Clinical Chemistry Siemens Point of Care - Urinalysis
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Clinical Chemistry 54: 396-405, 2008. First published December 18, 2007; 10.1373/clinchem.2007.093419
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
clinchem.2007.093419v1
54/2/396    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
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 Wright, C.
Right arrow Articles by Fare, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wright, C.
Right arrow Articles by Fare, T.
Related Collections
Right arrow Molecular Diagnostics and Genetics
(Clinical Chemistry. 2008;54:396-405.)
© 2008 American Association for Clinical Chemistry, Inc.

Molecular Diagnostics and Genetics

Characterization of Globin RNA Interference in Gene Expression Profiling of Whole-Blood Samples

Christopher Wright1, Donald Bergstrom2, Hongyue Dai1, Matthew Marton1, Mark Morris1, George Tokiwa1, Yanqun Wang1 and Thomas Fare1,a

1 Rosetta Inpharmatics, Merck & Co., Inc., Seattle, WA; 2 Merck Research Laboratories, Upper Gwynedd, PA.

aAddress correspondence to this author at: Rosetta Inpharmatics, Merck & Co., Inc., 401 Terry Ave. N., Seattle, WA 98109. Fax 206-802-6501; e-mail thomas_fare{at}merck.com.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Background: Blood-based biomarker discovery with gene expression profiling has been hampered by interference from endogenous, highly abundant {alpha}- and β-globin transcripts. We describe a means to quantify the interference of globin transcripts on profiling and the effectiveness of globin transcript mitigation by (a) defining and characterizing globin interference, (b) reproducing globin interference with synthetic transcripts, and (c) using ROC curves to measure sensitivity and specificity for a protocol for removing {alpha}- and β-globin transcripts.

Methods: We collected blood at 2 sites and extracted total RNA in PreAnalytiX PAXgene tubes. As a reference for characterizing interference, we supplemented aliquots of total RNA with synthesized globin transcripts and total RNA from human brain. Selected aliquots were processed with Ambion GLOBINclear to remove globin transcripts. All aliquots were labeled and hybridized to Agilent DNA microarrays by means of pooling schemes designed to quantify the mitigation of globin interference and to titrate gene expression signatures. Quantitative reverse transcription–PCR data were generated for comparison with microarray results.

Results: Our supplementation and pooling strategy for comparing the microarray data among samples demonstrated that mitigation could reduce an interference signature of >1000 genes to approximately 200. Analysis of samples of endogenous globin transcripts supplemented with brain RNA indicated that results obtained with the GLOBINclear treatment approach those of peripheral blood mononuclear cell preparations.

Conclusion: We confirmed that both the absolute concentrations of globin transcripts and differences in transcript concentrations within a sample set are factors that cause globin interference (Genes Immun 2005;6:588–95). The methods and transcripts we have developed may be useful for quantitatively characterizing globin mRNA interference and its mitigation.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Gene expression profiling of peripheral blood can be a powerful tool for biomarker discovery in clinical trials (1)(2)(3)(4). Although peripheral blood samples are collected routinely in clinical practice, maintaining RNA integrity in such samples and stabilizing their biological signatures have proved difficult. A widely used method for obtaining total RNA from leukocytes requires isolating peripheral blood mononuclear cells (PBMCs)1 by density-gradient centrifugation, which requires both instrumentation beyond what is available in the typical clinical setting and manipulations of 1– 2 h that may cause interfering signatures (5).

To reduce the occurrences of such signatures, investigators have introduced technologies that decrease the time between blood draw and RNA stabilization [e.g., PAXgene (PreAnalytiX), Tempus (Applied Biosystems)]. Although these systems may stabilize RNA at the point of collection, they introduce interfering signatures because of the abundance and variability in the amounts of {alpha}- and β-globin mRNA, which obscure signatures of biological interest (1)(6)(7). The degree of globin message representation in mRNA in whole blood can vary widely, with globin mRNA species constituting up to 70% of whole-blood mRNA in patients with high reticulocyte counts (8). We describe methods for quantifying and characterizing the globin message in samples, how to recognize a globin message–induced signature in profiling data, and the efficacies of measures for mitigating interference by globin messages.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
We describe samples, processing methods, pooling strategies, and experimental-design features for generating data that quantify globin RNA interference and its mitigation. Our strategy consists of pooling samples of total RNA to make homogeneous samples of RNA populations that we can supplement with different concentrations of globin transcripts, characterizing the degree of interference by the globin message, and measuring the effects of measures of globin transcript mitigation on expression profiles. We also used an RNA pool with endogenous globin transcripts to measure how mitigation measures affect the ability to detect an exogenous RNA signature (from brain tissue).

blood collection and extraction of total rna
We drew blood samples from volunteers after they had provided informed consent and the protocols had been approved. At site A (BSC Labs), 1 unit of whole blood was drawn from each of 10 healthy donors and preserved with EDTA. We filled 20 PAXgene tubes (VWR catalog no. 77776–026) for subsequent extraction of total RNA (sample sets A and D). The EDTA-preserved blood (75 mL) was added to 3 Accuspin tubes (Sigma-Aldrich, catalog no. A7054) within 2 h of collection for PBMC isolation and extraction of total RNA (sample set E) with an in-house semiautomated procedure that follows the protocol (9) accompanying the PAXgene 96 Blood RNA Kit (Qiagen, catalog no. 762331). We conducted no additional purification steps before treating samples with GLOBINclear (Ambion). At site B, we drew blood directly into 10 PAXgene tubes for each of 10 healthy donors and extracted total RNA (sample sets B and C). We created intradonor pools for each donor/extraction protocol set (see Pooling and Supplementation Strategy and Fig. 1 ).


Figure 1
View larger version (27K):
[in this window]
[in a new window]

  		Figure 1. Schematic of sample preparation for assessing interference by globin mRNA and assessing mitigation technologies for both endogenous and synthetic interferences.

(A), sample processing for creating and assessing an endogenous interference. Blood was first drawn from each donor into an EDTA-containing bag to obtain a volume sufficient for the experiment and then transferred into PAXgene tubes for stabilization. Total RNA was extracted from the samples in PAXgene tubes and pooled for each donor. Donor pools were amplified and hybridized against a multidonor cRNA pool in fluor-reverse pairs (FRPs). (B), sample processing for creating and assessing an exogenous interference. RNA from blood tubes was extracted and combined into a single all-donor pool, which was then split and used as the reference for hybridization experiments with samples supplemented with globin transcripts. (C), Splitting of donor sample to create brain RNA supplementation sets (0, 0.5, and 5.0 mg/g total RNA) to evaluate sensitivity. Supplement pools were amplified and hybridized in FRPs against a cRNA pool containing no supplement.

supplementation with globin transcripts
We added {alpha}- and β-globin transcripts to human whole-blood samples containing total RNA and used these supplemented samples to quantify the reduction in the globin message and the mitigation of the interference. We used multiple sequences of globin transcripts in the NCBI Reference Sequence collection (RefSeq) to construct consensus sequences for {alpha}- and β-globin mRNA (see Figs. 1 and 2 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol54/issue2 ) and added 30 nucleotides of poly(A) 3' to the sequence. We then submitted these sequences to Blue Sky Biotech for subcloning and in vitro transcription. They isolated full-length clones, subcloned them, and sequenced the clones for verification. After we verified the transcript sequences with our RefSeq consensus, a sequence analysis identified a point mutation in the 3' untranslated region of the β-globin clone that coincided with the array probe for the β-globin transcript; regardless, the array probe was saturated because of the abundance of the β-globin message.

supplementation with human brain total rna
We obtained human brain total RNA from BioChain (catalog no. R1234035–50, lot no. A703158) and added it to aliquots of peripheral blood total RNA at the following concentrations: 0 mg/g, 0.5 mg/g, and 5 mg/g. We then used the resulting exogenous brain RNA signature and ROC curve analysis to evaluate the sensitivity and specificity of each mitigation protocol (10).

globin transcript mitigation
Ambion’s GLOBINclear product, a globin transcript–mitigation technology for removing {alpha}- and β-globin transcripts, is used to pretreat total RNA before amplifying a target. We precipitated a 2-µg aliquot of each sample in ethanol, resuspended the sample in 14 µL water, and processed the sample with the GLOBINclear kit according to the manufacturer’s protocol (11). In brief, we mixed custom biotinylated oligonucleotides complementary to globin RNA sequences with RNA prepared from blood samples and annealed the oligonucleotides to {alpha}- and β-globin transcripts. We then added streptavidin-coated paramagnetic beads to bind the biotinylated duplexes and remove the captured globin transcripts from the preparations of total RNA, adjusted the GLOBINclear-treated samples of total RNA to 150 µL, and used aliquots for quantitative reverse transcription–PCR (qRT-PCR) and microarray experiments.

pooling and supplementation strategy
Characterization of globin transcript interference in microarray data (sample sets A and B).
We used several pooling strategies to generate reference channels for the 2-color array format used in this study. In one strategy (Fig. 1AUp ), we pooled total RNA extracted from blood that had been aliquoted into PAXgene tubes from a given donor, combined copy RNA (cRNA) generated from each donor pool of total RNA to form a multidonor pool [also referred to as a mass-balanced, self-referenced pool (10)] to make interdonor comparisons, and compared each donor with the pool. We used the resulting hybridization data to characterize globin transcript interference. We used this pooling strategy with sample sets A and B to quantify such interference, with 2 separate interdonor comparisons as examples.

Creating and mitigating interference by synthetic globin transcripts (sample set C).
In another strategy (Fig. 1BUp ), we combined total RNA from multiple donors to create a multidonor pool for supplementation experiments. We split aliquots of pooled total RNA into separate containers to which we had added a titration of synthetic globin transcripts. We generated cRNA from each supplemented sample and formed a mass-balanced, self-referenced pool from all cRNA preparations for fluor-reverse pairing in hybridization experiments. We used sample set C to quantify the effects of globin transcript mitigation on microarray data via comparison with untreated samples.

Mitigation of an endogenous globin interference (sample sets D and E).
In the third set of experiments, we quantified the effect of endogenous globin interference on the ability to recover an exogenous brain signature. Sample sets D and E were formed according to the pooling scheme summarized in Fig. 1CUp . We split a donor sample into aliquots and added different known amounts of brain RNA into the individual aliquots. cRNA was made from each aliquot, and a self-referenced pool of all donors was made from cRNA synthesized from aliquots with no added brain RNA. We hybridized cRNA from the supplemented samples against cRNA synthesized from samples with no supplemented brain RNA and measured the ability to identify brain-specific sequences for sample sets D and E, with and without mitigation of globin mRNA.

qRT-PCR analysis with taqman and sybr® green
We selected primer sets targeted to the second intron/exon junctions of both {alpha}- and β-globin mRNA. We submitted sequences for 200-nucleotide stretches spanning intron/exon junction 2 of {alpha}- and β-globins to Applied Biosystems’ Assay-by-Design service for TaqMan primer/probe sets, which produced the following forward, reverse, and reporter {alpha}-globin sequences, respectively: 5'-GCACGCGCACAAGCT-3', 5'-GGGTCACCAGCAGGCA-3', and 5'-ACTTCAAGCTCCTAAGCCAC-3'. The forward, reverse, and reporter β-globin sequences were 5'-AAGCTGCACGTGGATCCT-3', 5'-GATGGGCCAGCACACAGA-3', and 5'-CCCAGGAGCCTGAAGTT-3', respectively. We also used these primers for SYBR Green qRT-PCR analysis. For total RNA and cRNA, we used Applied Biosystems’ High Capacity Archive Kit (catalog no. 4322171) with 10 ng of each sample to generate T7- dT–primed cDNA. In short, total RNA was dried down and resuspended in 20 µL master mix containing 1x deoxynucleoside triphosphates and 1x RT buffer, 50 U MultiscribeTM, 40 U RNase OUT (Invitrogen, catalog no. 10777–019) and 5 pmol oligo-dT (Ambion, catalog no. AM5730G). Reactions were carried out on a DNA Engine Tetrad (MJ Research) at 70 °C for 5 min, 37 °C for 120 min, and 95 °C for 5 min).

We conducted quadruplicate TaqMan or SYBR Green assays with 100-pg cDNA aliquots and the primer or probe, respectively. We carried out all reactions simultaneously in an Applied Biosystems 7900HT Fast Real-Time PCR System and calculated results by means of a relative-abundance method (12)(13).

sample amplification and microarray analysis
We profiled and analyzed samples of total RNA extracted from human whole blood (14)(15). In brief, we amplified total RNA from blood by means of a modified 2-round reverse transcription reaction mediated by Moloney murine leukemia virus reverse transcriptase, followed by in vitro transcription and labeling with a Cy dye. Samples were hybridized to custom 25 000–probe oligonucleotide arrays (Agilent Technologies) in fluor-reverse pairs. Scanned microarray images were processed with a feature extractor developed in house with MATLAB (The MathWorks). The feature extractor automatically locates the arrayed features on a scanned image, calculates the mean pixel intensity for each feature, and flags features that either show artifacts or have intensities at the scanner’s background or saturation level. Feature intensities normalized by the mean intensity of the nonflagged features for the Cy3 and Cy5 channels are used to form ratios of the 2 channels for each reporter. An in-house error model based on a null hypothesis of no differential regulation was used to assess the significance of an observed ratio (16). Hybridization-ratio data were analyzed with MATLAB.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
characterization of globin transcript interference in microarray data
We first encountered globin transcript interference with microarray data from healthy donors during expression profiling of blood donor variability. We found globin mRNA interference in 2 separate sets of donor samples (Fig. 2 ). Fig. 2A is a 1-D agglomerative cluster of a panel of 7 (of a total of 10) donor blood samples from sample set A; all 7 samples were hybridized against a mass-balanced, self-referenced pool (16). The probe content chart on the horizontal axis displays the window-averaged nucleotide count (labeled A, T, C, and G) for each probe ordered by the 1-D clustering algorithm. We used these charts to assess possible content bias in clusters. For example, we observe in Fig. 2A that the GC content per probe to the left of the yellow (dot-dash) line is approximately 35% higher than the GC content of probes to the right of the line. Nonspecific probe-target annealing has been associated with a high GC content of probes (17).


Figure 2
View larger version (67K):
[in this window]
[in a new window]

  		Figure 2. Characterization of interference by synthetic globin message (sample sets A and B).

Heat maps (A, B) show qRT-PCR–sorted 1-D agglomerative clusters of 2 healthy multidonor, self-referenced experiments for sample sets A and B, respectively. Transcripts that were differentially regulated by 2-fold were selected to go into the heat map. The figure comprehensively shows a heat map of ratio profiles for a set of experiments and probes (vertical and horizontal axes, respectively), a bar chart of qRT-PCR–derived globin RNA content (as a proportion of the total RNA), and a probe content chart above the heat map. Cross-hybridization and normalization interferences (see text for details) are demarcated by hatched yellow lines. The line charts above the heat maps plot the nucleotide content of probes. (C), a histogram of globin RNA abundance as measured by qRT-PCR and sorted as a percentage of the total RNA. Sample sets for clusters A and B are noted. (D), plot showing correlation of array data with qRT-PCR data for combined {alpha}- and β-globin data.

We have arranged the experiments summarized in Fig. 2Up , A and B, on the vertical axis in descending order of qRT-PCR–derived globin transcript content. This ordering shows the interference as a globin content–dependent pattern of gene regulation. One can divide the pattern into 2 regions: cross-hybridization and normalization (Fig. 2AUp ). We call cross-hybridization a clustering of differentially expressed genes that can be associated with AT-poor/GC-rich probes when experiments are arranged in order by globin mRNA content in heat maps. Within the constraints of the 3'-biased probe selection, we obtained approximately 1250 cross-hybridization genes (P <0.01) and have observed gene sets as large as 2500 in more severe cases (data not shown). In ratio-based microarray experiments, we used the mean intensity of all biological features to normalize feature intensities in each channel over the entire chip (16). In this case, the intensity due to cross-hybridization is sufficient to skew channel normalization, producing an "inverse" effect. In other words, for a set of genes found to be up-regulated because globin transcript cross-hybridization, there is another set of genes calculated to be down- regulated, compensating for the skewed signal levels. In this sense, the normalization effect is a consequence of globin transcript cross-hybridization interference. Specifically, sample set B has fewer than 250 cross-hybridization genes, compared with 1250 in sample set A. Consequently, the normalization interference is not as pronounced in sample set B, and the globin-dependent pattern is not as robust.

To investigate further the difference between the sample sets, we analyzed the qRT-PCR data and their relationship to the microarray data. Fig. 2CUp is a bar chart of the qRT-PCR data for sample sets A and B and is arranged by globin transcript content. We note that the overall content of {alpha}- and β-globin transcripts as a percentage of total RNA is higher in sample set A. Furthermore, sample set A shows an absolute difference of approximately 3% in the proportion of globin RNA between the samples with the highest and lowest proportions (approximately 4.5% and approximately 1.5%, respectively; Fig. 2CUp ), which represents a 3-fold change in the relative amount of globin transcripts. We see with sample set B a corresponding absolute change of 0.7% or a relative change of just over 2-fold (approximately 1.3% and 0.6%, respectively). This analysis reveals that the total abundance of and variation in globin RNA content for sample set A is greater than for sample set B.

In Fig. 2DUp , we show how globin mRNA concentration expressed as a percentage of total RNA (as measured by qRT-PCR) correlates with the mean log ratio (MLR) for {alpha}-globin, β-globin, and {epsilon}-globin (an embryonic form of globin with 50% homology to β-globin). The microarray MLR of a gene is defined as the log10 ratio of the mean of the 2 normalized treatment-channel intensities to the mean of the 2 normalized control-channel intensities. Intensities are normalized by dividing the specific spot intensity by the global biological intensity for the entire chip. The qRT-PCR MLR is defined as the log10 of the specific abundance divided by the mean of all the abundance data for the experiment. Although the microarray and qRT-PCR data show good correlation for all 3 globins (R2{alpha} = 0.82; R2β = 0.75; R2{epsilon} = 0.81), the higher slope, m, of the trend line for {epsilon} globin (m{alpha} = 6.2; mβ = 10.1; m{epsilon} = 15.5) indicates a larger dynamic range for the MLR. The shallower slopes for {alpha}-globin and β-globin were due to saturation of the respective probes. From the data in Fig. 2Up , we conclude that qRT-PCR of HBE12 (hemoglobin, epsilon 1) MLR data can be used to assess the likelihood of globin transcript interference in a sample set.

creating and mitigating the interference by synthetic globin rna
To characterize the interference further and assess potential mitigation technologies, we developed a means to reliably recreate globin RNA interference by titrating synthesized globin gene transcripts into a background of total RNA extracted from pooled samples of whole blood. We measured the ratio of {alpha}-globin mRNA to β-globin mRNA for donor sets A and B via qRT-PCR analysis of total RNA. For each donor, we measured a consistent {alpha}-globin/β-globin mRNA ratio of 3:1 [for example, the mean {alpha}-globin/β-globin mRNA ratio for the data in Fig. 2CUp was 74.6:25.4 (n = 17), CV{alpha} = 3%, and CVβ = 9%]. We used this ratio for our globin mRNA–supplementation experiments. To the same parent pool of total RNA, we added synthesized globin RNA to concentrations of 0, 10, 25, 40, and 70 mg/g total RNA (sample set C). For simplicity, we refer to the experimental points by the supplementation amounts. Our intent was to create substantial interference because the highest proportion of endogenous globin RNA that we had previously encountered was 65 mg/g of the total RNA (data not shown). We also wanted to ensure large variation in the globin RNA content within a sample set.

For sample set C, we divided each sample into 2 aliquots, one of which we treated with GLOBINclear and the other we left untreated. BioAnalyzer traces with the RNA 6000 Pico LabChip (Agilent Technologies) showed that GLOBINclear effectively eliminates the globin RNA peak at all supplementation levels (see Fig. 3 in the online Data Supplement). Furthermore, a qRT-PCR analysis of total RNA before and after treatment demonstrates a reduction of representation in all samples (Table 1 ). In particular, 97% ({varsigma} = 0.23%) of the globin in the sample supplemented with globin mRNA at 70 mg/g was removed from total RNA. Absolute representation was reduced for all globin RNA additions by at least 94% of the original content. Not only are the magnitudes of globin mRNA representation lowered overall, but the difference between the samples with highest and lowest globin mRNA content ({alpha}-globin plus β-globin) was also greatly reduced after GLOBINclear treatment.


View this table:
[in this window]
[in a new window]

  		Table 1. The relative abundance of globin messages in total RNA and cRNA.1

We then amplified treated and untreated samples for 2 independent self-referenced pool hybridization plans (Fig. 1BUp ). We quantified the globin transcript content of the cRNA produced, including the self-referenced pools, with the same globin transcript assays used with total RNA. Table 1Up shows that the amplification product increases as the amount of added globin RNA increases. In addition, the self-referenced pool has the approximate mean globin transcript content (approximately 25 mg/g) expected from mixing the total set of supplemented samples together. We conclude from these data that GLOBINclear-treated samples have substantially reduced globin message, both in the starting total RNA and in the subsequent cRNA product. We hybridized samples according to a self-referenced pool scheme. We found in array hybridization data (Fig. 3 ) that a 1-D agglomerative cluster for the untreated controls (left heat map) demonstrated globin message interference; that is, a large signature (>1500 genes, P <0.01) correlated with A-poor/GC-rich probes and a corresponding normalization effect. The magnitudes and directions of the interference signatures correlate with the supplement quantity. A comparison with the GLOBINclear-treated sample (right heat map) demonstrates that the interference is essentially eliminated. Whereas the mean number of signatures per supplement for the untreated sample panel was 1126, the GLOBINclear-treated sample set had a mean of 210 signatures, a difference that is statistically significant (P <0.01). The remnant GLOBINclear signatures do not correlate with the A-poor/GC-rich probes and represent <1% of the total features differentially detected.


Figure 3
View larger version (40K):
[in this window]
[in a new window]

  		Figure 3. Mitigation of interference by synthetic globin transcript (sample set C).

1-D agglomerative clusters of self-referenced hybridizations according to treatment group. Experiments are sorted by the concentration of supplemented globin mRNA relative to the total RNA concentration. Each plot has the same gene set and order. Transcripts that were differentially regulated by 2-fold were selected to go into the no-treatment heat map; the same set of transcripts is represented in the GLOBINclear heat map. Signature counts for each supplement level are noted at the side of each heat map. The nucleotide compositions of the probes are noted above each plot. Cross-hybridization and normalization interferences are demarcated by hatched yellow lines.

mitigation of interference by endogenous globin transcripts
To quantify mitigation strategies for a traceable endogenous signature, we created sample sets D and E. We supplemented donor total RNA with brain total RNA (0, 0.5, and 5.0 mg/g) and then treated this sample set with GLOBINclear or not. In addition, we created sample set E (see Materials and Methods) as a reference for globin transcript mitigation. We processed each sample set as depicted in Fig. 1CUp , with the modification that the reference pool was not supplemented with brain RNA. This strategy induced a signature in total RNA from blood.

Fig. 4 is a combined 1-D agglomerative cluster with experiments organized first into 3 groups by treatment (no mitigation, GLOBINclear, PBMCs) and then by HBE1 MLR, according to the no-mitigation protocol. The cross-hybridization and normalization effects are marked in the figure. Without mitigation, there is substantial interference when the data are ordered by HBE1 expression. GLOBINclear treatment significantly reduces both globin message and normalization interferences to background levels (from approximately 1200 signatures to 0; P = 0.01).


Figure 4
View larger version (63K):
[in this window]
[in a new window]

  		Figure 4. Mitigation of endogenous globin transcript interference for samples with no mitigation and GLOBINclear-treated samples (sample set D) and for PBMCs (sample set E).

A 1-D agglomerative cluster with experiments sorted first by mitigation protocol; all 3 experiments were then sorted by no-mitigation HBE1 MLR order. The plot to the left is the HBE1 MLR for the 3 experiments. Transcripts that were differentially regulated by 2-fold were selected to go into the heat map. The probe content plot is sorted with the heat map; the inset to the immediate left is a detail of the brain signature. Yellow rectangles demarcate the globin message interference and the associated normalization. Yellow lines demarcate the brain signature.

Next, we identified a brain signature (18)(19) (Fig. 4Up inset). We used ROC curve analysis to measure the impact of GLOBINclear treatment on sensitivity and specificity, relative to the brain signature (Fig. 5 ). We specifically compared nontreated samples with GLOBINclear-treated samples and PBMC samples with respect to the 0.5-mg/g brain signature. The sensitivity and specificity of GLOBINclear approach those obtained with PBMCs and show a nominal improvement over no treatment. Each curve in the figure represents the mean of 36 arrays (9 samples, 4 arrays) for a given sample condition.


Figure 5
View larger version (36K):
[in this window]
[in a new window]

  		Figure 5. ROC curves of globin-mitigation protocols.

The y-axis is the number of verified brain signatures detected in samples supplemented with brain mRNA (0.5 mg/g total RNA). The x-axis is the number of false positives that arise from same-vs-same comparisons. The hatched line represents cases in which the number of false positives equals the number of true positives.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
The desire to use human whole blood collected in clinical settings for microarray profiling and biomarker discovery has stimulated the commercial development of several technologies that stabilize RNA and mitigate interference by globin transcripts. We selected PAXgene blood-collection tubes, Ambion’s GLOBINclear technology, and the Agilent Technologies microarray platform to characterize reagents and methods for the purposes of quantifying a globin message–induced interference and investigating its mitigation. The development of reproducible interference with synthetic globin transcripts was critical to evaluating globin interference–mitigation strategies. Throughout this study, we used these reagents in a model that was not influenced by the variability in donor samples. We used qRT-PCR and BioAnalyzer analyses in conjunction with heat maps, cross-hybridization analyses, and ROC curve analysis to confirm that a commercially available kit could mitigate interference by synthetic globin message. Our application of the technology to a "natural" endogenous interference demonstrates the potential usefulness of the tools developed in our investigation of this approach. We confirmed the observation (20) that both the absolute concentration of globin transcripts and the difference in transcript concentrations within a sample set are factors in causing interference. We conclude that the methods and reagents we have developed may be useful quantitative tools for characterizing globin mRNA interference and its mitigation.


   Acknowledgments
 
Grant/funding Support: This research was supported by Merck & Co.

Financial Disclosures: None declared.

Acknowledgments: We acknowledge Robert Rosler, Jennifer Garnett, Jaime Forbes, Lori Roadcap, and Bin Li for technical support, Mark Parrish for critical reviews of the manuscript, and the Gene Expression Laboratory for sample processing and data generation.


   Footnotes
 
1 Nonstandard abbreviations: PBMC, peripheral blood mononuclear cell; qRT-PCR, quantitative reverse transcription–PCR; cRNA, copy RNA; MLR, mean log ratio.

2 Human genes: HBE1, hemoglobin, epsilon 1.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Tang Y, Gilbert DL, Glauser TA, Hershey AD, Sharp FR. Blood gene expression profiling of neurologic diseases. Arch Neurol 2005;62:210-215.[Abstract/Free Full Text]
  2. Wang Z, Neuburg D, Li C, Su L, Kim JY, Chen JC, Christiani DC. Global gene expression profiling in whole-blood samples from individuals exposed to metal fumes. Environ Health Perspect 2005;113:233-241.[Web of Science][Medline] [Order article via Infotrieve]
  3. Batliwalla FM, Li W, Ritchlin CT, Xiao X, Brenner M, Laragione T, et al. Microarray analyses of peripheral blood cells identifies unique gene expression signature in psoriatic arthritis. Mol Med 2005;11:21-29.[Web of Science][Medline] [Order article via Infotrieve]
  4. Ghielmetti M, Bellis M, Spycher MO, Miescher S, Vergeres G. Gene expression profiling of the effects of intravenous immunoglobulin in human whole blood. Mol Immunol 2006;43:939-949.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  5. Debey S, Schoenbeck U, Hellmich M, Gathof BS, Pillai R, Zander T, Schultze JL. Comparison of different isolation techniques prior to gene expression profiling of blood derived cells: impact on physiological responses, on overall expression and the role of different cell types. Pharmacogenomics J 2004;4:193-207.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  6. Liu J, Walter E, Stenger D, Thach D. Effects of globin mRNA reduction methods on gene expression profiles from whole blood. J Mol Diagn 2006;8:551-558.[Abstract/Free Full Text]
  7. Feezor RJ, Baker HV, Mindrinos M, Hayden D, Tannahill C, Brownstein L, et al. Whole blood and leukocyte RNA isolation for gene expression analyses. Physiol Genomics 2004;19:247-254.[Abstract/Free Full Text]
  8. Wu K, Miyada G, Martin J, Finkelstein D. Globin reduction protocol: a method for processing whole blood RNA samples for improved array results. Affymetrix Technical Note.http://www.affymetrix.com/support/technical/technotes/blood2_technote.pdf (Accessed October 2007)..
  9. Rainen L, Oelmueller U, Jurgensen S, Wyrich R, Ballas C, Schram J, et al. Stabilization of mRNA expression in whole blood samples. Clin Chem 2002;48:1882-1890.
  10. Parrish ML, Wei N, Duenwald S, Tokiwa GY, Wang Y, Holder D, et al. A microarray platform comparison for neuroscience applications. J Neurosci Methods 2004;132:57-68.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  11. Whitley P, Moturi S, Santiago J, Johnson C, Setterquist R. Improved microarray sensitivity using whole blood RNA samples. Ambion TechNote 2005.http://www.ambion.com/techlib/tn/123/10.html (Accessed October 2007)..
  12. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2–{Delta}{Delta}CT method. Methods 2001;25:402-408.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  13. Schmittgen TD, Jiang J, Liu Q, Yang L. A high-throughput method to monitor the expression of microRNA precursors. Nucleic Acids Res 2004;32:e43.[Abstract/Free Full Text]
  14. Hughes TR, Mao M, Jones AR, Burchard J, Marton MJ, Shannon KW, et al. Expression profiling using microarrays fabricated by an ink-jet oligonucleotide synthesizer. Nat Biotechnol 2001;19:342-347.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  15. Brandish P, Su M, Holder DJ, Hodor P, Szumlloskl J, Kleinhanz R, et al. Regulation of gene expression by lithium and depletion of inositol in slices of adult rat cortex. Neuron 2005;45:861-872.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  16. He Y, Dai H, Schadt E, Cavet G, Edwards S, Stepaniants S, et al. Microarray standard data set and figures of merit for comparing data processing methods and experiment designs. Bioinformatics 2002;19:956-965.[CrossRef][Web of Science]
  17. Tijssen P. Overview of principles of hybridization and the strategy of nucleic acid probe assays. Laboratory techniques in biochemistry and molecular biology: hybridization with nucleic acid probes, part I: theory and nucleic acid preparation 1993;Vol. 24:19-78 Elsevier Science Amsterdam, The Netherlands. .
  18. Schadt E, Edwards S, GuhaThakurta D, Holder D, Ying L, Svetnik V, et al. A comprehensive transcript index of the human genome generated using microarrays and computational approaches. Genome Biol 2004;5:R73.[CrossRef][Medline] [Order article via Infotrieve]
  19. Bleasby K, Castle J, Roberts C, Cheng C, Bailey W, Sina J, et al. Expression profiles of 50 xenobiotic transporter genes in humans and pre-clinical species; a resource for investigations into drug disposition. Xenobiotica 2006;36:963-988.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  20. Thach DC, Agan BK, Olsen C, Diao J, Lin B, Gomez J, et al. Surveillance of transcriptomes in basic military trainees with normal, febrile respiratory illness, and convalescent phenotypes. Genes Immun 2005;6:588-595.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]



The following articles in journals at HighWire Press have cited this article:


Home page
 Clin. Chem.Home page
H. P.Y. Fan, C. Di Liao, B. Y. Fu, L. C.W. Lam, and N. L.S. Tang
Interindividual and Interethnic Variation in Genomewide Gene Expression: Insights into the Biological Variation of Gene Expression and Clinical Implications
Clin. Chem., April 1, 2009; 55(4): 774 - 785.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Chem.Home page
A. Kohlmann, E. Haschke-Becher, B. Wimmer, A. Huber-Wechselberger, S. Meyer-Monard, H. Huxol, U. Siegler, M. Rossier, T. Matthes, M. Rebsamen, et al.
Intraplatform Reproducibility and Technical Precision of Gene Expression Profiling in 4 Laboratories Investigating 160 Leukemia Samples: The DACH Study
Clin. Chem., October 1, 2008; 54(10): 1705 - 1715.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
clinchem.2007.093419v1
54/2/396    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
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 Wright, C.
Right arrow Articles by Fare, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wright, C.
Right arrow Articles by Fare, T.
Related Collections
Right arrow Molecular Diagnostics and Genetics

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS