Surgical
Oncology 1992 ; 1 : 309-314
309
Expression polymerase chain reaction : a sensitive method for analysis of gene expression in human tumours W . G . CANCE`f, R . J . CRAVEN`$ AND E . T . LIUft
Departments of "Surgery and tMedicine, and the "Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
To analyse the expression of individual genes in small tumour samples, we have used the method of RNA polymerase chain reaction (PCR) to develop a technique which we have termed expression PCR . With this technique, specific cDNA sequences of a target gene are amplified, analysed by gel electrophoresis, and semi-quantitated using laser densitometry. Alpha-actin is amplified as a reference gene to control for template RNA and each target gene is analysed at several cycle numbers to optimize PCR dynamics . In this study, we have demonstrated expression PCR by analysing the levels of expression of tyrosine kinase genes in a panel of human tumours . We have compared expression PCR with Northern analysis to show that these techniques provide equivalent information on relative levels of gene transcription, with expression PCR requiring 100fold less RNA- This technique is sufficiently sensitive to detect and compare the levels of expression of genes not seen on Northern analysis and is ideally suited for analysing the expression of multiple genes within the same portion of a tumour . Surgical Oncology 1992 ; 1 : 309-314 .
Keywords : gene expression, RNA polymerase chain reaction .
INTRODUCTION The pathogenesis of human tumours involves the activation and expression of numerous oncogenes and growth factors, as well as the inactivation of suppressor genes . As our understanding of this process has progressed, many novel genes have been identified and characterized which relate to the initiation and progression of tumours . Since the levels of gene expression can provide important insight into the biology and clinical behaviour of a tumour [1-3], it is essential to have a method for rapidly analysing gene expression . We have used the RNA polymerase chain reaction (PCR) technique to develop a sensitive method for
Correspondence : William
G . Lance, MD, Department of Surgery, 3010 Old Clinic Bldg ., CB #7210, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA .
determining the relative levels of gene transcription in a tumour . We have termed this technique expression PCR, as it is semi-quantitative and can compare the levels of gene expression between multiple tumour samples or can analyse the levels of expression of multiple genes in a single tumour, Expression PCR is simple to accomplish, does not require radionucleotides and provides a more sensitive technique than Northern analysis, while requiring only miniscule amounts of RNA to analyse each gene . The technique of expression PCR is applicable to any gene in any tumour. In this study, we describe the method of expression PCR, the optimization of the PCR dynamics, and demonstrate this technique for detecting the levels of RNA expression of a variety of tyrosine kinase genes (e .g . HER-2/neu, epidermal growth factor receptor) in human tumours .
W. G. Cance et al .
310 MATERIALS AND METHODS
5 mm MgCI 2 , 0 .01% gelatin, 0.1% Triton X-100), 200 µM dNTP, 1 um of each PCR primer, 2 .5 U Taq poly-
Human tumours, RNA extraction and Northern
merase (Promega) .
analysis PCR reaction. One microlitre of the cDNA sample Primary human tumours were obtained from operative specimens at the University of North Carolina Hospitals . RNA was extracted from the tumours using the guanidinium isothiocyanate/caesium chloride method
[4] .
prepared by the method above was used in each PCR reaction . Following a 5 minute cycle at 94°C to denature the DNA, each subsequent PCR cycle consisted of annealing at 55°C for 1 min, extension at
Northern analysis was per-
72°C for 1 min and denaturation at 94°C for 1 min .
formed for the HER-2/neu gene using standard tech-
The total number of cycles used for each target
niques [5] .
gene was determined in a separate experiment (see results below) .
cDNA synthesis Po/yacrylamide gel analysis of expression PCR Three micrograms of total RNA was denatured at
results . The PCR products were analysed by electro-
95°C for 5 min in the presence of 100 pmol of random hexamers (Promega) and then cooled on ice
phoresis on a 10% polyacrylamide gel (Mighty Small
for 2 min . The RNA was reverse transcribed in a
microlitres of each PCR reaction were mixed with 1
total volume of 20 pi
,ul
containing 1 x PCR buffer
mini gel electrophoresis system, Hoeffer) . Five of sample buffer (10% glycerol, 20% ficoll and
(50 mm KCI, 10 mm Tris, pH 9.0, 5 mm MgCI 2 , 0 .01 gelatin, 0 .1 % Triton X-100), 625 pM each deoxyribo-
0 .25% bromophenol blue) . The samples were
nucleotide triphosphate (dNTP), 10 mm dithiothreitol, 200 U Superscript Moloney Murine Leukemia Virus
ethidium bromide and photographed using Polaroid 665 film to obtain a photographic negative. The peak
Reverse Transcriptase (BRL), and 20 U RNasin
area of the resulting bands on the negative was
RNase inhibitor (Promega) . This mixture was
quantified using a LKB ultroscan XL laser den-
incubated at room temperature for 10 min, 42°C for
sitometer . For some genes, it was necessary to
45 min, and 95°C for 5 min .
adjust the amount of PCR product loaded on the
electrophoresed at 70 Volts, stained in 0 .5 µg ml - '
To control for the possibility of genomic DNA con-
polyacrylamide gel in order to obtain accurate den-
tamination, cDNA reactions were also performed without reverse transcriptase .
sitometry . For each tumour, the results were expressed as a ratio of the peak areas of the target gene to actin after correction for the amount of PCR product loaded .
Expression PCR
Oligonucleotide
primers .
For each target gene
analysed, specific oligonucleotide primers were syn-
RESULTS
thesized . The primers were designed using the cDNA sequences of the selected genes and amplified a 150-200 base pair (bp) cDNA fragment . The
Optimization of the expression PCR technique
amplimer sequences for HER-2/neu were as follows : 5'AAGAGTCCCAACCATGTCAAA, 3'ACTCTGGTGGGTGAACCGCCG . For the alpha-actin control amplifications in each experiment (see below), the following amplimers were used : 5'CCTTCCT000CATGGAGTCCT, and 3'GGAGCAATGATCTTGATCTTC . Master mix. For each expression PCR experiment, a master mix of all PCR reagents was prepared and individually aliquoted to each reaction tube . The basic PCR reaction mix was a 50 uI reaction containing 1 x PCR buffer (50 mm KCI, 10 mm Tris, pH 9 .0,
Since the PCR amplification process is saturable at high cycle numbers (6], it is necessary to optimize the cycle number for each gene . To identify the cycle numbers that gave optimal quantitation of HER-2/neu, expression PCR was performed on RNA from two cell lines with varying levels of HER-2/neu expression . In SK-BR-3, (a HER-2/neu overexpressor) PCR products were detected at 20 cycles, peaking at 32 cycles, and plateauing at greater than 32 cycles (Fig . 1) . For MCF-7, (a low HER-2/neu expressor), PCR products were detected only at 26
Expression PCR analysis of human tumours cycles, peaking at 32 cycles, and plateauing thereafter . Thus, when SK-BR-3 and MCF7 were compared between cycle numbers 26 and 30, expression PCR permitted the appropriate detection of overexpression in SK-BR-3 as compared with MCF-7 . However at greater than 32 cycles, MCF-7 and SKBR-3 appeared to give the same expression results . Thus, quantitation of RNA levels by expression PCR should be performed at the cycle number giving a proportional signal :cycle number relationship . We have empirically determined these optimal cycle numbers to be 18-22 for actin and 26-32 for most tyrosine kinases we have examined .
Comparison of expression PCR and Northern analysis To compare the ability of expression PCR and Northern analysis to determine the relative levels of gene transcription, we analysed the expression of the HER-2/neu oncogene in 10 human tumours . Those tumours with high levels of HER-2/neu transcripts by Northern analysis were identified as high expressors by expression PCR (Fig . 2, lanes 2, B, 9 and 10) . Conversely, tumours with low amounts of HER-2/neu transcripts were identified as low expres-
31 1
sors by expression PCR (lanes 5, 6 and 7) . The Northern analysis, however, required 15 mg of total RNA from each tumour sample . In contrast, expression PCR was able to give comparable information using 150 ng of RNA . Thus, 100 times less RNA was used with expression PCR than with Northern analysis .
Analysis of genes with low levels of expression Since the PCR technique is exquisitely sensitive, it was possible to use expression PCR to analyse genes which are transcribed at low levels and are undetectable by Northern analysis . We determined the levels of expression of a novel tyrosine kinase isolated in our laboratory, TK1 [7], in the same panel of 10 tumours used in the HER-2/neu experiment a bove . TM expression was not seen in Northern blot analyses, but was detectable by expression PCR in 6 of the 10 tumours (Fig . 3) . This gene was expressed at variable levels among the tumours, with the highest levels in tumours 1 and 9 . By expressing the densitometric data for HER-2/ neu and TK1 as kinase :actin ratios, the variable
levels of expression of these genes were demonstrated in a graphic form (Fig . 4) . This extends the
Expression PLR 1 .4
1 .2
Northern analysis 0 .8
a 0.6
Y 0.4
0 .2
• SK-BR-3 O MCF-7
0 20 25
30
1 35
Noo cycles
Figure 1 . Optimization of the PCR dynamics for HER-2/ neu. Expression PCR was performed at several cycle numbers to analyse levels of expression in MCF-7, a low expressor of HER-2/neu and SK-BR-3, an overexpressor of HER-2/neu . Between 26 and 30 cycles, there is a proportional signal :cycle number relationship, with saturation of the PCR after 32 cycles .
Figure 2 . Comparison of expression PCR and Northern analysis in determining the relative levels of expression of the HER-2/neu oncogene in 10 human tumours . The expression PCR figure shows an ethidium bromide-stained polyacrylamide gel on which 2 pl of PCR products obtained at 26 cycles were analysed in each lane . The Northern analysis figure shows an autoradiograph of a blot which was probed with a specific 32 P-labelled cDNA fragment after analysing 15 pg of total RNA in each lane . The panel of tumours represents 9 human breast carcinomas and 1 metastatic melanoma (lane 7) .
31 2
W. G. Cance
I
2
3
4
5
6
7
a
9
10
TKI
et al . 5
4
a 3 2
Actin
Figure 3 . Expression PCR analysis of TK1, a gene with low levels of expression, in the 10 human tumours. The ethidium bromide-stained gel for TK1 was obtained by loading 10 ul of PCR products obtained at 26 cycles, and the gel for actin was obtained by loading 5 ul of PCR products obtained at 20 cycles . The actin amplifications serve as a control for the amount of input RNA from each tumour . The same panel of tumours was used for the TK1 and the HER-2/neu experiment described above .
0 I
2
3
4
5
6
7
a
9
10
Tumour
Figure 4. Comparison of the levels of expression of HER2/neu and TK1 in the panel of 10 tumours . From the results of the laser densitometry of the gels shown in Figs 2 and 3, the HER-2/neu/actin and TK1/actin ratios were calculated and corrected for the variations in gel loading . The variable levels of expression of these genes is shown in the tumours, as well as the relatively higher levels of HER-2/neu .
Northern blot results and demonstrates that HER-2/ neu is expressed at significantly higher levels than TK1 in this panel of tumours .
Analysis of a panel of genes in individual tumours by expression PCR We used expression PCR to analyse 3 human breast tumours for the levels of expression of a panel of tyrosine kinases including HER-2/neu, the epidermal growth factor receptor (EGFR), the insulin-like growth factor I receptor, the fer oncogene, and TK1 (Fig . 5) . The relative expression of these genes was highly variable . For example, EGFR was expressed at an equivalent level as HER-2/neu in tumour 3, whereas tumour 8 has over 10 times more HER-2/ neu transcript than EGFR . Furthermore, tumour 8 expressed low levels of all of the kinases, except HER-2/neu, in contrast to tumour 1 which had a more equivalent distribution of expression . Thus, expression PCR can be applied to determine the relative expression of a panel of genes in primary tumour samples.
DISCUSSION Expression PCR is a technique based on RNA PCR, which has been optimized to determine the relative levels of gene expression in human tumours . This technique is applicable to any gene family and provides a simple, non-radioactive method of semi-
ereasr tumour
Figure 5 . Expression PCR analysis of a panel of tyrosine kinase genes in 3 human breast tumours . Relative levels of expression of the epidermal growth factor receptor (EGFR), the fer oncogene, the insulin-like growth factor I receptor (IGF1-R), HER-2/neu, and TK1 are compared by their kinase :actin ratios .
quantitating the levels of gene expression, while requiring only small amounts of RNA . These features make expression PCR particularly suited for analysis of genes with low levels of expression or tumours with limited amounts of RNA available for study . We have been able to perform expression PCR using as little as 30 ng of tumour RNA to determine the expression of a single gene (data not shown) . Other investigators have used a similar technique in cell lines and found that 2000 cells provides sufficient RNA to analyse a single gene [8] . This contrasts to Northern analysis which requires approximately 15 ug of RNA per gene . We
Expression PCR analysis of human tumours
313
have shown that both expression PCR and Northern
whether a gene is variably expressed in a group of
analysis are equivalent techniques for determining
tumours .
relative levels of gene expression in tumours . Northern analysis does determine the size of the
quantitative PCR which have been reported for
mRNA transcript, which is not possible to assess by
calculating the amounts of mRNA in a sample .
expression PCR . Thus, expression PCR cannot detect
These other methods use a synthetic internal RNA
aberrantly spliced transcripts or gene rearrange-
standard containing a small deletion or insertion in
ments . However, expression PCR can be used to
the gene of interest (10, 11) . This synthetic gene is
compare levels of expression of genes whose tran-
reverse transcribed and co-amplified with the target
scripts are undetectable by Northern analysis, such
DNA, acting as a competitor fragment in the PCR
as the DCC gene in colon cancer [9], which is
reaction . Since a known amount of synthetic RNA is
expressed at only low levels in primary tumours .
co-amplified, a standard curve for each gene is
This technique contrasts with other methods of
Because of the minimal requirements for amounts
generated and used to calculate the amount of
of RNA, it is possible to use expression PCR to
target mRNA . With an internal RNA standard, these
analyse multiple genes within the same area of
techniques can be used to determine the number of
tumour, obviating the potential problem of sampling
mRNA molecules in a given specimen [12] . How-
from different sites within a heterogeneous tumour .
ever, different synthetic standards need to be
Furthermore, this allows these analyses to be performed from individual tissue sections, making it
synthesized and quantitated for each gene studied, which would not be practical for analysing the
feasible for expression PCR to be used in a clinical
expression of a large number of different genes .
setting .
Furthermore, for many genes of interest, such as the
Expression PCR is a semi-quantitative technique .
HER-2/neu gene in breast cancer, this level of sensi-
The results are calculated from the densitometric
tivity is unnecessary, since the relative level of
ratios of the target gene to actin . We have assumed
expression is most important [1], rather than the
that actin is a 'housekeeping' gene which is
absolute number of molecules of mRNA .
expressed at a reasonably constant level in each tumour . By amplifying actin, we not only have con-
In conclusion, expression PCR is a technique that
trolled for the cDNA reactions, but also have a
provides the capability for simple and rapid analysis of levels of gene expression and which is easily
means for comparing the relative levels of gene
adaptable for the analysis of small tissue sections .
expression between different tumour samples, simi-
Such an approach may aid in determining the
lar to that used in standard Northern analysis [3] .
importance of certain genes in the process of
In order for this technique to be semi-quantitative, it is essential to calculate PCR amplification
tumourigenesis, as new oncogenes, growth factors and suppressor genes are identified .
dynamics for each gene so that amplification is measured at a point before the PCR reaction has been saturated . We have found that the dynamics
ACKNOWLEDGEMENTS
for most tyrosine kinase genes are optimal between 26 and 30 cycles . Furthermore, kinases which are
This work was supported by American Cancer
easily detectable at 26 cycles can be visualized on
Society Grant #PDT-395 (E .T .L.) . WGC is a recipient
Northern blots, whereas kinases not detectable at 26
of an American Cancer Society Clinical Oncology Career Development Award .
cycles but visualized at 30 cycles are not detectable by standard Northern blot analysis (data not shown) . We have used expression PCR as a rapid method of determining the levels of expression of novel tyro-
REFERENCES
sine kinase genes in human tumours . With this degree of semi-quantitation, we can analyse a
1 . Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A,
variety of tumour types to determine the expression
McGuire WL . Human breast cancer : correlation of relapse and survival with amplification of the HER-2/ neu oncogene . Science 1987 ; 235 : 177-82 . 2 . Sainsbury JRC, Farndon JR, Needham GK, Malcolm
profile of particular kinases in these tissues . In addition, we can determine whether a gene is preferentially expressed by tumour or normal tissues, and
314
W. G. Cance et al . AJ, Harris AL . Epidermal-growth-factor receptor status
8. Golay J, Passerini F, Introna M . A simple and rapid
as predictor of early recurrence of and death from
method to analyze specific mRNAs from few cells in a semi-quantitative way using the polymerase chain
breast cancer.
Lancet
1987 ; 1 : 1398-402 .
3 . Liu C, Park M, Tsao MS . Overexpression of c-met proto-oncogene but not epidermal
growth factor
receptor or c-erb-B2 in primary human colorectal carcinoma .
Oncogene
1992 ; 7 : 181-5 .
cancers .
4 . Glisin V, Crkvenjakow R, Byus C . Ribonucleic acid isolated
by cesium
chloride
centrifugation .
Bio-
chemistry 1974 ; 13 : 2633-7-
Method Appl
1991 ; 1 : 144-5 .
Science
1990 ; 247 : 49-56-
10 . Wang AM, Doyle MV, Mark DF . Quantitation of mRNA by the polymerase chain reaction . Proc Nati Acad Sci USA 1989 ; 86 : 9717-21 .
5 . Sambrook J, Fritsch EF, Maniatis T . Molecular cloning : a laboratory manual, New York, Cold Spring Harbor
11 . Uberla K, Platzer C, Diamantstein, Blankenstein T. Generation of competitor DNA fragments for quantitative PCR . PCR Method Appi 1991 ; 1 : 136-9 .
Laboratory Press 1989 ; 7 .39-7.52 . 6 . Frye RA, Benz CC, Liu ET. Detection of amplified oncogenes by differential polymerase chain reaction .
reaction . PCR
9 . Fearon ER, Cho KR, Nigro JM, et al. Identification of a chromosome 18q gene that is altered in colorectal
Onco-
gene 1989 ; 4 : 1153-7 . 7 . Cance WG, Craven RJ, Weiner TM, Liu ET. Novel protein kinases in human breast cancer (to be submitted) .
12 . Baldwin GS, Zhang QX . Measurement of gastrin and transforming growth factor alpha messenger RNA levels in colonic carcinoma cell lines by quantitative polymerase chain reaction . 2261-7 .
Cancer Res 1992 ; 52 :
Oncology 1992 ; 1 : 309-314
309
Expression polymerase chain reaction : a sensitive method for analysis of gene expression in human tumours W . G . CANCE`f, R . J . CRAVEN`$ AND E . T . LIUft
Departments of "Surgery and tMedicine, and the "Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
To analyse the expression of individual genes in small tumour samples, we have used the method of RNA polymerase chain reaction (PCR) to develop a technique which we have termed expression PCR . With this technique, specific cDNA sequences of a target gene are amplified, analysed by gel electrophoresis, and semi-quantitated using laser densitometry. Alpha-actin is amplified as a reference gene to control for template RNA and each target gene is analysed at several cycle numbers to optimize PCR dynamics . In this study, we have demonstrated expression PCR by analysing the levels of expression of tyrosine kinase genes in a panel of human tumours . We have compared expression PCR with Northern analysis to show that these techniques provide equivalent information on relative levels of gene transcription, with expression PCR requiring 100fold less RNA- This technique is sufficiently sensitive to detect and compare the levels of expression of genes not seen on Northern analysis and is ideally suited for analysing the expression of multiple genes within the same portion of a tumour . Surgical Oncology 1992 ; 1 : 309-314 .
Keywords : gene expression, RNA polymerase chain reaction .
INTRODUCTION The pathogenesis of human tumours involves the activation and expression of numerous oncogenes and growth factors, as well as the inactivation of suppressor genes . As our understanding of this process has progressed, many novel genes have been identified and characterized which relate to the initiation and progression of tumours . Since the levels of gene expression can provide important insight into the biology and clinical behaviour of a tumour [1-3], it is essential to have a method for rapidly analysing gene expression . We have used the RNA polymerase chain reaction (PCR) technique to develop a sensitive method for
Correspondence : William
G . Lance, MD, Department of Surgery, 3010 Old Clinic Bldg ., CB #7210, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA .
determining the relative levels of gene transcription in a tumour . We have termed this technique expression PCR, as it is semi-quantitative and can compare the levels of gene expression between multiple tumour samples or can analyse the levels of expression of multiple genes in a single tumour, Expression PCR is simple to accomplish, does not require radionucleotides and provides a more sensitive technique than Northern analysis, while requiring only miniscule amounts of RNA to analyse each gene . The technique of expression PCR is applicable to any gene in any tumour. In this study, we describe the method of expression PCR, the optimization of the PCR dynamics, and demonstrate this technique for detecting the levels of RNA expression of a variety of tyrosine kinase genes (e .g . HER-2/neu, epidermal growth factor receptor) in human tumours .
W. G. Cance et al .
310 MATERIALS AND METHODS
5 mm MgCI 2 , 0 .01% gelatin, 0.1% Triton X-100), 200 µM dNTP, 1 um of each PCR primer, 2 .5 U Taq poly-
Human tumours, RNA extraction and Northern
merase (Promega) .
analysis PCR reaction. One microlitre of the cDNA sample Primary human tumours were obtained from operative specimens at the University of North Carolina Hospitals . RNA was extracted from the tumours using the guanidinium isothiocyanate/caesium chloride method
[4] .
prepared by the method above was used in each PCR reaction . Following a 5 minute cycle at 94°C to denature the DNA, each subsequent PCR cycle consisted of annealing at 55°C for 1 min, extension at
Northern analysis was per-
72°C for 1 min and denaturation at 94°C for 1 min .
formed for the HER-2/neu gene using standard tech-
The total number of cycles used for each target
niques [5] .
gene was determined in a separate experiment (see results below) .
cDNA synthesis Po/yacrylamide gel analysis of expression PCR Three micrograms of total RNA was denatured at
results . The PCR products were analysed by electro-
95°C for 5 min in the presence of 100 pmol of random hexamers (Promega) and then cooled on ice
phoresis on a 10% polyacrylamide gel (Mighty Small
for 2 min . The RNA was reverse transcribed in a
microlitres of each PCR reaction were mixed with 1
total volume of 20 pi
,ul
containing 1 x PCR buffer
mini gel electrophoresis system, Hoeffer) . Five of sample buffer (10% glycerol, 20% ficoll and
(50 mm KCI, 10 mm Tris, pH 9.0, 5 mm MgCI 2 , 0 .01 gelatin, 0 .1 % Triton X-100), 625 pM each deoxyribo-
0 .25% bromophenol blue) . The samples were
nucleotide triphosphate (dNTP), 10 mm dithiothreitol, 200 U Superscript Moloney Murine Leukemia Virus
ethidium bromide and photographed using Polaroid 665 film to obtain a photographic negative. The peak
Reverse Transcriptase (BRL), and 20 U RNasin
area of the resulting bands on the negative was
RNase inhibitor (Promega) . This mixture was
quantified using a LKB ultroscan XL laser den-
incubated at room temperature for 10 min, 42°C for
sitometer . For some genes, it was necessary to
45 min, and 95°C for 5 min .
adjust the amount of PCR product loaded on the
electrophoresed at 70 Volts, stained in 0 .5 µg ml - '
To control for the possibility of genomic DNA con-
polyacrylamide gel in order to obtain accurate den-
tamination, cDNA reactions were also performed without reverse transcriptase .
sitometry . For each tumour, the results were expressed as a ratio of the peak areas of the target gene to actin after correction for the amount of PCR product loaded .
Expression PCR
Oligonucleotide
primers .
For each target gene
analysed, specific oligonucleotide primers were syn-
RESULTS
thesized . The primers were designed using the cDNA sequences of the selected genes and amplified a 150-200 base pair (bp) cDNA fragment . The
Optimization of the expression PCR technique
amplimer sequences for HER-2/neu were as follows : 5'AAGAGTCCCAACCATGTCAAA, 3'ACTCTGGTGGGTGAACCGCCG . For the alpha-actin control amplifications in each experiment (see below), the following amplimers were used : 5'CCTTCCT000CATGGAGTCCT, and 3'GGAGCAATGATCTTGATCTTC . Master mix. For each expression PCR experiment, a master mix of all PCR reagents was prepared and individually aliquoted to each reaction tube . The basic PCR reaction mix was a 50 uI reaction containing 1 x PCR buffer (50 mm KCI, 10 mm Tris, pH 9 .0,
Since the PCR amplification process is saturable at high cycle numbers (6], it is necessary to optimize the cycle number for each gene . To identify the cycle numbers that gave optimal quantitation of HER-2/neu, expression PCR was performed on RNA from two cell lines with varying levels of HER-2/neu expression . In SK-BR-3, (a HER-2/neu overexpressor) PCR products were detected at 20 cycles, peaking at 32 cycles, and plateauing at greater than 32 cycles (Fig . 1) . For MCF-7, (a low HER-2/neu expressor), PCR products were detected only at 26
Expression PCR analysis of human tumours cycles, peaking at 32 cycles, and plateauing thereafter . Thus, when SK-BR-3 and MCF7 were compared between cycle numbers 26 and 30, expression PCR permitted the appropriate detection of overexpression in SK-BR-3 as compared with MCF-7 . However at greater than 32 cycles, MCF-7 and SKBR-3 appeared to give the same expression results . Thus, quantitation of RNA levels by expression PCR should be performed at the cycle number giving a proportional signal :cycle number relationship . We have empirically determined these optimal cycle numbers to be 18-22 for actin and 26-32 for most tyrosine kinases we have examined .
Comparison of expression PCR and Northern analysis To compare the ability of expression PCR and Northern analysis to determine the relative levels of gene transcription, we analysed the expression of the HER-2/neu oncogene in 10 human tumours . Those tumours with high levels of HER-2/neu transcripts by Northern analysis were identified as high expressors by expression PCR (Fig . 2, lanes 2, B, 9 and 10) . Conversely, tumours with low amounts of HER-2/neu transcripts were identified as low expres-
31 1
sors by expression PCR (lanes 5, 6 and 7) . The Northern analysis, however, required 15 mg of total RNA from each tumour sample . In contrast, expression PCR was able to give comparable information using 150 ng of RNA . Thus, 100 times less RNA was used with expression PCR than with Northern analysis .
Analysis of genes with low levels of expression Since the PCR technique is exquisitely sensitive, it was possible to use expression PCR to analyse genes which are transcribed at low levels and are undetectable by Northern analysis . We determined the levels of expression of a novel tyrosine kinase isolated in our laboratory, TK1 [7], in the same panel of 10 tumours used in the HER-2/neu experiment a bove . TM expression was not seen in Northern blot analyses, but was detectable by expression PCR in 6 of the 10 tumours (Fig . 3) . This gene was expressed at variable levels among the tumours, with the highest levels in tumours 1 and 9 . By expressing the densitometric data for HER-2/ neu and TK1 as kinase :actin ratios, the variable
levels of expression of these genes were demonstrated in a graphic form (Fig . 4) . This extends the
Expression PLR 1 .4
1 .2
Northern analysis 0 .8
a 0.6
Y 0.4
0 .2
• SK-BR-3 O MCF-7
0 20 25
30
1 35
Noo cycles
Figure 1 . Optimization of the PCR dynamics for HER-2/ neu. Expression PCR was performed at several cycle numbers to analyse levels of expression in MCF-7, a low expressor of HER-2/neu and SK-BR-3, an overexpressor of HER-2/neu . Between 26 and 30 cycles, there is a proportional signal :cycle number relationship, with saturation of the PCR after 32 cycles .
Figure 2 . Comparison of expression PCR and Northern analysis in determining the relative levels of expression of the HER-2/neu oncogene in 10 human tumours . The expression PCR figure shows an ethidium bromide-stained polyacrylamide gel on which 2 pl of PCR products obtained at 26 cycles were analysed in each lane . The Northern analysis figure shows an autoradiograph of a blot which was probed with a specific 32 P-labelled cDNA fragment after analysing 15 pg of total RNA in each lane . The panel of tumours represents 9 human breast carcinomas and 1 metastatic melanoma (lane 7) .
31 2
W. G. Cance
I
2
3
4
5
6
7
a
9
10
TKI
et al . 5
4
a 3 2
Actin
Figure 3 . Expression PCR analysis of TK1, a gene with low levels of expression, in the 10 human tumours. The ethidium bromide-stained gel for TK1 was obtained by loading 10 ul of PCR products obtained at 26 cycles, and the gel for actin was obtained by loading 5 ul of PCR products obtained at 20 cycles . The actin amplifications serve as a control for the amount of input RNA from each tumour . The same panel of tumours was used for the TK1 and the HER-2/neu experiment described above .
0 I
2
3
4
5
6
7
a
9
10
Tumour
Figure 4. Comparison of the levels of expression of HER2/neu and TK1 in the panel of 10 tumours . From the results of the laser densitometry of the gels shown in Figs 2 and 3, the HER-2/neu/actin and TK1/actin ratios were calculated and corrected for the variations in gel loading . The variable levels of expression of these genes is shown in the tumours, as well as the relatively higher levels of HER-2/neu .
Northern blot results and demonstrates that HER-2/ neu is expressed at significantly higher levels than TK1 in this panel of tumours .
Analysis of a panel of genes in individual tumours by expression PCR We used expression PCR to analyse 3 human breast tumours for the levels of expression of a panel of tyrosine kinases including HER-2/neu, the epidermal growth factor receptor (EGFR), the insulin-like growth factor I receptor, the fer oncogene, and TK1 (Fig . 5) . The relative expression of these genes was highly variable . For example, EGFR was expressed at an equivalent level as HER-2/neu in tumour 3, whereas tumour 8 has over 10 times more HER-2/ neu transcript than EGFR . Furthermore, tumour 8 expressed low levels of all of the kinases, except HER-2/neu, in contrast to tumour 1 which had a more equivalent distribution of expression . Thus, expression PCR can be applied to determine the relative expression of a panel of genes in primary tumour samples.
DISCUSSION Expression PCR is a technique based on RNA PCR, which has been optimized to determine the relative levels of gene expression in human tumours . This technique is applicable to any gene family and provides a simple, non-radioactive method of semi-
ereasr tumour
Figure 5 . Expression PCR analysis of a panel of tyrosine kinase genes in 3 human breast tumours . Relative levels of expression of the epidermal growth factor receptor (EGFR), the fer oncogene, the insulin-like growth factor I receptor (IGF1-R), HER-2/neu, and TK1 are compared by their kinase :actin ratios .
quantitating the levels of gene expression, while requiring only small amounts of RNA . These features make expression PCR particularly suited for analysis of genes with low levels of expression or tumours with limited amounts of RNA available for study . We have been able to perform expression PCR using as little as 30 ng of tumour RNA to determine the expression of a single gene (data not shown) . Other investigators have used a similar technique in cell lines and found that 2000 cells provides sufficient RNA to analyse a single gene [8] . This contrasts to Northern analysis which requires approximately 15 ug of RNA per gene . We
Expression PCR analysis of human tumours
313
have shown that both expression PCR and Northern
whether a gene is variably expressed in a group of
analysis are equivalent techniques for determining
tumours .
relative levels of gene expression in tumours . Northern analysis does determine the size of the
quantitative PCR which have been reported for
mRNA transcript, which is not possible to assess by
calculating the amounts of mRNA in a sample .
expression PCR . Thus, expression PCR cannot detect
These other methods use a synthetic internal RNA
aberrantly spliced transcripts or gene rearrange-
standard containing a small deletion or insertion in
ments . However, expression PCR can be used to
the gene of interest (10, 11) . This synthetic gene is
compare levels of expression of genes whose tran-
reverse transcribed and co-amplified with the target
scripts are undetectable by Northern analysis, such
DNA, acting as a competitor fragment in the PCR
as the DCC gene in colon cancer [9], which is
reaction . Since a known amount of synthetic RNA is
expressed at only low levels in primary tumours .
co-amplified, a standard curve for each gene is
This technique contrasts with other methods of
Because of the minimal requirements for amounts
generated and used to calculate the amount of
of RNA, it is possible to use expression PCR to
target mRNA . With an internal RNA standard, these
analyse multiple genes within the same area of
techniques can be used to determine the number of
tumour, obviating the potential problem of sampling
mRNA molecules in a given specimen [12] . How-
from different sites within a heterogeneous tumour .
ever, different synthetic standards need to be
Furthermore, this allows these analyses to be performed from individual tissue sections, making it
synthesized and quantitated for each gene studied, which would not be practical for analysing the
feasible for expression PCR to be used in a clinical
expression of a large number of different genes .
setting .
Furthermore, for many genes of interest, such as the
Expression PCR is a semi-quantitative technique .
HER-2/neu gene in breast cancer, this level of sensi-
The results are calculated from the densitometric
tivity is unnecessary, since the relative level of
ratios of the target gene to actin . We have assumed
expression is most important [1], rather than the
that actin is a 'housekeeping' gene which is
absolute number of molecules of mRNA .
expressed at a reasonably constant level in each tumour . By amplifying actin, we not only have con-
In conclusion, expression PCR is a technique that
trolled for the cDNA reactions, but also have a
provides the capability for simple and rapid analysis of levels of gene expression and which is easily
means for comparing the relative levels of gene
adaptable for the analysis of small tissue sections .
expression between different tumour samples, simi-
Such an approach may aid in determining the
lar to that used in standard Northern analysis [3] .
importance of certain genes in the process of
In order for this technique to be semi-quantitative, it is essential to calculate PCR amplification
tumourigenesis, as new oncogenes, growth factors and suppressor genes are identified .
dynamics for each gene so that amplification is measured at a point before the PCR reaction has been saturated . We have found that the dynamics
ACKNOWLEDGEMENTS
for most tyrosine kinase genes are optimal between 26 and 30 cycles . Furthermore, kinases which are
This work was supported by American Cancer
easily detectable at 26 cycles can be visualized on
Society Grant #PDT-395 (E .T .L.) . WGC is a recipient
Northern blots, whereas kinases not detectable at 26
of an American Cancer Society Clinical Oncology Career Development Award .
cycles but visualized at 30 cycles are not detectable by standard Northern blot analysis (data not shown) . We have used expression PCR as a rapid method of determining the levels of expression of novel tyro-
REFERENCES
sine kinase genes in human tumours . With this degree of semi-quantitation, we can analyse a
1 . Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A,
variety of tumour types to determine the expression
McGuire WL . Human breast cancer : correlation of relapse and survival with amplification of the HER-2/ neu oncogene . Science 1987 ; 235 : 177-82 . 2 . Sainsbury JRC, Farndon JR, Needham GK, Malcolm
profile of particular kinases in these tissues . In addition, we can determine whether a gene is preferentially expressed by tumour or normal tissues, and
314
W. G. Cance et al . AJ, Harris AL . Epidermal-growth-factor receptor status
8. Golay J, Passerini F, Introna M . A simple and rapid
as predictor of early recurrence of and death from
method to analyze specific mRNAs from few cells in a semi-quantitative way using the polymerase chain
breast cancer.
Lancet
1987 ; 1 : 1398-402 .
3 . Liu C, Park M, Tsao MS . Overexpression of c-met proto-oncogene but not epidermal
growth factor
receptor or c-erb-B2 in primary human colorectal carcinoma .
Oncogene
1992 ; 7 : 181-5 .
cancers .
4 . Glisin V, Crkvenjakow R, Byus C . Ribonucleic acid isolated
by cesium
chloride
centrifugation .
Bio-
chemistry 1974 ; 13 : 2633-7-
Method Appl
1991 ; 1 : 144-5 .
Science
1990 ; 247 : 49-56-
10 . Wang AM, Doyle MV, Mark DF . Quantitation of mRNA by the polymerase chain reaction . Proc Nati Acad Sci USA 1989 ; 86 : 9717-21 .
5 . Sambrook J, Fritsch EF, Maniatis T . Molecular cloning : a laboratory manual, New York, Cold Spring Harbor
11 . Uberla K, Platzer C, Diamantstein, Blankenstein T. Generation of competitor DNA fragments for quantitative PCR . PCR Method Appi 1991 ; 1 : 136-9 .
Laboratory Press 1989 ; 7 .39-7.52 . 6 . Frye RA, Benz CC, Liu ET. Detection of amplified oncogenes by differential polymerase chain reaction .
reaction . PCR
9 . Fearon ER, Cho KR, Nigro JM, et al. Identification of a chromosome 18q gene that is altered in colorectal
Onco-
gene 1989 ; 4 : 1153-7 . 7 . Cance WG, Craven RJ, Weiner TM, Liu ET. Novel protein kinases in human breast cancer (to be submitted) .
12 . Baldwin GS, Zhang QX . Measurement of gastrin and transforming growth factor alpha messenger RNA levels in colonic carcinoma cell lines by quantitative polymerase chain reaction . 2261-7 .
Cancer Res 1992 ; 52 :
