 7 . 3 . ) Using the technique described, the optimal concentrations for the one-, two-, and three-layer amplification methods were found to be 1 x lo-*, and 0.1 p ~ respectively. , These represent 4x 1/50, 1/12, and l / 5 of the amount of streptavidin originally used.
Relationship between reaction product and amount of message [ 7 . 4 . ) The results of this experiment are shown in Figs 2 and 3 . Figure 2 shows the differences in minimum message concentration that can be detected using the four disclosing systems under the same conditions. When compared with the directly labelled probe, the two-layer technique offers a ten-fold increase in sensitivity and the three-layer technique a 100-fold increase. Figure 3 relates known concentrations of the artificial ‘mRNA’ disclosed following in situ hybridization using the two-layer technique to the silver grain density on X-ray film. There is a direct association between the measured reaction product density and the log of the ‘mRNA’ concentration.
57
AMPLIFICATION OF IN SITU HYBRIDIZATION
160
W
140
120
GREY LEVEL (MACIIINE UNITS)
'O0
80
60
0 0
40
0
20
- I'
I'
2.3
I'O
15 i o
l'oo
CONC ( n g / p l ) OF AI!TIFICIAI, "mRNA"
Fig. 3-Graph comparing the amount of reaction product (assessed by measuring the mean grey level of each plaque autoradiograph) with the known concentration of poly d(A) molecules in each plaque using the two-layer disclosing technique
DISCUSSION Oligodeoxynucleotide probes represent a new and potentially powerful histological tool for the study of cell function in disease state^.^ A major criticism of the use of these small probes is that the amount of label that can, in practice, be added to the probe is much smaller than that for an equivalent cDNA or riboprobe (cRNA). (For instance, a 1 kb cDNA probe with equal amounts of each base will acquire approximately 250 labelled bases during random primer labelling, whereas a 30 base oligonucleotide probe labelled can, in practice, be given no more than 20 labelled bases.) Although the sensitivity of the oligonucleotide probe will be no less than that of the cDNA probe (indeed, in
formalin-fixed tissue it may be higher because of the greater penetration of the smaller probe), the sensitivity of the detection system will be between one and two orders of magnitude smaller. One way of increasing the detectability of bound oligonucleotide probe is to amplify the signal after hybridization. Post-hybridization amplification of biotinylated oligonucleotides using the streptavidin biotin enzyme complex has already been reported."6 This type of amplification system usually gives a coloured reaction product, which although demonstrating very adequately the presence of a hybridization reaction, is not easily or reliably quantified. We have adapted this technique in two ways. First, by constructing the complex in stages on the slide, thus allowing the size of the complex
58
J. A. HOYLAND AND A. J. FREEMONT
(and thereby the number of streptavidin molecules) due to the autoradiographic exposure time has to be controlled. And second, by replacing strepta- always been a disadvantage in an autoradiographic vidin with "S-radiolabelled streptavidin, permiting disclosure system. The amplification technique could also be used to autoradiographic disclosure of hybridized probe aid in the detection of low numbers of mRNA and quantification. In practice, the amplification technique is copies by increasing either the exposure time or the straightforward (for a laboratory routinely using number of layers of streptavidin molecules further. In summary, a technique has been developed and immunohistochemistry) and gives reproducible investigated that allows the signal from oligoresults. The reaction product can be quantified by assess- nucleotide probes to be amplified and quantified. It ing the autoradiographic silver grain density in is simple, reproducible, and easily adapted to the emulsion or on overlaid X-ray film. On the nitro- nature of the mRNA being investigated. cellulose gel, increasing the number of streptavidin molecules resulted in a predictable increase in the number of autoradiographic grains; and for any ACKNOWLEDGEMENTS particular number of streptavidin molecules, the silver grain density increased directly with the This work was funded by the Arthritis and number of hybridization target molecules. Whether this happens in tissue is more difficult to assess Rheumatism Council for Research, and the North because the amount of mRNA even in sequential Western Regional Research Grants Committee. sections may vary and there is an unpredictable quenching effect of a proportion of radioactive emissions by tissue. REFERENCES Our data indicate that when compared with the directly labelled probe the radiostreptavidin-biotin 1. Caruthers MH. Synthesis of oligonucleotides and oligonucleotide amplification system can give up to a 100-fold analogues. In: Cohen JS. ed. Oligodeoxynucleotides: Antisense Inhibitors of Gene Expression. Basingstoke: Macmillan, 1989: 7-25. increase in signal per labelled base, without an M. Application of avidin-biotin complex technique for the appreciable increase in noise. By amplification the 2. Sar localisation of estradiol receptor in target tissues using monoclonal signal from an oligonucleotide tagged with as few antibodies In: Bullock G R , Petrusr P, eds. Techniques in ImmunoVol 3. London: Academic Preaa. 1988: 43-54. as 2 bases can become, in effect, comparable to 3. cytochemistry, Pringle J H , Primrose L, Kind CN. Talbot IC, Lauder 1 In-situ that from a 1 kb cDNA probe containing 200 hybridization demonstration of poly-adenylated R N A sequences in formalin-fixed paraffin sections using a biotinylated oligonucleotide radiolabelled bases. poly d(T) probe. J Pathol 1989; 158: 279-286. The efficiency of the detection system was such 4. Van Prooijen-Knegt AC, Raap AK, Van der Burg MJM. Spreading that in this study the grain density achieved under and staining of human metaphase chromosomes on aminoalkylsilane treated glass slides. Hi.ytockem J 1983; 1 4 333-334. some of the amplification conditions was so great Deng G, Wu R.Terminal transferase: use in the tailing of D N A and for that cellular detail was obscured. This aspect of the 5. in-viiro niutagenesis. Metiioih Enzjnzol1983; 9 6 96.- I 16. amplification system could be exploited to shorten 6. Mougin C . Guitteny AF. Fouque B. Viennet G, Teoule R, Bloch B. Histochemical detection of the messenger RNAs coding for calcitonin the exposure time if the technique is to be used with a and calcitonin gene-related peptide in medullary thyroid carcinomas highly transcribed message. This may be a signifiwith radioactive and biotinylated oligonucleotide probes. J Polhol 1990; 160: 187-194. cant advantage as the delay in obtaining results
164: 5 1-58 (1 99 1)
INVESTIGATION OF A QUANTITATIVE POST-HYBRIDIZATION SIGNAL AMPLIFICATION SYSTEM FOR mRNA-OLIGODEOXY RIBONUCLEOTIDE IN SITU HYBRIDIZATION JUDITH A. HOYLAND AND ANTHONY J. FREEMONT
Departments of Internal Medicine and Pathological Sciences, University of Manchester Medical School, Stopford Building, Oxford Road, Manchester M 1 3 9 P T , U.K. Received 13 June 1990 Accepted 26 November I990
SUMMARY This study was undertaken to assess a post-hybridization amplification system for increasing the sensitivity of in-situ hybridization with oligodeoxyribonucleotide (oligonucleotide) probes. The technique employs a multilayer streptavidin-biotinylated link protein amplification technique, similar to that used in the ABV histochemical amplification system, to attach increasing numbers of radiolabelled streptavidin molecules to a biotinylated ologonucleotide probe. In tissue sections the amplification technique increases the signal-to-noise ratio dramatically, increasing the sensitivity of in situ hybridization and reducing the autoradiographic exposure time. On an artificial medium the amplification technique has been shown to increase the hybridization signal 100-fold when compared with a directly radiolabelled oligonucleotide probe. The technique could be used to compare relative amounts of target molecule, there being a linear relationship between the detectable signal and the log of the concentration of the target molecule. KEY
WORDS-In situ hybridization, oligonucleotide probes, amplification, quantification.
INTRODUCTION
ZH situ hybridization (ISH) for cytoplasmic and nuclear messenger R N A (mRNA) in tissue sections is a histopathological technique for studying cell function in normal and diseased tissues and complements existing tissue probing techniques. In theory, it should permit any cell to be examined for expression of the m R N A specific for exported o r retained proteins including intracellular enzymes. m R N A ISH is limited by the availability of probes against specific m R N A molecules. Until a wide range of conventional c D N A o r c R N A (riboprobe) probes becomes generally available, the histopathologist, without access to a dedicated Addressee for correspondence: A. J. Freemont, Department of Pathological Sciences, University of Manchester Medical School, Stopford Building, Oxford Road, Manchester MI 3 9PT, U.K.
0022-341 7/9 1 jOl005 1-08 $05.00 0 1991 by John Wiley & Sons, Ltd
molecular biology laboratory, cannot take full advantage of the potential of the technique to explore alterations in cell function in disease states. One solution that avoids the need for complex molecular biology is to use short synthetic probes to specific sequences of the m R N A under investigation (oligonucleotide probes). With access to suitable computational facilities it is relatively straightforward to design oligonucleotide probes between 25 and 30 bases long, complementary to segments of the m R N A being studied.' Careful cross-checking against all known m R N A sequences enables oligonucleotide probes t o be designed that have minimal homology with any base sequence other than that of the specific m R N A . Control of the conditions of in situ hybridization further ensures specificity of the hybridization reaction. F o r the histopathologist these short probes have the advantage'of good penetration into sections of
52
J. A . HOYLAND AND A. J. FREEMONT
tissue fixed using cross-linking fixatives. The major problem encountered with oligonucleotide probes is in their detection. Large probes labelled with a series of radioisotope or chemically tagged bases can be disclosed using autoradiography or a suitable chemical system.’ Although oligonucleotide probes can be labelled with similar molecules, the nature of the labelling technique, in practical terms, means that significantly less labelled bases are added than wiih larger probes, thus reducing the relative sensitivity of oligonucleotide ISH. As in immunohistochemistry, methods are available for amplifying the signal from non-radiolabelled probes during the disclosing reaction, producing a coloured final reaction p r ~ d u c tUsing . ~ this type of system, quantification of the reaction product is difficult and, as a result, comparison of the amounts of cellular mRNA becomes unreliable. It would be useful to know if the amount of cellular mRNA is altered in disease states as this may indicate changes in cell activity that have a significant bearing on understanding disease mechanisms. If quantitation of in sifu product were possible, it would therefore be a bonus. Radiolabelled probes, detected by autoradiography, lend themselves to quantification by some form of silver grain counting technique, but with oligonucleotide probes the signal is often too low to make this practicable. The signal from the radiolabelled base can be amplified by the use of high energy labels, e.g., 32P, but with a resultant decrease in signal localization, or by increasing the exposure time to several weeks, which is often an unexceptably long period between initiating an experiment and data analysis. Ideally what is required is a technique that is sensitive, that allows quantification, and that can be completed in a reasonable length of time. In the study described here we explore a quantitative method for increasing the sensitivity of oligonucleotide in situ hybridization by posthybridization signal amplification.
MATERIALS AND METHODS In general terms, the practical aspects of in .ritu hybridization involve a series of steps which can be summarized as: specimen preparation: (2) probe formulation; (3) probe labelling; (4) hybridization: (1)
(5) disclosure of bound probe; and (6) quantification of probe signal. I . Specimen prepurution Two types of specimen were employed in this study: (a) randomized, serial sections from conventionally fixed and processed tissue from active Dupuytren’s nodules. Dupuytren’s tissue was selected because the cells are actively biosynthetic and there is a reasonable degree of cell separation which facilitates autoradiographic grain localization (see Fig. 1). The tissue was fixed for 24 h in neutral buffered formalin, processed overnight into paraffin wax, and 5 p m serial tissue sections were mounted onto aminopropyltriethoxysilane-coated glass slide^.^ Sections were randomized and arranged into groups of five. All studies were performed on sets of five sections and the data generated in each part of each study are a mean of the five. (b) An artificial medium onto which had been baked known concentrations of a synthetic homologue of poly A + RNA (i.e., the sequence common to all molecules of mRNA) between 25 and 30 bases long. 2. Probe formulation The oligonucleotide probes were purchased from Pharmacia Ltd (Pharmacia LKB Biotechnology Division, Milton Keynes, Bucks.). One consisted of between 25 and 30 thymidine bases, and was designed to react with the repetitive sequence of adenine bases that form the ‘poly A tail’ present on all mRNA molecules. The second probe consisted of 25-30 adenine bases, a complementary base sequence to the first probe. It was used both labelled as a hybridization control (Section 7.2) and unlabelled as an artificial reaction medium for assessing the sensitivity of post-hybridization amplification systems. 3. Labelling of probes
Probes were labelled at the 3‘ end using terminal deoxynucleotidyl transferase (TdT) by a standard method;’ the poly T probe with deoxythmidine 5’- [a trip ho s p ha t e ([ S]- dTTP) (Amersham International, Aylesbury, Bucks.; SJ307) or with biotinylated deoxyuridine triphosphate (b-dUTP) (Boehringer Mannheim U.K., Lewes, East Sussex; 1093-070); and the poly-A probe with biotinylated
’’
AMPLIFICATION OF IN m
u HYBRIDIZATION
53
Fig. I-Comparison of the directly radiolabelled probe (A) and the two-layer amplification system (B). Autoradiograph counterstained with H&E
deoxyadenine triphosphate (b-dATP) (Boehringer Mannheim 1093468) or with deoxyadenine 5'-[a''SJtriphosphate (t3'S]-dATP) (Amersham SJ304). Radiolabelling with [35S]-dTTP gave a specific activity of 4.8 x lo6 dpmlpg. 4. Hybridization Tissue sections were dewaxed and rehydrated through xylene, alcohol, and water, with a final wash in diethyl pyrocarbonate (DEPC)-treated twice-distilled water. Hybridization was similar to that described by Pringle et al.' Slides were covered with hybridization buffer at 37°C containing oligonucleotide probe at a concentration of 160ng/ml and 1 0 m ~ dithiothreitol at 37°C for 18h, then washed with SSC at 37"C, dehydrated, and air-dried. 5. Disclosure of boundprobe 5.1. General description of disclosure systems-
Throughout this study the presence of probe was detected autoradiographically. The probes were either radiolabelled directly, requiring only that autoradiography be applied to the section, or labelled with a biotinylated base which was then detected using radiotagged chemicals. These methods are detailed in Section 7.2.
5.2. Autoradiography-Slides were dipped in a photographic emulsion consisting of 50 per cent Ilford K5 emulsion in water, then placed in lighttight boxes containing desiccant for 2-8 days, removed, developed, stained with H&E, dehydrated, and mounted. 6. Quantijication ofprobe signal
The signal was measured by assessing the density of silver grains over individual cells using a Sight Systems Image Manager image analysis system (Sight Systems, Newbury, Berkshire). Because the emission from "S (the radiolabel used throughout the study) has a maximum penetration of 10 pm in emulsion, the area studied included the cell plus a zone 10pm wide outside the periphery of the cell. Sections were examined by epi-illumination using a x 50 oil immersion objective. The study zone of 100 cells was delineated in transmitted light and stored in the computer. The lighting was then changed to epi-illumination and the computer was used to measure the grain density over each cell by establishing the fractional area of the study zone occupied by silver grains. The silver grain count over 100 cells was measured for each slide. Zeroing measurements and measurements of background
54
J. A. HOYLAND AND A. J. FREEMONT
‘noise’, defined as the grain density over acellular areas, were also performed. The results are expressed as ratios of signal to noise. 7. Development and assessment of the signal amplijication system
7.1. General principles-Signal amplification systems (such as immunoalkaline phosphatase and avidin-biotin complex techniques) are widely used in immunohistochemistry to improve epitope detection. They all exploit the principle that increased sensitivity can be obtained by linking several histochemically active molecules to one bound antibody molecule. The disclosing system is usually an enzyme which gives a coloured reaction product when the section is incubated in a suitable medium. The disclosing steps employed following in situ hybridization with non-radioactively labelled probes are similar to those used for recognizing bound antibody. In the experiments described here, we exploited the same amplification systems used to improve the visualization of antibody binding, to attach radioisotopes to the biotin tagged base of single hybridized oligonucleotide molecules in tissue sections and we assessed whether or not the results are reproducible. In addition, we used a similar system to relate the amount of autoradiographic reaction product to the quantity of a homologue of mRNA in an artificial medium treated as if it were a tissue section. The disclosing system used is based on a multistep streptavidin-biotin binding technique. Simply, the initial step in the disclosing system is the binding of a molecule of radiolabelled streptavidin to the biotin molecule on the hybridized probe. To this can be bound a link molecule (because of availability we used alkaline phosphatase) containing more than one biotin molecule. Further radiolabelled streptavidin molecules can be layered on to bind with the free biotinylated regions on the link molecule. The link protein-streptavidin steps can then be repeated. In this way, a large network of radiolabelled streptavidin molecules can be bound to the single biotin molecule on the probe.
7.2. Assessment of the ampllfication system on tissue sections-In this study a comparison was made between the density of autoradiographic final reaction product using sections in which mRNA is:
(a) hybridized with j5S-labelled poly T probe; (b) hybridized with biotinylated poly T probe reacted with (i) one streptavidin molecule (one-layer technique), (ii) two layers of streptavidin-biotinylated alkaline phosphatase complexes (SBAPC) (two-layer technique), or (iii) three layers of SBAPC (three-layer technique). The reaction product was also examined at different exposure times of 2, 5 , and 8 days. At each time interval, 15 sections (three sets of five sections (see Section la) were treated with each of the four disclosing systems (a, bi, bii, and biii above). For the directly labelled probe the hybridization method detailed above was followed directly by autoradiography. For the biotinylated probe, following hybridization each of the 45 sections was incubated with a 0.5 p~ solution of ”S-labelled streptavidin (Amersham SJ436) for 30 min at room temperature. Sections were then washed in 0.125 M Tris-buffered saline (TBS) for 5 rnin. The labelling was either terminated at this point and the autoradiographic steps were undertaken (1 5 sections), or the sections were treated with biotinylated link protein. The link protein used was biotinylated alkaline phosphatase (from the Dakopatts ‘strep AB complex AP’ kit K391; Dako Ltd., High Wycombe, Bucks.) at the product concentration (1 in 1000 dilution of the contents of the biotinylated alkaline phosphatase vial) recommended by Dako, for 30 min at room temperature. The sections were washed in 0.125 M TBS for 5 min and the streptavidin reaction was repeated. Fifteen sections were processed for autoradiography and the last 15 treated with the biotinylated link protein-streptavidin procedure once more before autoradiography. 7.3. Optimal streptavidin concentration-The major disadvantage of using the techniques described in Section 7.2 is that radiolabelled streptavidin is expensive. An experiment was therefore designed to assess whether smaller quantities of streptavidin could be used. Five-pm-thick tissue sections of Dupuytren’s tissue were hybridized as described above. A singlelayer disclosing procedure was used, autoradiography was undertaken, and the reaction product quantified.
55
AMPLIFICATION OF IN SITU HYBRIDIZATION
The technique was repeated with ten-fold dilutions of streptavidin until a fall of more than 10 per cent in the autoradiographic grain density was noted (1 0 per cent was chosen as being a limit outside which the fall was unlikely to be due to chance). The concentration was then increased by doubling until the grain count returned to the original amount. This concentration was then determined to be the optimal streptavidin concentration. The procedure was repeated for the two- and three-layer disclosing techniques. 7.4. Relationship between reaction product and amount of message-In this experiment the effect of varying the numbers of bound streptavidin molecules on the sensitivity of the detection system was tested. A diluent (9 parts 20 x SSC and 1 part 10 mg/ml of salmon sperm DNA) was used to produce varying concentrations of the poly d(A) probe in the range of 500 ng/,d to 100 pglpl. The solutions were boiled for 10 min and then chilled rapidly on ice. One p1 aliquots of each dilution were applied to nine nitrocellulose strips 8 cm x 1 cm where each formed a round plaque 2.0mm in diameter. The strips were then placed between paper and baked at 80°C for 2 h, after which they were immersed in 4 x SSC, removed, and placed in a plastic bag with 4ml of pre-hybridization fluid for 1 h. After this time, the pre-hybridization fluid was removed and replaced with hybridization fluid containing biotinylated poly d(T) probe at a concentration of 1000 ng/ml. Hybridization was allowed to proceed for 18 h at 37°C. The strips were then treated as if they were slides with the exception that all the reactions were carried out within sealed plastic bags. Disclosure of bound probe was undertaken using the single-layer technique on three strips, the two-layer technique on a further three strips and the three-layer technique on the remaining three. An identical experiment was performed using the radiolabelled poly d(T) probe for comparison. At the end of the hybridization all 12 nitrocellulose strips were wrapped in cling film and exposed to X-ray film for 4 days at - 70°C. The reaction product was assessed microdensitometrically by grey level analysis of autoradiographic grain density over each plaque, and the minimum concentration at which product was detected was noted.
RESULTS Assessment of the amplijkation on tissue sections (7.2.) The results of this experiment are summarized in Table I. All controls, with the exception of those treated with DNase, were negative. The results for the DNase-treated tissue were similar to those of the test systems. Table I-A comparison of the signal, noise, and signal-tonoise ratio for each of the four labelling techniques at different exposure times
2
No. of days exposure 5 8
Direct label
Signal Noise SIN
3.3 f 1.4 2.1 f0.6 1.6f0.2
5.1 f 1.9 3.0 f0.7 1.7 f 0 . 1
16.7 f5.9 5.4 f2.1 3.1 f 0 . 3
One layer
Signal Noise SIN
12.5f3.6 6.1 f 1.0 2.1 f0.4
36.4f8.1 8.4k4.2 4.3 f0.4
74.3 f9.6 10.6 f2.1 7.0 f0.5
Signal Noise SIN
33.5 f8.4 8.0 f3.1 4.2 f0.6
67.9 f6.8 10.4k2.2 6.5 f2.4
95.4 f7.3 9.4 f5.3 10.1 f 3 . 4
Signal Noise SIN
50.2 f 17.3 96.5 f8.6 8.4 f3.4 10.7 & 3.0 9.0f 1.1 5.3 f2.7
98.9 f 5.3 8.8 f2.6 11.2f2.3
Two layer
Three layer
As a general rule, the detected signal increased with the exposure time and the number of streptavidin molecules used in the disclosing system. Even after 2 days exposure all the disclosing systems registered grain counts over cells in excess of background, (i.e. values for S/N 3 l), indicating hybridization to have occurred. The background grain counts were similar for all the streptavidin-biotin techniques and did not vary significantly with the number of SBAPC layers or the exposure time. The background was, however, between two and three times higher than that obtained using the directly labelled probe, presumably because of non-specific binding of streptavidin to the tissue. The increase in signal was much greater than the increase in noise, making the signal-tonoise ratio much higher. Unfortunately, with the two- and three-layer techniques the silver grain densities were very
56
J. A. HOYLAND AND A. J. FREEMONT
100
-
50
-
043
t3
o
n
CONC OI:
ARTIFICIAL “mRNA“
-
ndpl
10
W
5-
2.5-
DIRECT
ONE
LABEL
LAYER
Two L A Y FR
.,.
1IIREE LAYER
DISCLOSING TECHNIOUP. Fig. 2-Graph comparing the lowest concentration of artificial mRNA detected in each of the autoradiographic strips examined with the four disclosing techniques: (0)direct labelling; (0) one layer; ( 0 )two layer; and (H) three layer
high at long exposures and the cellular detail was obscured by overlapping grains. This made grain counting inaccurate and probably explains the smaller difference between the 5- and 8-day signalto-noise ratios and the corresponding 2- and 5-day ratios. The best definition was obtained with the three-layer technique at 2 days and with the two-layer technique at 5 days. Optimal streptavidin concentration ( 7 . 3 . ) Using the technique described, the optimal concentrations for the one-, two-, and three-layer amplification methods were found to be 1 x lo-*, and 0.1 p ~ respectively. , These represent 4x 1/50, 1/12, and l / 5 of the amount of streptavidin originally used.
Relationship between reaction product and amount of message [ 7 . 4 . ) The results of this experiment are shown in Figs 2 and 3 . Figure 2 shows the differences in minimum message concentration that can be detected using the four disclosing systems under the same conditions. When compared with the directly labelled probe, the two-layer technique offers a ten-fold increase in sensitivity and the three-layer technique a 100-fold increase. Figure 3 relates known concentrations of the artificial ‘mRNA’ disclosed following in situ hybridization using the two-layer technique to the silver grain density on X-ray film. There is a direct association between the measured reaction product density and the log of the ‘mRNA’ concentration.
57
AMPLIFICATION OF IN SITU HYBRIDIZATION
160
W
140
120
GREY LEVEL (MACIIINE UNITS)
'O0
80
60
0 0
40
0
20
- I'
I'
2.3
I'O
15 i o
l'oo
CONC ( n g / p l ) OF AI!TIFICIAI, "mRNA"
Fig. 3-Graph comparing the amount of reaction product (assessed by measuring the mean grey level of each plaque autoradiograph) with the known concentration of poly d(A) molecules in each plaque using the two-layer disclosing technique
DISCUSSION Oligodeoxynucleotide probes represent a new and potentially powerful histological tool for the study of cell function in disease state^.^ A major criticism of the use of these small probes is that the amount of label that can, in practice, be added to the probe is much smaller than that for an equivalent cDNA or riboprobe (cRNA). (For instance, a 1 kb cDNA probe with equal amounts of each base will acquire approximately 250 labelled bases during random primer labelling, whereas a 30 base oligonucleotide probe labelled can, in practice, be given no more than 20 labelled bases.) Although the sensitivity of the oligonucleotide probe will be no less than that of the cDNA probe (indeed, in
formalin-fixed tissue it may be higher because of the greater penetration of the smaller probe), the sensitivity of the detection system will be between one and two orders of magnitude smaller. One way of increasing the detectability of bound oligonucleotide probe is to amplify the signal after hybridization. Post-hybridization amplification of biotinylated oligonucleotides using the streptavidin biotin enzyme complex has already been reported."6 This type of amplification system usually gives a coloured reaction product, which although demonstrating very adequately the presence of a hybridization reaction, is not easily or reliably quantified. We have adapted this technique in two ways. First, by constructing the complex in stages on the slide, thus allowing the size of the complex
58
J. A. HOYLAND AND A. J. FREEMONT
(and thereby the number of streptavidin molecules) due to the autoradiographic exposure time has to be controlled. And second, by replacing strepta- always been a disadvantage in an autoradiographic vidin with "S-radiolabelled streptavidin, permiting disclosure system. The amplification technique could also be used to autoradiographic disclosure of hybridized probe aid in the detection of low numbers of mRNA and quantification. In practice, the amplification technique is copies by increasing either the exposure time or the straightforward (for a laboratory routinely using number of layers of streptavidin molecules further. In summary, a technique has been developed and immunohistochemistry) and gives reproducible investigated that allows the signal from oligoresults. The reaction product can be quantified by assess- nucleotide probes to be amplified and quantified. It ing the autoradiographic silver grain density in is simple, reproducible, and easily adapted to the emulsion or on overlaid X-ray film. On the nitro- nature of the mRNA being investigated. cellulose gel, increasing the number of streptavidin molecules resulted in a predictable increase in the number of autoradiographic grains; and for any ACKNOWLEDGEMENTS particular number of streptavidin molecules, the silver grain density increased directly with the This work was funded by the Arthritis and number of hybridization target molecules. Whether this happens in tissue is more difficult to assess Rheumatism Council for Research, and the North because the amount of mRNA even in sequential Western Regional Research Grants Committee. sections may vary and there is an unpredictable quenching effect of a proportion of radioactive emissions by tissue. REFERENCES Our data indicate that when compared with the directly labelled probe the radiostreptavidin-biotin 1. Caruthers MH. Synthesis of oligonucleotides and oligonucleotide amplification system can give up to a 100-fold analogues. In: Cohen JS. ed. Oligodeoxynucleotides: Antisense Inhibitors of Gene Expression. Basingstoke: Macmillan, 1989: 7-25. increase in signal per labelled base, without an M. Application of avidin-biotin complex technique for the appreciable increase in noise. By amplification the 2. Sar localisation of estradiol receptor in target tissues using monoclonal signal from an oligonucleotide tagged with as few antibodies In: Bullock G R , Petrusr P, eds. Techniques in ImmunoVol 3. London: Academic Preaa. 1988: 43-54. as 2 bases can become, in effect, comparable to 3. cytochemistry, Pringle J H , Primrose L, Kind CN. Talbot IC, Lauder 1 In-situ that from a 1 kb cDNA probe containing 200 hybridization demonstration of poly-adenylated R N A sequences in formalin-fixed paraffin sections using a biotinylated oligonucleotide radiolabelled bases. poly d(T) probe. J Pathol 1989; 158: 279-286. The efficiency of the detection system was such 4. Van Prooijen-Knegt AC, Raap AK, Van der Burg MJM. Spreading that in this study the grain density achieved under and staining of human metaphase chromosomes on aminoalkylsilane treated glass slides. Hi.ytockem J 1983; 1 4 333-334. some of the amplification conditions was so great Deng G, Wu R.Terminal transferase: use in the tailing of D N A and for that cellular detail was obscured. This aspect of the 5. in-viiro niutagenesis. Metiioih Enzjnzol1983; 9 6 96.- I 16. amplification system could be exploited to shorten 6. Mougin C . Guitteny AF. Fouque B. Viennet G, Teoule R, Bloch B. Histochemical detection of the messenger RNAs coding for calcitonin the exposure time if the technique is to be used with a and calcitonin gene-related peptide in medullary thyroid carcinomas highly transcribed message. This may be a signifiwith radioactive and biotinylated oligonucleotide probes. J Polhol 1990; 160: 187-194. cant advantage as the delay in obtaining results
