| Nucleic Acids Research | Pages |
In vivo analysis of cancerous gene expression by RNA-polymerase chain reaction
Introduction
Materials And Methods
Sequences of oligonucleotides
In situ hybridization and cell preparations
RNA-PCR
Northern blotting
Results
High yield of amplified mRNAs by RNA-PCR
Comparison of mRNA products prepared by chemical extraction, chromatography and RNA-PCR
In vivo analysis of cancerous gene expression in human prostatic epithelium
Comparison of data consistency between in vitro and in vivo analysis
Discussion
References
In vivo analysis of cancerous gene expression by RNA-polymerase chain reaction
Received August 13, 1999; Revised and Accepted October 12, 1999
ABSTRACT An easy and routine procedure to amplify messenger RNA (mRNA) libraries from a few tissue cells can provide molecular gene expression profiles at high resolution. A novel PCR-like method, the RNA-PCR, was developed to generate high quality and quantity mRNAs from as few as 20 cells (2 pg mRNAs). The principle relies upon the cycling steps of promoter-linked double-stranded cDNA synthesis and promoter-driven transcription to amplify mRNAs up to 250-fold/cycle with good representation of high and low copy mRNAs. The amplified mRNA libraries were shown to possess high fidelity, purity, specificity and reproducibility for in vivo analyses of cancerous gene expression in human prostate cancers.
INTRODUCTION
The ability to amplify a complete mRNA repertoire from a few homogeneous cells (single cells) isolated under microscopic examination of pathological tissues enables the comparison of gene expression patterns between normal and disordered cells at high resolution. Single cell cDNA libraries can be applied widely to a variety of biological problems, including the molecular profiling of cancer by mapping stage-specific molecular signatures, the developmental staging and control of stem cells, single cell molecular and pathological phenotyping to aid in identifying targets for gene therapy and pharmaceutical design. With the rapid development of gene array techniques for high throughput analyses, another component that urgently needed to be developed was the ability to provide mRNAs from particular cell types.
Currently, techniques based on in-cell PCR are applied to generate complementary DNA (cDNA) libraries from relatively large numbers of cells (103-105) and only offer partial sequence information with limited fidelity (1,2). This is useful when a homogeneous cell population such as in vitro cultured cell lines is used. However, molecular analyses of in vivo specimens, including normal and pathological tissues, have been limited by the resolution of biochemical methods and the difficulty in obtaining homogeneous cell populations. Methods such as laser-capture microdissection (3,4) have been developed to isolate small numbers of cells from tissue sections; however, the analysis is still limited to the detection of abundantly expressed genes. Another advance is the use of antisense RNA (aRNA) amplification (5), which has led to the identification of some abundant markers for tuberous sclerosis (6) and apoptosis (7). However, this procedure only generates 50-75% of the total mRNAs from a single neuron (8) and rare mRNAs are inaccessible (7). Also, aRNA products are generated by random priming and represent only partial complementary sequences of mRNAs. Therefore, aRNA technology cannot be applied to functional cloning assays such as in vitro transcription/translation, transfection or full-length gene cloning.
To this end, a novel PCR-like reaction performed on mRNAs was devised, named the RNA-polymerase chain reaction (RNA-PCR). This technique provides a highly efficient amplification (~250-fold/cycle) of the whole mRNA repertoire. The elevated thermocycling temperature prevents rapid degradation of short-lived mRNAs (9) and further reduces the secondary structure of mRNAs to increase the accessibility of enzyme interactions and the production of more complete full-length mRNAs. Moreover, the application of novel thermostable enzymes was tested successfully in this procedure, including Tth-like DNA polymerases with reverse transcriptase activity and thermostable RNA polymerases. Thus the use of proof-reading RNA polymerases for amplification not only provides higher fidelity but also eliminates preferential amplification of abundant mRNA species. Additionally, we devised rapid and simple cell fixation and permeabilization steps (2) to inhibit any alterations in gene expression during specimen handling or genomic contamination.
The procedure is as follows (Fig. 1): (1) prevention of mRNA degradation; (2) first reverse transcription; (3) a tailing reaction to add 5[prime]-poly(dT) and 3[prime]-poly(dG) to the first strand cDNAs; (4) denaturation and then cDNA double-stranding by the extension of an oligo(dC)-promoter primer complementary to the 3[prime]-poly(dG) tail; (5) promoter-driven transcription to amplify mRNAs up to 2000-fold in one cycle; (6) repeating steps 2, 4 and 5 (without 3) to achieve the desired mRNA amplification.
Figure 1. RNA-PCR illustrated by schematic representations. The pre-cycling procedure comprises steps (1)-(4) to form promoter-linked double-stranded cDNAs, while the cycling procedure consists of steps (5) and (6) to generate mRNAs from the above cDNAs (see also text). To reiterate another round of amplification, the amplified mRNA products can be used as templates following the same cycling steps to generate more amplified mRNAs, and so on.
MATERIALS AND METHODS
Sequences of oligonucleotides
Primers used in RNA-PCR were as follows: a poly(dT)24 primer (5[prime]-dTTTTTTTTTTTTTTTTTTTTTTTT-3[prime]) and an oligo(dC)10N-promoter primer mixture comprising equal amounts of oligo(dC)10G-T7 primer (5[prime]-dCCAGTGAATTGTAATAC-GACTCACTATAGGGAAC10G-3[prime]), oligo(dC)10A-T7 primer (5[prime]-dCCAGTGAATTGTAATACGACTCACTATAGGGAAC10A-3[prime]) and oligo(dC)10T-T7 primer (5[prime]-dCCAGTGAATTGTAATA-CGACTCACTATAGGGAAC10T-3[prime]). The poly(dT)24 primer was used to reverse transcribe mRNAs into first-strand cDNAs, while the oligo(dC)10N-promoter primers functioned as a forward primer for second-strand cDNA extension from the poly(dG) end of the first-strand cDNAs and therefore RNA promoter incorporation. All oligonucleotides were synthetic and purified by high performance liquid chromatography (HPLC).
In situ hybridization and cell preparations
Fresh formaldehyde prefixed paraffin-embedded sections were dewaxed, dehydrated and refixed with 4% PFA, and then permeabilized with proteinase K (10 µg/ml; Roche) after rinsing with 1× PBS. In situ hybridization was achieved with a denatured hybridization mixture within a 200 µl coverslip chamber, containing 40% formamide, 5× SSC, 1× Denhardt's reagent, 50 µg/ml salmon testis DNA, 100 µg/ml tRNA, 120 pmol/ml poly(dT)24 primer, 10 pmol/ml biotin-labeled activin antisense probe (~700 bases in size) and tissue. After 10 h incubation at 65°C, sections were washed once with 5× SSC at 25°C for 1 h and once with 0.5× SSC, 20% formamide at 60°C for 30 min to remove unbound probes. A pre-heating step (68°C, 3 min) immersing the sections in a mild denaturing solution (25 mM Tris-HCl, pH 7.0, 1 mM EDTA, 20% formamide, 5% DMSO and 2 mM ascorbic acid) was performed to minimize secondary structures (including crosslinks) and to reduce the background. After the temperature was lowered to 45°C, 2,5-diaziridinyl-1,4-benzoquinone (200 µM; Sigma Chemical Co., St Louis, MO) was added to each incubation for a further 30 min. Finally, 0.1× SSC, 20% formamide was applied at 60°C for 30 min to clean sections for chromogenic detection with straptavidin-alkaline phosphatase and Fast Red staining (Roche Biochemicals, Indianapolis, IN). Positive and negative results were observed and recorded under a microscope. RNase-free enzymes and DEPC-treated materials were required throughout the procedure.
Prostate cancer cells (20-150 cells) from in situ sections of patient tissues were isolated with a micromanipulator and directly used in RNA-PCR, while cultured LNCaP cells were preserved in 500 µl of ice-cold 10% formaldehyde in suspension buffer (0.15 M NaCl pH 7.0, 1 mM EDTA) for the following fixation and permeabilization procedure (2). After 1 h incubation with occasional agitation, fixed LNCaP cells were collected with microcon-50 filters (Amicon, Beverly, MA) and washed with 350 µl of ice-cold PBS with vigorous pipetting. The collection and wash were repeated at least once. The fixed cells were then permeabilized in 500 µl of 0.5% NP-40 for 1 h with frequent agitation. After that, three collections and washes were given to cells as before but using 350 µl of ice-cold PBS containing 0.1 M glycine instead. The cells were finally mixed with 0.1 µM poly(dT)24 primer and resuspended in the same buffer with vigorous pipetting to evenly distribute them into small aliquots (~20 cells in 10 µl) for RNA-PCR. They could be stored at -80°C for up to 2 weeks.
RNA-PCR
For amplification of intracellular mRNAs, more than 20 fixed cells were preheated at 94°C for 5 min and applied to a reverse transcription (RT) reaction (50 µl) on ice, comprising 10 µl of 5× RT&T buffer [100 mM Tris-HCl, pH 8.5 at 25°C, 600 mM KCl, 300 mM (NH4)2SO4, 25 mM MgCl2, 5 M betaine, 35 mM dithiothreitol, 10 mM spermidine and 25% dimethylsulphoxide (DMSO)], 1 µM poly(dT)24 primer, dNTPs (1 mM each dATP, dGTP, dCTP and dTTP) and RNase inhibitors (10 U). After 6 U Caxboxydothermus hydrogenoformans (C.therm.) polymerase (Roche) was added, the reaction was incubated at 52°C for 3 min and shifted to 65°C for another 30 min. The first-strand cDNAs so obtained were collected with a microcon-50 microconcentrater filter, washed once with 1× PBS and suspended in a tailing reaction (50 µl), comprising 10 µl of 5× tailing buffer (250 mM KCl, 100 mM Tris-HCl, 4 mM CoCl2, 10 mM MgCl2, pH 8.3 at 20°C) and 0.5 mM dGTP. After 75 U terminal transferase (Roche) was added, the reaction was incubated at 37°C for 15 min, stopped by denaturation at 94°C for 2 min and instantly mixed with 1 µM oligo(dC)10N-promoter primer mixture. After briefly centrifuging, 3.5 U Taq DNA polymerase (Roche) and 1 mM each dNTPs were added to form promoter-linked double-stranded cDNAs at 52°C for 3 min and then 72°C for 7 min. The cells were broken by adding 1 vol of 2% Nonidet P-40 (NP-40; Sigma Chemical Co.) for 10 min, and then the double-stranded cDNAs were washed and recollected with a microcon-50 in autoclaved ddH2O. This completed the pre-cycling steps for the following cycling amplification.
A transcription reaction (50 µl) was prepared, containing 10 µl of 5× RT&T buffer, rNTPs (1 mM each ATP, GTP, CTP and UTP), RNA inhibitors (10 U), T7 RNA polymerase (200 U; Roche) and the double-stranded cDNAs. After 2 h incubation at 37°C, the cDNA transcripts were isolated with a microcon-50 filter in 20 µl of DEPC-treated TE buffer (pH 7.0) and used directly for the next round of RNA-PCR without the tailing reaction, containing 10 µl of 5× RT&T buffer, 1 µM poly(dT)24 primer, 1 µM oligo(dC)10N-promoter primers, dNTPs (1 mM each), rNTPs (1 mM each), C.therm. polymerase, Taq DNA polymerase and the transcription products (20 pg). T7 RNA polymerase was renewed in every transcription step due to prior denaturation. The quality of mRNA products (20 µg) after three rounds of amplification was assessed on a 1% formaldehyde-agarose gel.
Northern blotting
mRNAs were fractionated on 1% formaldehyde-containing agarose gels and transferred to nylon membranes (Schleicher & Schuell, Keene, NH). Probes were labeled with the Prime-It II kit (Stratagene, La Jolla, CA) by random primer extension in the presence of [32P]dATP (>3000 Ci/mM; Amersham International, Arlington Heights, IL). Hybridization was carried out in the mixture of 50% freshly deionized formamide (pH 7.0), 5× Denhardt's solution, 0.5% SDS, 4× SSPE and 250 µg/ml denatured salmon sperm DNA (18 h, 42°C). Membranes were washed twice in 2× SSC, 0.1% SDS (15 min, 25°C), followed by once each in 0.2× SSC, 0.1% SDS (15 min, 25°C) and 0.2× SSC, 0.1% SDS (30 min, 65°C) before autoradiography.
RESULTS
In the current procedure, we used a poly(dT)24 primer to generate the first-strand cDNAs and then used another oligo(dC)-promoter primer to generate the second-strand cDNAs. Both strands together formed the promoter-linked double-stranded cDNAs from the original mRNAs. The oligo(dC)-promoter primer was an equal mixture of oligo(dC)10N sequences (N = dG, dA or dT) coupled to an RNA promoter for in vitro transcription along the double-stranded cDNA templates. Because the promoter region was incorporated in the 5[prime]-end of the second-strand cDNAs which had the same sequence and composition as the original mRNAs, the transcription products were all in the form of mRNAs, not aRNAs. These amplified mRNAs not only shared the same properties but also had the full integrity of their original mRNAs, depending on the quality of the first promoter-linked double-stranded cDNAs. It has been noticed that steps (1)-(4) are critical for the quality of the resulting mRNAs.
High yield of amplified mRNAs by RNA-PCR
mRNAs generated with this procedure are well amplified and well represented. According to the high efficiency of transcriptional amplification (up to 2000-fold/cycle) (8), three rounds of RNA-PCR are theoretically equivalent to 33 cycles of PCR amplification (2-fold/cycle). The advantages of this amplification cycling are as follows: (i) single copy mRNAs can be increased up to 2000-fold in one round of amplification with proof-reading activity; (ii) the mRNA amplification is linear and does not result in preferential amplification; (iii) the mRNA products are potentially of full length and can be directly used to generate a complete cDNA library. Theoretically, we can multiply a single copy of mRNA more than 1 billion-fold. However, the high amplification rate also results in rapid consumption of the starting materials and accumulation of products in the reactions, resulting in lower efficiency following multiple cycles. Another issue to improve is the prevention of hydrolysis of RNA sequences at elevated temperature. Still, in our experience, we have acquired 30 µg of amplified mRNAs in one 50 µl reaction after three rounds of RNA-PCR amplification from 20 cells. This represents a 15 million-fold increase based upon a comparison between the amount of synthesized mRNAs and that of theoretically presumed mRNAs within a cell (0.1 pg).
Comparison of mRNA products prepared by chemical extraction, chromatography and RNA-PCR
mRNAs generated by RNA-PCR from 20 cells were compared with those by traditional RNA extraction from 105 cells (Fig. 2). Three different RNA profiles from LNCaP cells, a human prostatic cancer cell line, were isolated either by phenol/chloroform extraction (TRIzol reagent; Gibco BRL), oligo(dT) column chromatography (Qiagen) or RNA-PCR, and fractionated on a denaturing agarose gel. RNAs prepared by each of the three methods ranged in size from 300 bases to >7.4 kb. In general, good integrity of total cellular mRNAs should appear as a smear between approximately 300 bases and 8 kb. The uniform smearing pattern of the RNA-PCR products showed no contamination with either rRNAs or genomic DNAs, indicating that RNA-PCR had acquired better quality mRNAs than the other methods. Based on our results from northern blotting, the correctly sized gene transcripts of RB, [beta]-actin and GAPDH were detected (Fig. 2B), demonstrating potential full-length specificity up to 5 kb.
Figure 2. Comparison of RNA prepared by RNA-PCR and traditional methods. (A) Comparison between RNA repertoires prepared by phenol/chloroform extraction (lane 2), on an oligo(dT) chromatographic column (lane 3) and by RNA-PCR (lane 4) fractionated on a 1% formaldehyde-agarose gel, all ranged from 300 bases to >7.4 kb based on RNA markers. A uniform smearing pattern of the RNA-PCR products indicates better mRNA quality than produced by the other methods, containing no contamination from non-poly(A)+ nucleotide species. (B) Correctly sized gene transcripts of RB (4.9 kb), [beta]-actin (2.2 kb) and GAPDH (1.7 kb) were also detected by northern blot hybridization, demonstrating the high specificity of using RNA-PCR products in gene expression analyses.
In vivo analysis of cancerous gene expression in human prostatic epithelium
One of the most powerful applications of this procedure is in the analysis of pathological sections. In our model, pathological sections were stained by in situ hybridization (Fig. 3A) with probes to activin to identify activin-positive and activin-negative cells in vivo. Following microdissection, mRNAs were amplified from the activin-positive and activin-negative epithelial prostatic cancer cells, respectively. The amplified mRNAs from malignant, intermediate and neoplastic prostate cancers were generated using RNA-PCR. As seen in Figure 3B, the amplified mRNAs were of similar quality to those from the cell lines (compare with Fig. 2).
Figure 3. RNA-PCR using cells microdissected from specific regions of pathological sections. (A) Identification of activin-positive (red) and activin-negative (blue) prostate cancer epithelial cells by in situ hybridization. Dotted circles demarcate the regions for cell isolation with a micromanipulator under a microscope. For clarity, the background of this section was stained with hematoxylin. This staining was not used in the sections used for cell isolation. (B) Denaturing agarose gel electrophoresis (1%) of RNA-PCR products from the above isolated few cells. Three stages of prostate cancers were identified under a microscope and labeled as malignant (M), intermediate (I) and prostatic intra-epithelial neoplasia (PIN). Lane 1, RNA markers; lanes 2, 4 and 6, mRNAs from activin-negative prostatic cancer cells; lanes 3, 5 and 7, mRNAs from activin-positive prostatic cancer cells; lane 8, negative control of RNA-PCR without cells.
We further verified the results of this procedure by investigating the behavior of known genes in prostate cancers (10,11). For LNCaP cells, we observed the time course of alterations in p53 and p16 expression by northern blotting at different times after activin treatment, consistent with previous reports demonstrating by traditional chromatography methods that both genes were slightly up-regulated after activin treatment (Fig. 4A, left). For tissue sections, a similar up-regulation pattern was also seen in the patients with intermediate and neoplastic prostate cancers, showing a good correlation (Fig. 4, middle). However, the expression of p53 was identical in activin-positive and activin-negative cells derived from the malignant cancer patient, suggesting a possible loss of apoptotic regulation. This presumption is likely related to another observation that the more progressive the cancer, the less activin-positive cells were detected by in situ hybridization.
Figure 4. Northern blot analyses of gene expression changes. (A) Comparison between mRNAs made by oligo(dT) column chromatography and RNA-PCR. In vitro mRNAs (left) were generated by chromatography from 106 LNCaP cells at different times after activin treatment (0, 12, 24, 48, 68 and 120 h). In vivo mRNAs were amplified by RNA-PCR from 20 isolated cancer cells with (Ac+) or without (Ac-) expression of activin. p53 and [beta]-actin templates (Ambion, Austin, TX) and synthetic p16 were used as probes in (A). (B) Comparison between in vitro and in vivo mRNAs made by RNA-PCR from 20 cells. The blots of in vitro mRNAs (right) displayed differential alterations in gene expression in LNCaP cells before and after activin treatment (120 h), while the blots of in vivo mRNAs (middle) showed the actual differential expression in patient tissue cells. Probes (~500-750 bases in length) for (B) were isolated from RNA-PCR-derived cDNA libraries prepared from LNCaP cells and amplified by PCR with sequence-specific primers. The sequences of the probes were confirmed by automated sequencing. All detected gene transcripts matched their original mRNA sizes, indicating that they were of good integrity and most likely full length.
Comparison of data consistency between in vitro and in vivo analysis
mRNAs generated from prostate cancer sections were further compared with RNA-PCR-derived mRNAs prepared from LNCaP cells (Fig. 4B). Both sets of data showed consistent levels of p53 at the neoplastic stage, apoptosin at the malignant and intermediate stages, and p16 and apoptostatin at all three stages. Some inconsistency was observed; the expression of p21 was weak in vivo but it was clearly expressed in vitro. This may result from individual and stage-related variations, demonstrating the need to maintain caution in extrapolating cell culture data to tissue sections. Now, with this new, easy and reliable procedure, we and the scientific community can re-evaluate cancer molecular phenotypes at the single cell level without artifacts produced in culture.
DISCUSSION
Based on the comparisons of in vivo to in vitro results, the RNA-PCR-derived mRNA libraries truly provide a high resolution profiling for cancerous gene expression. A good correlation was well maintained in the comparisons of both methods and results. The gene expression patterns of p53 and p16 from 1 million LNCaP cells determined by chromatography (Fig. 4, left) resemble those from 20 LNCaP cells determined by RNA-PCR (Fig. 4, right), further indicating a potential replacement for traditional chromatography methods. Since the RNA-PCR-derived mRNAs have better quality, quantity and specificity (up to 5 kb), sample preparation for a variety of genetic research studies can be produced and reproduced by this method without deviations due to cell culture or tissue heterogeneity. Following the high demand for precise diagnosis of cancers, RNA-PCR is expected to reach this goal by offering a clear look into the molecular basis of stage-specific gene expression in specific cell populations.
Like other promoter-driven nucleic acid sequence amplification methods (12,13), RNA-PCR is useful for cloning a specific nucleic acid sequence. When performed with a specific primer complementary to the 3[prime]-margin of the desired mRNA sequence, mRNA or cDNA fragments of an appropriate size can be generated by RNA-PCR for further genetic analysis. The design of these sequence-specific primers is based on the same principles used for PCR. Since reverse transcriptase extension of the specific primers forms mRNA-cDNA hybrids and in vitro transcription through the promoter primers forms mRNAs, we can acquire the desired single-stranded nucleotide probes by adding either DNase or RNase to digest the unwanted half of the amplified products.
We now routinely use RNA-PCR to generate high purity nucleotide probes because it alternates between synthesizing mRNAs and cDNAs, depending on the stopping point chosen during the amplification cycle. The labeling of cDNAs can be easily accomplished by incorporating labeled nucleotide analogs during reverse transcription, whereas the labeling of mRNAs is completed during transcription. Typically, these probes can be useful in a variety of applications, such as hybridization blotting, antisense knock-out transfection, in situ detection and genetic cloning.
This method can also be used to generate full-length sequences easily. For example, full-length cDNA products (<5 kb) can be cloned into a competent vector for functional assays using transfection or viral transduction. For in vitro transcription, the linear cDNA products usually give higher yields and do not require endonuclease restriction, which may cleave internal sites of inserted nucleotide sequences. The amplified full-length mRNAs also provide transcripts for protein production in vitro. Many commercial in vitro translation systems provide a cap nucleotide which can be added to the 5[prime]-end of the amplified mRNAs during the transcription step of RNA-PCR. Unlike aRNAs, the capped mRNAs are useful in protein synthesis with labeling and may help the detection of protein activity if the proteins fold correctly.
Many applications and technological combinations can be facilitated by applying RNA-PCR. In conjunction with gene chip analysis and laser capture microscopy, RNA-PCR will be an indispensable component for a new level of molecular analyses in the future. The possibilities are unlimited.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 323 442 1982; Fax: +1 323 442 3158; Email: shilungl{at}hsc.usc.edu
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S.-L. Lin, D. Chang, and S.-Y. Ying Hyaluronan stimulates transformation of androgen-independent prostate cancer Carcinogenesis, February 1, 2007; 28(2): 310 - 320. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. C. Carter, S. Hutchison, F. L. Henriquez, D. Legare, M. Ouellette, C. W. Roberts, and A. B. Mullen Resistance of Leishmania donovani to Sodium Stibogluconate Is Related to the Expression of Host and Parasite {gamma}-Glutamylcysteine Synthetase Antimicrob. Agents Chemother., January 1, 2006; 50(1): 88 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Toledo-Rodriguez, B. Blumenfeld, C. Wu, J. Luo, B. Attali, P. Goodman, and H. Markram Correlation Maps Allow Neuronal Electrical Properties to be Predicted from Single-cell Gene Expression Profiles in Rat Neocortex Cereb Cortex, December 1, 2004; 14(12): 1310 - 1327. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y.-G. Chen, H. M. Lui, S.-L. Lin, J. M. Lee, and S.-Y. Ying Regulation of Cell Proliferation, Apoptosis, and Carcinogenesis by Activin Experimental Biology and Medicine, February 1, 2002; 227(2): 75 - 87. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||