Clinical Cancer Research Targets Advances in Breast Cancer
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kumar, S. V.
Right arrow Articles by Spivack, S. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kumar, S. V.
Right arrow Articles by Spivack, S. D.
Clinical Cancer Research Vol. 12, 5033-5039, September 1, 2006
© 2006 American Association for Cancer Research

Human Cancer Biology

Validity of Messenger RNA Expression Analyses of Human Saliva

Shalini V. Kumar1, Gregory J. Hurteau2 and Simon D. Spivack1,3,4

Authors' Affiliations: 1 Laboratory of Human Toxicology and Molecular Epidemiology, Wadsworth Center, New York State Department of Health, Albany, New York, 2 Ordway Research Institute, 3 Pulmonary and Critical Care Medicine, Albany Medical College, Albany, New York, and 4 Environmental Health Sciences, School of Public Health, State University of New York, Rensselaer, New York

Requests for reprints: Simon D. Spivack, Laboratory of Human Toxicology and Molecular Epidemiology, Wadsworth Center, New York State Department of Health, P.O. Box 509, Empire State Plaza, Albany, NY 12201-0509. Phone: 518-473-0782; Fax: 518-486-1505; E-mail: spivack{at}wadsworth.org.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
Purpose: The origins of expression microarray and reverse transcription-PCR (RT-PCR) signals in human saliva were evaluated.

Experimental Design: The "RNA" extracts from human saliva samples were treated with vehicle, DNase, or RNase. Two-step amplification and hybridization to Affymetrix 133A cDNA microarrays were then done. Confirmatory RT-PCR experiments used conventionally designed PCR primer pairs for the reference housekeeper transcripts encoding 36B4, ß-actin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA sequences, which are known to be homologous to genomic DNA pseudogene sequences. Negative controls included the omission of reverse transcriptase ("no-RT") to detect any DNA-derived signal. Finally, an RNA-specific RT-PCR strategy eliminated confounding signals from contaminating genomic DNA.

Results: Microarray experiments revealed that untreated, DNase-treated, and RNase-treated "RNA" extracts from saliva all yielded negligible overall signals. Specific microarray signals for 36B4, ß-actin, and GAPDH were low, and were unaffected by RNase. Real-time quantitative RT-PCR reactions using conventional, non–RNA-specific primers on saliva samples yielded PCR products for 36B4, ß-actin, and GAPDH; DNase-treated saliva samples did not yield a PCR product, and the "no-RT" and "+RT" conditions yielded similar amounts of PCR product. The RNA-specific RT-PCR strategy, across all conditions, yielded no PCR product from saliva.

Conclusions: The combination of (a) a minimal microarray signal, which was unaffected by RNase treatment, (b) the presence of a conventional RT-PCR housekeeper product in both RNase-treated and no-RT saliva samples, (c) the absence of a conventional RT-PCR housekeeper product in DNase-treated conditions, and (d) the absence of a RNA-specific RT-PCR product shows that any microarray or RT-PCR signal in the saliva must arise from genomic DNA, not RNA. Thus, saliva extracts do not support mRNA expression studies.

Large-scale human cancer studies for the assessment of functional genomics or gene-environment interactions require a reliable source of RNA that can be accessed with minimal invasiveness, and is compatible with downstream, high-throughput applications. Options that have emerged include brush-exfoliated buccal cells (1), urine (2), and saliva (3, 4). Given the potential translational importance of mRNA expression biomarkers in such noninvasively acquired biospecimens, it is imperative that the RNA specificity of gene expression studies from these specimens be confirmed.

Many human transcripts have homologous sequences in nontranscribed genomic DNA sequences, remote from the structurally active coding genes, which are termed pseudogenes. Common examples include ß-actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), acidic ribosomal phosphoprotein P0 (36b4), hypoxanthine phosphoribosyl transferase, glutathione S-transferases M1 and P1, and glutathione peroxidase. These pseudogenes are intronless, often have poly(A) tails, and have extremely high homology with cDNA sequences.

Pseudogenes are very commonly the source of primer-confounded, false-positive reverse transcription-PCR (RT-PCR) results from genomic DNA–contaminated total RNA extracts (5). Although the use of DNase treatment of the "RNA" extract is a simple approach to the elimination of genomic DNA contamination, it has been shown to be both incompletely effective and to compromise the RNA-derived RT-PCR signal yield, particularly in trace mRNA template situations (1, 5).5 To circumvent this problem, we have developed an RNA-specific RT-PCR strategy that amplifies RNA specifically in a mixed nucleic acid background. The sensitivity of our method is equivalent to that of conventional RT-PCR, and is compatible with current real-time quantitative RT-PCR methods (1, 5–7).

In microarray transcriptome studies, the fluorescent signal may be RNA-specific or RNA-nonspecific, depending on the particular steps used for RNA isolation, template amplification, and labeling. Because published microarray data generally do not typically address RNA specificity, there is a presumption that T7-tagged odT-based reverse transcription and subsequent cRNA amplification by in vitro transcription approaches yield RNA-specific signals (Affymetrix Expression Analysis Technical Manual, p. 2.1.5, available from: http://www.affymetrix.com/; Arcturus RiboAmp manual, p. 11). The processes also entail an intermediate step of random primed cDNA {Rightarrow} double-stranded cDNA. These amplification approaches have some potential to produce false-positives due to carry-over double-stranded genomic DNA pseudogene sequence contaminants. As noted, these DNA sequences are incompletely extinguished by DNase treatment of the original RNA extract.

In addition, homologous genomic DNA or RNA sequences from mouth microorganisms (bacteria and viruses) could potentially confound gene expression studies that have not been designed to specifically exclude this possibility.

In this report, we evaluate the ability of saliva to reliably support gene expression studies by analyzing whether the microarray or RT-PCR signal previously reported by other laboratories does indeed arise from the RNA template, or alternatively, whether it arises from genomic DNA contaminants. Experiments using reciprocal RNase and DNase treatments of the saliva extract, omission of the reverse transcription step, and employment of a previously developed RNA-specific RT-PCR strategy definitively show the origin of both the microarray signal and the RT-PCR signal.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Human saliva collection
Volunteer subjects were recruited anonymously, as per Institutional Review Board–approved protocols. Individual saliva samples for each of the five healthy volunteers were processed separately. Each spontaneous, unstimulated 3 to 5 mL saliva specimen was expectorated into a tube containing 10 mL of RNAlater (Ambion, Inc., Austin, TX), as per Li and coauthors (4), and placed immediately on ice at 4°C.

Overall strategy
In separate saliva specimens from each individual processed in parallel, mRNA expression was qualitatively and quantitatively evaluated by both cDNA microarray (human genome U133A array; Affymetrix, Inc., Santa Clara, CA) and real-time RT-PCR. Each RNA extract was divided into three separate treatments: untreated, DNase-treated, and RNase-treated representing the hypothesized RNA/DNA mixture, pure RNA if present in saliva, and pure genomic DNA contaminant if present in saliva, respectively. Commercially available human lung total RNA (Clontech, Mountain View, CA) was used as positive control RNA.

For microarray experiments, the amounts of RNA (in the microgram range) required for conventional amplification and labeling prompted the use of a GeneChip Two Cycle Target Labeling kit (Affymetrix), which gives sufficient cRNA yield from a total RNA quantity as low as 10 to 100 ng. Two rounds of cDNA synthesis prior to in vitro transcription and labeling allow higher amplification than is achievable with other protocols.

The RT-PCR experiments used standard-design primers as well as mRNA-specific primer pairs (Table 1 ). The expression of three reference housekeeper transcripts, 36B4, ß-actin, and GAPDH, was measured under all three RNA treatment conditions using two types of primer sets: (a) a standard-design primer set (in which at least one oligomer spans an exon/exon splice site), that can coamplify both RNA template and the corresponding contaminating genomic DNA pseudogenes for these transcripts, and (b) an RNA-specific RT-coupled PCR primer set previously shown to be RNA-specific (5). As further controls, reverse transcription reactions either included the reverse transcriptase enzyme (RT+), or omitted the enzyme ("no-RT"); the latter therefore amplified only contaminating genomic DNA.


View this table:
[in this window]
[in a new window]
  		Table 1. Universal (UR) and standard-design (std) PCR primer sequences used in the current study

 
RNA isolation from saliva
Individual saliva samples for each of the five volunteers were immediately centrifuged at 2,600 x g for 15 minutes at 4°C. For saliva supernatant, total RNA was isolated from 500 µL of sample using the standard Qiagen RNeasy kit (Qiagen, Valencia, CA) with minor modifications to the manufacturer's protocol. Briefly, RLT-ßME buffer (350 µL) was added to the saliva sample (500 µL) and vortexed for 10 seconds. Although the results for the supernatant are displayed, identical extraction procedures were employed for the saliva "pellet" and whole saliva samples. The lysate was homogenized by passage (10 times) through a 22-gauge needle. Poly(C) RNA carrier (1 µg) was added to this lysate, followed by 350 µL of 70% ethanol. The sample was vortexed again for 10 seconds and was held at room temperature for 1 minute. The lysate was then transferred to a column provided in the kit and centrifuged at 10,000 x g for 30 seconds. RW1 buffer (700 µL) was added to the column after the flow-through from the previous spin had been discarded, followed by centrifugation at 10,000 x g for 30 seconds. RPE buffer (500 µL) was added after the collection tube had been changed, and the sample was centrifuged at 10,000 x g for 10 seconds before the final elution into RNase-free water. Each isolated RNA extract was divided into three equal volumes maintained on ice, for different treatments. One volume was left untreated, the second one was DNase-treated, and the third was treated with RNase, as described below.

DNase treatment
In an attempt to remove any contaminating genomic DNA, we treated RNA with 10 units of RNase-free DNase (Ambion) in a volume of 50 µL, and then incubated it at 37°C for 30 minutes. This was followed by the addition of 0.1 volume of DNase Inactivation Reagent, and incubation at room temperature for 2 minutes. The DNase Inactivation Reagent was subsequently removed by centrifugation at 10,000 x g for 1 minute, followed by transfer of the RNA solution into a fresh tube. Given the published data demonstrating incomplete DNase efficacy (5), we did this DNase treatment twice for samples intended for both microarray analysis and RT-PCR.

RNase treatment
To further determine the relative contributions of DNA and RNA to the PCR product, we treated the isolated RNA solution with 1 µL of RNase A (4 mg/mL; Gentra Systems, Inc., Minneapolis, MN) and incubated it at 37°C for 30 minutes. RNase was removed from the RNA solution by use of a RNA cleanup protocol described by the manufacturer (RNeasy Mini Handbook, Qiagen).

Microarray analysis
RNA isolated from saliva, as well as the human lung total RNA standard purchased from Clontech, was either left untreated or treated with DNase or RNase, with cleanup as described above. An Agilent Capillary Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA) was used to assess the RNA quality of each treatment condition in both RNA types. One cycle of double-stranded cDNA synthesis was done using the GeneChip Two Cycle Target Labeling kit (Affymetrix) for all of the treated and untreated RNA samples. This was followed by in vitro transcription amplification of antisense RNA (cRNA) using the MEGA script T7 kit (Ambion). The samples were then subjected to a second cycle of cDNA synthesis, followed by biotin labeling of antisense cRNA. Subsequently, the labeled RNA samples were hybridized, washed, stained, and then scanned using an Affymetrix GCS 3000 scanner, housed in the Genomics Microarray Core facility at the Wadsworth Center. GeneSpring GX software (Agilent) was used for data normalization and initial data analysis.

Conventional RT-PCR
Reverse transcription. The Superscript II reverse transcriptase RNase H– (Invitrogen Life Technologies, Inc., Carlsbad, CA) protocol, with minor modifications, was followed, to reverse transcribe both treated and untreated RNA. All steps were done in a thermal cycler [Perkin-Elmer 9700 or real-time LightCycler (Roche, Indianapolis, IN), as appropriate]. cDNA was prepared in a 20 µL reaction containing 1 µL of oligo(dT) (0.5 µg/µL), RNA (up to 0.3 µg), and RNase-free/DNase-free water to make up the final volume to 11 µL. The RNA was denatured at 65°C for 5 minutes and then at 4°C for 5 minutes, before the addition of the following: 4 µL of 5x Superscript II first strand buffer (Invitrogen Life Technologies), 2 µL of 0.1 MDTT, and 2 µL of deoxynucleoside triphosphate mix (10 mmol/L). Incubation at 42°C for 2 minutes was followed by the addition of 1 µL of Superscript II RT to each tube, and each tube was individually mixed gently by pipetting up and down. In the no-RT reactions, 1 µL of water was added instead of the reverse transcriptase. This was followed by incubation at 42°C for 50 minutes, and then at 70°C for 15 minutes.

Quantitative real-time cDNA-PCR protocol. Real-time quantitative RT-PCR was done using SYBR Green chemistry (Molecular Probes, Invitrogen). The following were added in a glass capillary (Roche) for a final reaction volume of 25 µL: 5.0 µL of the 5x one-step RT-PCR buffer (Qiagen), to yield 2.5 mmol/L of MgCl2; 15.5 µL of RNase-free/DNase-free H2O; a total of 1.6 µL of deoxynucleoside triphosphate mix (10 mmol/L each triphosphate nucleotide); 1.25 µL of SYBR Green dye (1:10,000); 0.4 µL of Platinum Taq (Invitrogen Life Technologies) DNA polymerase; 0.25 µL of 50 µmol/L from each oligo of the primer pair; and 1.0 µL template cDNA from the reverse transcriptase reaction. The thermocycling reaction conditions included one denaturing cycle (20°C/s up-ramp) at 95°C for 30 seconds; PCR for 50 cycles consisting of up-ramp (20°C/s) to melt at 95°C for 10 seconds, down-ramp (3.5°C/s) to anneal at 59°C; and up-ramp (3.5°C/s) to anneal at 72°C for 30 seconds. The melting analysis for one cycle was as follows: up-ramp (20°C/s) to melt at 95°C for 10 seconds, down-ramp (20°C/s) to anneal at 58°C/s, and then slow up-ramp (0.1°C/s) for continuous acquisition to 95°C. For a run to be scored as positive, and for quantitative data to be entered, the PCR product had to display both a single peak with characteristic melt temperature on LightCycler melting analysis, and to be corroborated by agarose EtBR electrophoresis, displaying a single product at the appropriate size.

RNA-specific RT-PCR
Parallel experiments were done employing an RNA-specific RT-PCR technique, as previously described (1, 5–7). Briefly, the approach was to take conventional total RNA extracts, which invariably also contain genomic DNA contamination, and to reverse-transcribe the poly(A)-containing mRNA with a 5'-tagged and 3'-anchored oligo-dT reverse transcription primers (100 µmol/L; URT). The unique 5' tag sequence does not occur anywhere in the human genome, so that with the poly(T) region annealing to poly(A)-capped RNA, only RNA-derived cDNA is uniquely tagged, whereas genomic DNA is not. The subsequent PCR step uses a reverse (antisense) primer (UR) targeting this tag, and a transcript-specific forward (sense) primer. Therefore, only those DNA sequences derived from mRNA containing the tag (i.e., cDNA) are amplified. The method does not require DNase treatment of the original nucleic acid extract for RNA specificity.

Thus, cDNA synthesis for RT-PCR was done accordingly in two parallel sets of samples, using either a conventional oligo-dT primer, or alternately, the tagged URT primer as an RNA-specific control. Housekeeper reference transcripts (36b4, ß-actin, and GAPDH; GAPDH results not shown) were amplified using standard-design housekeeper cDNA primers when the cDNA template was prepared with oligo-dT primer, whereas UR (tag compatible) reverse primer pairs were used when cDNA was generated using the tagged-RT primer. Primer sets used in both the conventional and RNA-specific RT-PCR are listed in Table 1.


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
RNA quality
For the RNA (Clontech) standard, the electropherograms obtained from the Agilent Bioanalyzer 2100 gave clear and well-defined 28S and 18S RNA peaks, with high 28S/18S RNA ratio for the untreated and DNase-treated RNA standards. The RNase-treated Clontech RNA standard showed complete degradation of the RNA. For the saliva RNA extracts, the bioanalyzer did not detect meaningful 28S/18S bands (data not shown).

Microarray results
The overall HG U133A microarray patterns were used to initially screen for the presence or absence of RNA in both the RNA standard and the saliva extract samples (Fig. 1 ). Additionally, signals from the internal reference housekeeper genes [ribosomal phosphoprotein large subunit P0 (36b4), ß-actin, and GAPDH] in the untreated, DNase-treated, and RNase-treated fractions were used to assess the presence of DNA and RNA in the saliva samples. The Clontech human lung total RNA sample was used as a positive control (Fig. 2 ).


Figure 1
View larger version (52K):
[in this window]
[in a new window]
  		Fig. 1. Overall heat maps of the hybridization of human RNA standard (lung total RNA, Clontech, top) and human saliva (bottom) to same-lot Affymetrix U133A expression microarrays. Each sample was partitioned into untreated, DNase-treated, and RNase-treated conditions. Present (%), number of probes displaying a signal above the Affymetrix software–defined threshold. For the RNA standard, DNase had no effect, whereas RNase quenched the overall signal, implying that the signal derives exclusively from RNA, as expected of an RNA standard. The saliva signals under all conditions were very low, similar to the signal from the RNase-treated signal of the RNA standard, and were largely unaffected by RNase treatment, implying a lack of RNA in the samples.

 

Figure 2
View larger version (14K):
[in this window]
[in a new window]
  		Fig. 2. Graphical representation of the U133A microarray signals obtained for reference housekeeper transcripts 36B4, ß-actin, and GAPDH for both Clontech RNA standard and saliva samples, for untreated, DNase-treated, and RNase-treated conditions. Columns, mean of the RMA normalized values (GeneSpring GX software, Agilent Technologies). P values for the signal difference (ANOVA, Tukey post hoc; SigmaStat, Systat Software, Inc., Point Richmond, CA) are as indicated. NS, "nonsignificant" values. For ß-actin, the microarray signal difference between the saliva DNase-treated condition and the other comparable conditions (RNase-treated Clontech RNA standard, untreated saliva, and RNase-treated saliva) is of unclear significance.

 
For the known, pure total RNA (Clontech) standard, the untreated and DNase-treated conditions had 54.6% and 48.2% of probes displaying present signals, respectively, whereas the RNase-treated condition gave only a 5.9% present signal (Fig. 1). For saliva RNA extracts, in comparison, untreated saliva RNA gave only 7.4% present signal and the DNase-treated condition gave only about 5.3% present signal, both of which were close in value to the RNase-treated condition of the control RNA. The RNase condition gave similar (8.1%) present signals. The low overall signal intensity for all saliva specimens can be visualized in the respective heat maps (Fig. 1).

The microarray data were normalized using RMA normalization, a quantile normalization procedure that normalizes across all chips in a given set, in the GeneSpring GX software. The mean values obtained from the three probes (5', middle, 3' positions for ß-actin and GAPDH) or the five probes (for the multiple 36b4 probe sets) were plotted against the treatment conditions (Fig. 2). The RNA standard (Clontech) showed maximum signals in the untreated condition for all three housekeeping genes; there was no significant decrease in the DNase-treated condition, whereas the RNase-treated condition showed a significant decrease in signal intensity. The untreated saliva RNA showed much lower signal intensities than did the untreated Clontech standard RNA (Fig. 2). There was no significant difference between the RNase-treated condition of the Clontech standard RNA and the untreated saliva RNA. The difference between the untreated saliva extract and the RNase-treated saliva extract was not significant.

RT-PCR results
Quantitative RT-PCR was used to validate the microarray results. For the untreated saliva RNA extract (Figs. 3A and 4A ), standard-design PCR product was present in both the RT and no-RT conditions, implying at least a contribution to the PCR product by contaminating a genomic DNA template of identical size, i.e., processed pseudogene sequence. When DNase was applied to each of the saliva RNA extracts, all PCR product signals were extinguished. When RNase was applied to the saliva RNA extract, PCR product signal was still preserved across RT and no-RT conditions. This implies that any positive signal for the salivary extracts was genomic DNA–contaminant in origin. Results were consistent across all five volunteers, and for the supernatant, "whole saliva," and the centrifuged pellet fraction RNA extracts; results for the supernatant fraction are displayed (Figs. 3–5 ).


Figure 3
View larger version (36K):
[in this window]
[in a new window]
  		Fig. 3. Electrophoresis gel display (A and B) of real-time quantitative RT-PCR products, and real-time display (C) done on salivary RNA extracts (top, gel A) and positive RNA control (human lung commercial RNA standard; Clontech; bottom, gel B), under various conditions. A, using standard-design (std) 36B4 PCR primers, PCR products of untreated (Non), DNase-treated, and RNase-treated RNA extracts from saliva were compared using no-RT and RT conditions. MCF, positive control cDNA derived from a MCF-7 breast cancer cell line. B, Clontech RNA extracts subjected to conventional RT-PCR using the same primers as positive controls. C, real-time (LightCycler) curves of 36B4 (std) RT-PCR primer product from the nontreated total RNA extracts from saliva (A, top left), comparing those reverse transcription reactions containing reverse transcriptase (+RT) and omitted reverse transcriptase (no-RT) conditions. There is no detectable RNA-derived signal, implying that there is no lower starting template in the no-RT condition, in which only genomic DNA is amplified, than in the RT condition, in which both genomic DNA and RNA are amplified. Note that a lower CRO implies a higher amount of starting template, in which k / 2(CRO) {approx} (starting template).

 

Figure 4
View larger version (40K):
[in this window]
[in a new window]
  		Fig. 4. Electrophoresis gel display (A and B) of ß-actin RT-PCR products, and real-time display (C) done on salivary RNA extracts under various conditions. A, using conventional ß-actin (std) PCR primers; PCR products of untreated, DNase-treated, and RNase-treated total RNA extracts from saliva, comparing no-RT and RT conditions. B, Clontech RNA standard, as a positive control, was subjected to conventional (std) RT-PCR. Results for GAPDH were identical (data not shown). C, real-time (LightCycler) curves of ß-actin (std) RT-PCR primer product from the nontreated total RNA extracts from saliva (A, top left), comparing those reverse transcription reactions containing reverse transcriptase (RT) and omitted reverse transcriptase (no-RT) conditions. There is no detectable RNA-derived signal, implying that there is no lower starting template in the no-RT condition, in which only genomic DNA is amplified, than in the RT condition, in which both genomic DNA and RNA are amplified. Note that a lower CRO implies a higher amount of starting template, in which k / 2(CRO) {approx} (starting template).

 

Figure 5
View larger version (42K):
[in this window]
[in a new window]
  		Fig. 5. Electrophoresis gel display of RNA-specific RT-PCR products, using the tagged URT primer set done on salivary RNA extracts under various conditions. A, using RNA-specific 36B4 (UR) PCR primers; PCR products of untreated, DNase-treated, and RNase-treated total RNA extracts from saliva, comparing no-RT and RT conditions. B, RNA-specific ß-actin (UR) PCR primers; PCR products of untreated, DNase-treated, and RNase-treated total RNA extracts from saliva, comparing no-RT and RT conditions. There is no appropriate-size PCR product resulting from the use of either the 36B4 or ß-actin cDNA-specific primer set.

 
For the Clontech control RNA sample (Figs. 3B and 4B), the untreated and DNase-treated samples gave an RT product, but there was no PCR product resulting from the no-RT conditions. DNase did not extinguish this signal, whereas RNase extinguished the signal completely.

We attempted to quantitate the amounts of RNA-derived signal, if any, using real-time RT-PCR employing untreated saliva RNA extracts coupled with standard-design transcript primers (Fig. 3C shows 36B4; Fig. 4C shows ß-actin). In the no-RT condition, in which only homologous pseudogene-derived genomic DNA template is being amplified, there was consistently a higher quantity of original template (lower CRO) than in the "RT" condition, in which both RNA- and homologous pseudogene genomic DNA–derived signals are amplified. This result indicates that there is no incremental PCR signal detectable that is of RNA origin. Similarly, for RNase-treated saliva samples, there was no difference in PCR product between the no-RT and the RT conditions (data not shown). This result again implied that there was no RNA-derived signal.

When the RNA-specific strategy was used for RT-PCR, the mRNA-specific primers yielded no detectable product in any of the three treatments of the salivary RNA extracts across all conditions and all five individuals (Fig. 5A and B).


   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
There is an intensive search ongoing to identify noninvasively acquired biological fluids that can serve as surrogates for adjacent or distant visceral cancer-prone organs. Saliva is one such candidate body fluid. Recently Li and coauthors (3, 4) reported the use of salivary transcriptome diagnostics as well as RT-PCR in oral cancer detection. In the current study, we were unable to corroborate their assertion that saliva reliably supports RNA-specific amplification strategies by microarray or RT-PCR. In exploring the discrepancies between our saliva microarray and RT-PCR studies, and those of Li and coauthors (3, 4), we expended considerable effort to identify the origins of the signal.

Our microarray results indicate that saliva has no detectable RNA. Whereas for the Clontech control RNA standard, the hybridization signal was abolished with RNase treatment, there was very little signal loss when DNase treatment was used (Figs. 1 and 2). That pattern stands in contrast with the stability of the microarray signal intensity in saliva extracts treated with RNase (Figs. 1 and 2). The fact that the saliva signal was not abolished after RNase treatment suggests that whatever low concentration of nucleic acid was present in saliva was DNA in nature. It should be noted that the specificity of microarrays for RNA-derived signal has been under-studied; reliance on the efficacy of DNases (3–5), and on conventional probe design, is commonly assumed to be sufficient to ensure RNA specificity.

The results of the RT-PCR studies were also unambiguous; we showed the genomic DNA origin of the signal arising in the RT-PCR. After the salivary RNA extract had been isolated by a standard solid phase technique (eukaryotic and nonviral), a comprehensive series of conditions were then applied to untreated saliva RNA extracts, including the subjection of parallel samples to reciprocal DNase or RNase treatments, as well as RT and no-RT conditions. Our results clearly indicate that the saliva RT-PCR signal originates from genomic DNA and not RNA; DNase abolished the entire signal, whereas RNase did not. In addition, the presence of a signal in the no-RT controls shows that the template cannot arise from RNA because there is no PCR amplification from RNA without the reverse transcription step. The lack of any signal from the mRNA-specific strategy further confirms that there is no detectable RNA transcript present in the saliva samples for the three highly expressed reference genes studied (36b4, ß-actin, and GAPDH).

The lack of an apparent RNA-derived signal in this study is unlikely to be due to a lack of sensitivity; RT-PCR is widely accepted as the most sensitive method for measuring RNA, with high technical sensitivity (<5 transcript copies/cell), wide dynamic range of quantification (7-8 logarithmic decades), and high precision (<2% SD; ref. 8). Translationally, PCR-based methods are widely used in forensic and trace detection situations as the reference standard, and RT-PCR is capable of exquisitely sensitive cancer cell detection, down to one cancer cell per 106 normal cells (9, 10).

The specificity of many human sample RT-PCR assays remains open to question, due to the coamplification of carry-over genomic DNA–derived pseudogene sequences that contaminate the cDNA mixture when conventional RNA extraction approaches are used in combination with conventional RT-PCR (5, 11–13). The use of any generic RT primer, such as oligo-dT or random hexamers, renders the cDNA specificity of the reaction totally dependent on the specificity of the conventional PCR primer toward the cDNA sequence in question versus homologous genomic DNA sequence. The common, conventional approaches to the design of PCR primers for RT-PCR, such as the use of an individual oligomer spanning an exon/exon splice site, or the primer pair spanning an intron, however, are not sufficient to ensure mRNA-specific amplification in the context of homologous contaminating genomic DNA sequences in the reaction mix, as occurs for 36B4, ß-actin, and GAPDH reference amplifications (5).

There are several steps at which verification of the RNA specificity of a PCR product can be done. In the RT step, the reverse transcriptase can be omitted, yielding a PCR product that, if present, must be derived from genomic DNA. Similarly, parallel samples can be treated reciprocally with DNase (whereby only RNA-derived PCR products will be generated) or RNase (whereby only genomic DNA–derived PCR products will be generated). In addition, as previously shown by our laboratory (5), a tagged RT primer approach to reverse transcription of mRNA, and subsequent RT-coupled PCR, confer mRNA specificity to the quantitative PCR.

The quality of RNA is an important determinant of the reproducibility and relevance of subsequent quantitative RT-PCR experiments (14), yet RNA is widely recognized as being more labile than DNA, due to ubiquitous RNases. Li and coauthors acknowledge the common presumption that human mRNA does not survive extracellularly in saliva, as the original motivation for their study to prove otherwise (3). Unfortunately, their study does not report control RT-PCR experiments, particularly for the reference internal housekeeper transcripts ß-actin and GAPDH that they employed. Use of such controls might have addressed the RNA specificity of the RT-PCR in their saliva samples.

We cannot completely rule out the possibility that there was some potential for RNA-derived signals in that groups' microarray experiments. For example, conventional oral fluid (saliva) isolation can potentially be enhanced by the admixture of agitation-exfoliated buccal cells; brush-exfoliated buccal cell quantitative expression was first reported by our laboratory (1). It is therefore possible that more aggressive, unconventional oral fluid (saliva) collection techniques could have enhanced RNA yield. However, no such maneuvers were described (3). Furthermore, this should also have been apparent in the RT-PCR experiments that we attempted to replicate. Given the widely accepted idea that RT-PCR has much greater dynamic range and sensitivity than do microarray approaches, the absence of RNA-specific RT-PCR signals suggests that any RNA-derived signals must be below extraordinarily low detection limits.

It should be noted that RNA isolation in the studies of Li and coauthors (3) were done using a viral RNA isolation kit (QiaAmp, Qiagen), suitable for isolation of RNA from cell-free bodily fluids such as saliva. The protocol is not selective for human sequences over bacterial or viral sequences, nor for RNA over DNA, except to the extent that the use of DNase confers such specificity in the protocol. In view of the fact that most of the adult population harbors bacteria and viruses in saliva (15), we speculate that some of the signals reported by Li and coauthors (3) could potentially have been from bacterially or virally derived nucleic acids. For example, a BLAST search of several probe sequences present on the Affymetrix U133A chip that were used in the Li et al. (3) report for expression microarray–detected transcripts (e.g., DUSP1, IL-8, and GAPDH) showed close homology with DNA sequences from certain bacteria and viruses that are commonly found in human saliva (e.g., Streptococcus pyogenes and foot and mouth disease virus; http://www.ncbi.nlm.nih.gov/, accessed 9/22/2005). Therefore, false-positive microarray detection of such nonhuman sequences could have occurred, in addition to those derived from genomic DNA, in their study.

In conclusion, we found that there is no significant, intact human mRNA present in conventional saliva samples. The signal arising from microarray analysis is very low and likely nonspecific; the signal from RT-PCR in our study clearly originates from genomic DNA. Considerable caution should therefore be used in interpreting saliva-based mRNA expression studies. Because cancer diagnostics rely heavily on both the specificity of biomarker assays, and the quantitative nature of these markers' expression (16), careful consideration of these technical validation issues is a prerequisite before resource-intensive clinical utility validation efforts are undertaken.


   Acknowledgments
 
Acknowledgement is offered to Michael Ryan, Ph.D., for Wadsworth Center Genomics Core (Core MG) aid in microarray studies, the Wadsworth Center Biochemistry Core (Core BIO) for RNA quantification, Adrianna Verschoor, Ph.D., of the Wadsworth Research Office, for manuscript review, and laboratory members Weiguo Han and Stephane Cauchi for helpful discussions.


   Footnotes
 
Grant support: NIH R21CA94714, R21CA104812, and R01CA106186 (S.D. Spivack).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

5 Unpublished results. Back

Received 3/ 2/06; revised 5/19/06; accepted 6/23/06.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Spivack SD, Hurteau GJ, Jain R, et al. Gene-environment interaction signatures by quantitative mRNA profiling in exfoliated buccal mucosal cells. Cancer Res 2004;64:6805–13.[Abstract/Free Full Text]
  2. Weikert S, Cristoph F, Schraeder M, et al. Qualitative analysis of survivin mRNA expression in urine and tumor tissue of bladder cancer patients and its potential relevance for disease detection and prognosis. Int J Cancer 2005;116:100–4.[CrossRef][Medline]
  3. Li Y, St. John MAR, Zhou X, et al. Salivary transcriptome diagnostics for oral cancer detection. Clin Cancer Res 2004;10:8442–50.[Abstract/Free Full Text]
  4. Li Y, Zhou X, St. John MAR, et al. RNA profiling of cell-free saliva using microarray technology. J Dent Res 2004;83:199–203.[Abstract/Free Full Text]
  5. Hurteau GJ, Spivack SD. mRNA-specific reverse transcription-polymerase chain reaction from human tissue extracts. Anal Biochem 2002;307:304–15.[CrossRef][Medline]
  6. Han W, Pentecost BT, Pietropaolo RL, et al. Estrogen receptor {alpha} increases basal and cigarette smoke extract-induced expression of CYP1A1 and CYP1B1, but not GSTP1, in normal human bronchial epithelial cells. Mol Carcinog 2005;44:202–11.[CrossRef][Medline]
  7. Spivack SD, Hurteau GJ, Fasco MJ, et al. Phase I and II carcinogen metabolism gene expression in human lung tissue and tumors. Clin Cancer Res 2003;9:6002–11.[Abstract/Free Full Text]
  8. Klein D. Quantification using real-time PCR technology: applications and limitations. Trends Mol Med 2002;8:257–60.[CrossRef][Medline]
  9. Schoenfeld A, Kruger KH, Gomm J, et al. The detection of micrometastases in the peripheral blood and bone marrow of patients with breast cancer using immunohistochemistry and reverse transcriptase polymerase chain reaction for keratin 19. Eur J Cancer 1997;33:854–61.[CrossRef][Medline]
  10. Stathopoulou A, Vlachonikolis I, Mavroudis D, et al. Molecular detection of cytokeratin-19-positive cells in the peripheral blood of patients with operable breast cancer: evaluation of their prognostic significance. J Clin Oncol 2002;20:3404–12.[Abstract/Free Full Text]
  11. Lacroix J, Becker HD, Woerner SM, et al. Sensitive detection of rare cancer cells in sputum and peripheral blood samples of patients with lung cancer by preproGRP-specific RT-PCR. Int J Cancer 2001;92:1–8.[CrossRef][Medline]
  12. Lupberger J, Kreuzer KA, Baskaynak G, et al. Quantitative analysis of ß-actin, ß-2-microglobulin and porphobilinogen deaminase mRNA and their comparison as control transcripts for RT-PCR. Mol Cell Probes 2002;16:25–30.[CrossRef][Medline]
  13. Lion T. Debate round table on RT-PCR controls: concluding remarks and mini-review—Current recommendations for positive controls in RT-PCR assays. Leukemia 2001;15:1033–7.[CrossRef][Medline]
  14. Bustin SA, Nolan T. Pitfalls of quantitative real-time reverse-transcription polymerase chain reaction. J Biomol Tech 2004;15:155–66.[Medline]
  15. Miller CS, Avdiushko SA, Kryscio RJ, et al. Effect of prophylactic valacyclovir on the presence of human herpesvirus DNA in saliva of healthy individuals after dental treatment. J Clin Microbiol 2005;43:2173–80.[Abstract/Free Full Text]
  16. Westra WH, Califano J. Toward early oral cancer detection using gene expression profiling of saliva: a thoroughfare or dead end? Clin Cancer Res 2004;10:8130–1.[Free Full Text]



This article has been cited by other articles:


Home page
 Clin. Chem.Home page
Z. Hu, B. G. Zimmermann, H. Zhou, J. Wang, B. S. Henson, W. Yu, D. Elashoff, G. Krupp, and D. T. Wong
Exon-Level Expression Profiling: A Comprehensive Transcriptome Analysis of Oral Fluids
Clin. Chem., May 1, 2008; 54(5): 824 - 832.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Chem.Home page
S. Rani, M. Clynes, and L. O'Driscoll
Detection of Amplifiable mRNA Extracellular to Insulin-Producing Cells: Potential for Predicting Beta Cell Mass and Function
Clin. Chem., November 1, 2007; 53(11): 1936 - 1944.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Cancer Res.Home page
J. Ballantyne
Validity of Messenger RNA Expression Analyses of Human Saliva
Clin. Cancer Res., February 15, 2007; 13(4): 1350 - 1350.
[Full Text] [PDF]

Home page
 Clin. Cancer Res.Home page
S. D. Spivack

Clin. Cancer Res., February 15, 2007; 13(4): 1351 - 1351.
[Full Text] [PDF]

Home page
 Clin. Cancer Res.Home page
D. T. Wong
Salivary Transcriptome
Clin. Cancer Res., February 15, 2007; 13(4): 1350 - 1350.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kumar, S. V.
Right arrow Articles by Spivack, S. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kumar, S. V.
Right arrow Articles by Spivack, S. D.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online