Impact of blood collection and processing on peripheral blood gene expression profiling in type 1 diabetes
© The Author(s). 2017
Received: 7 April 2017
Accepted: 17 July 2017
Published: 18 August 2017
The natural history of type 1 diabetes (T1D) is challenging to investigate, especially as pre-diabetic individuals are difficult to identify. Numerous T1D consortia have been established to collect whole blood for gene expression analysis from individuals with or at risk to develop T1D. However, with no universally accepted protocol for their collection, differences in sample processing may lead to variances in the results. Here, we examined whether the choice of blood collection tube and RNA extraction kit leads to differences in the expression of genes that are changed during the progression of T1D, and if these differences could be minimized by measuring gene expression directly from the lysate of whole blood.
Microarray analysis showed that the expression of 901 genes is highly influenced by sample processing using the PAXgene versus the Tempus system. These included a significant number of lymphocyte-specific genes and genes whose expression has been reported to differ in the peripheral blood of at-risk and T1D patients compared to controls. We showed that artificial changes in gene expression occur when control and T1D samples were processed differently. The sample processing-dependent differences in gene expression were largely due to loss of transcripts during the RNA extraction step using the PAXgene system. The majority of differences were not observed when gene expression was measured in whole blood lysates prepared from blood collected in PAXgene and Tempus tubes.
We showed that the gene expression profile of samples processed using the Tempus system is more accurate than that of samples processed using the PAXgene system. Variation in sample processing can result in misleading changes in gene expression. However, these differences can be minimized by measuring gene expression directly in whole blood lysates.
Type 1 Diabetes (T1D) is characterized by the gradual autoimmune destruction of insulin-producing pancreatic beta cells. It is the most common autoimmune disease in children, with approximately 1 in 500 affected in the USA . However, the disorder’s pathogenesis and natural history remains unclear due, in part, to the difficulty in identifying and obtaining samples from at-risk individuals prior to the onset of hyperglycemia, and the lack of validated biomarkers of the disease. To fill the void and meet this need, various multicenter T1D consortia have been established to screen and follow susceptible individuals and recently diagnosed subjects, and to collect biological samples from such persons (e.g., peripheral blood cells (PBC)). For example, the NIH TrialNet Pathway to Prevention study, NIH Environmental Determinants of Diabetes in the Young study (TEDDY), Diabetes Autoimmune Study in the Young (DAISY), and Type 1 Diabetes Prediction and Prevention Project (DIPP) have collectively enrolled more than 20,000 individuals with an increased risk of developing T1D. The samples provided by these repositories represent a valuable resource, allowing researchers to control, at a geographically diverse level, for ethnic or environmental factors that might confound samples collected at a single center.
Whole blood gene expression analysis may be used to identify biomarkers of disease risk and progression and to potentially help characterize the pathogenesis of T1D [2–4]. Developing diagnostic or prognostic tools based on peripheral blood gene expression is particularly appealing since blood is easily obtained and requires minimal processing. PBC RNA is considered stable for extended periods of time when stored in either PAXgene or Tempus RNA tubes . These tubes contain reagents that stabilize RNA and prevent induction or degradation of transcripts that normally would occur within minutes of blood collection in conventional tubes [5–7]. PAXgene tubes (used by TEDDY, DAISY, and DIPP) contain tetradecyl trimethyl-ammonium oxalate and tartaric, while Tempus tubes (used by TrialNet) contain detergent and guanidine. Both solutions immediately lyse cells, inactivate RNases, and precipitate RNA [2, 4, 8–10]. RNA can then be extracted using kits that are optimized for each collection tube . Previous studies have shown that PBC gene expression can be influenced by the choice of collection tube and/or the RNA extraction system used [12–16]. However, it is unclear which system most accurately reflects the true gene expression profile and if sample-processing-related differences might mask the true gene expression profile or result in artifactual changes in gene expression. It is also unclear if sample-processing dependent differences in gene expression could be minimized by omitting the RNA extraction step. This is possible by measuring gene expression directly from whole blood lysate using NanoString arrays. We believe these represent important issues for any gene expression profiling study utilizing both Tempus and PAXgene-processed samples. Here, we examined the impact of sample collection and processing on PBC gene expression analysis in T1D.
RNA Yield (ng/ml)
Normal controls in PAXgene and Tempus tubes for RNA extraction
TrialNet type 1 diabetes patient samples in Tempus tubes
GAD65, ICA512, mIAA, ICA
GAD65, ICA512, mIAA
ICA512, mIAA, ICA
GAD65, ICA512, mIAA
GAD65, ICA512, mIAA
ICA512, mIAA, ICA
GAD65, ICA512, mIAA
University of Florida type 1 diabetes patient samples in PAXgene tubes
GADA, IA-2, ZnT8
GADA, IA-2, ZnT8
GADA, IA-2, ZnT8
Normal controls for RNA extraction and lysate preparation
RNA Yield (ng/ml)
Sample collection and RNA extraction
To compare gene expression in PAXgene vs. Tempus-processed samples, whole blood of 9 healthy individuals was collected into both PAXgene (Preanalytix) and Tempus (Applied Biosystems) tubes (Table 1 and Fig. 1). Total RNA was extracted using the PAXgene Blood RNA Kit (Qiagen) or the Tempus Spin RNA Isolation Kit (Applied Biosystems), as appropriate, according to manufacturer’s instructions. TrialNet T1D (TN-T1D; Table 1) samples were collected in Tempus tubes and RNA was extracted by TrialNet using the KingFisher Purification system (Thermo Scientific). University of Florida T1D samples (UF-T1D; Table 1 and Fig. 1) were collected in PAXgene tubes and isolated using the PAXgene Blood RNA Kit (Qiagen). Total RNA concentrations were determined using the Nanodrop (Thermo Fisher) and RNA quality was assessed using the 2100 Bioanalyzer and the RNA 6000 Nano Reagent Kit (Agilent).
Whole blood lysate preparation
To compare gene expression in matching samples of total RNA and whole blood lysate, 4 blood samples were collected from each of 11 control subjects, 2 into PAXgene and 2 into Tempus tubes (Table 1 and Fig. 1). RNA was extracted from a single PAXgene and Tempus tube as described above. Whole blood lysates were prepared from the remaining 2 tubes as follows: PAXgene samples were pelleted, washed and dried according to manufacturer’s instructions, and resuspended in 150 μl Buffer RLT (Qiagen) containing β-mercaptoethanol. Tempus samples were diluted in PBS and pelleted according to manufacturer’s instructions. RNA pellets were dried and resuspended in 50 μl of the same buffer.
One color microarray analysis was performed at the Stanford Human Immune Monitoring Center using the SurePrint G3 Human Gene expression 8×60K one-color microarray kit and the Low Input QuickAmp Labeling Kit (Agilent). This system utilizes OligodT-promoter primer and T7 RNA polymerase to generate cRNA, and produces reliable and consistent gene expression results without globin removal, a procedure that substantially reduces the quantity and quality of extracted RNA [13, 14]. Total RNA (300 ng) containing 2 μl of spike-in control was amplified, fluorescently labeled, and hybridized to the microarray chips according to manufacturer’s instructions. Data were processed with Feature Extraction Software (version 12.0, Agilent), quantile-normalized and analyzed using GeneSpring GX 12.6 (Agilent). To compare gene expression between PAXgene and Tempus RNA samples, samples were filtered for detected entities (detected in ≥5 of 9 samples, Table 1, Control 1–9). Entities representing 17,843 unique genes that were detected using both collection tubes were further analyzed for differential gene expression by performing a moderated T-test with Benjamini-Hochberg multiple testing correction (P < 0.05) with a fold-change cut-off of 2. Hierarchical clustering (Euclidean Distance Metric, Ward’s Linkage Rule) was performed using GeneSpring GX 12.6.
To compare TN-T1D samples against Tempus controls (Control 1–9), gender-specific genes (see Additional file 1) were first omitted due to the imbalance of male to female subjects in the two groups. Gender-specific genes were identified by comparing male to female samples in the Tempus control and TN-T1D groups (Moderated T-test, Benjamini-Hochberg multiple testing correction, P < 0.05). Differentially expressed genes in TN-T1D vs. healthy Tempus controls were identified, and statistical analysis and hierarchical clustering were performed as described above. Enrichment of differentially expressed genes for B-cell, T-cell, granulocyte, and lymphocyte signature genes was examined using previously published lists of cell-type specific genes . Significant enrichment was determined using the Chi-squared test with Yates’ correction (two-tailed, P < 0.05). All microarray data are available at NCBI Gene Expression Omnibus (GEO) Database (GSE89021 and GSE89022).
First strand cDNA was generated using Superscript III Supermix, containing a mixture of random hexamers and Oligo(dT)20 (Invitrogen). QPCR was performed using the 7900HT Fast Real Time PCR System, Taqman Gene Expression Mastermix and Taqman gene expression assays (all Applied Biosystems) as previously described . Genes that were changed by at least 2-fold in PAXgene vs. Tempus RNA samples (COX6C, COX7B, COMMD6, LSM3, RPS24, UQCRB and RPL31), and in TN-T1D vs. Tempus controls (FASLG, FCRL6, GZMB, and KLRD1) were measured, along with GUSB, a stably expressed housekeeping gene . QPCR data were normalized using the housekeeping gene ACTB. The comparative Ct method (ΔΔCt) was used for relative quantification, and statistical analysis was performed using the Wilcoxon-matched pairs test or the Mann-Whitney test, where appropriate (P < 0.05).
NanoString arrays using total RNA and blood lysate
NanoString Probe Sequences
Gene expression in samples processed using the PAXgene versus tempus system
The 901 genes affected by sample processing were also significantly enriched for genes that have previously been reported to be changed in the PBC of recent onset T1D patients and at-risk individuals: 26 genes (29%) overlapped with the 90 genes (present on our array) found by Reynier and colleagues  to be altered in the PBC of autoantibody-positive first-degree relatives of T1D patients and T1D patients at clinical onset compared to healthy controls (P < 0.0001, two-tailed Chi-square with Yates’ correction, Fig. 3d). 54 genes (12%) overlapped with the 460 genes (present on our array) found by Kallionpaa and colleagues  to be changed in the PBC of individuals who later progressed to T1D compared to controls (P < 0.0001, two-tailed Chi-square with Yates’ correction, Fig. 3d). These overlapping genes include lymphocyte-signature genes COX6C, COX7B, LSM3, RPS24, RPL31, RPS3A and UQCRB (Fig. 3a and d ). 7 genes overlapped in all 3 datasets (COMMD6, COX7B, RPS24, SUB2, RPS3A, SCYL2 and RPS27L).
Differential expression of sample-processing-dependent genes in the peripheral blood of established T1D patients
Comparison of gene expression in blood lysate and extracted RNA
Comparison of the PAXgene and Tempus blood lysate samples showed that the expression of COX6C, COX7B, COMMD6, LSM3, and SUB1 was not significantly different (Fig. 7a-d, f, white bars). This suggests that there is selective loss of these gene transcripts during the PAXgene RNA extraction procedure. RPS24, KIF20B and KIAA1841 expression was still significantly different between PAXgene and Tempus whole blood lysates (Fig. 7e, g-h, white bars). However, the fold change differences were drastically lower. KIF20B and KIAA1841 were approximately 11-fold and 14-fold lower in PAXgene vs. Tempus RNA samples, but only 4.1-fold and 3.8-fold lower in PAXgene vs. Tempus whole blood lysates, respectively (Fig. 7j). This suggests that a large amount of KIF20B and KIAA1841 transcripts may have already degraded in the PAXgene collection tube prior to RNA extraction. However, some amount of KIF20B and KIAA1841 was also lost during RNA extraction. Expression of RPS24 was ~20% lower in PAXgene vs. Tempus lysate, and was 4-fold higher in PAXgene lysate vs. PAXgene RNA, suggesting that most of this transcript is lost during RNA extraction, and only a small amount was degraded in the PAXgene tube (Fig. 7e, j).
ZNF680 was the only gene examined where expression was higher in PAXgene than in Tempus RNA samples (Fig. 7i). ZNF680 was also higher in PAXgene vs. Tempus lysate samples, suggesting that this transcript may not be as well preserved in Tempus collection tubes. Expression was only slightly higher in Tempus blood lysates compared to Tempus RNA samples, indicating only slight loss (<20%) of ZNF680 during the Tempus RNA isolation procedure. Interestingly, a significant increase of ZNF680 (~40%) was observed in PAXgene RNA vs. PAXgene lysate samples. The reason for this is unclear, but it is possible that loss of high expressing transcripts, such as ribosomal proteins (Fig. 3b), in the PAXgene RNA samples may result in the relative enrichment of other genes.
These data demonstrate that differences in gene expression seen in PAXgene vs. Tempus-processed samples are mainly due to loss of transcripts during the RNA extraction step using the PAXgene system. This indicates that the gene expression profile of Tempus-processed samples more accurately reflects the true gene expression profile, and that sample-processing dependent changes in gene expression could be minimized by measuring gene expression directly in the whole blood lysate.
We observed that approximately 5% of genes expressed in PBC are affected by sample processing, and that a significant number of these overlapped with genes that have previously been reported to change in PBC of at-risk AA+ first-degree relatives of T1D patients, and recent onset T1D patients compared to controls [3, 4]. Using samples from established T1D patients and healthy controls, we showed that combining PAXgene and Tempus-processed samples affected the gene expression profile and resulted in misleading changes in gene expression. We showed that differences in PBC gene expression can result from differences in transcript preservation in PAXgene or Tempus tubes, but were mainly due to loss or incomplete recovery of transcripts when RNA was extracted using the PAXgene system. Thus, the gene expression profiles of Tempus-processed RNA samples or minimally-processed whole blood lysates more accurately reflect the actual transcriptome than those of PAXgene-processed RNA.
This information is exceedingly important because gene expression profiling of peripheral blood in high-risk T1D-related individuals can potentially lead to the identification of diagnostic biomarkers of disease risk and progression – information that may provide insight into disease pathogenesis. Even using multi-center T1D consortia that have been established to collect biological samples from different cohorts of at-risk individuals, samples are frequently limited or collected from a particular demographic. Therefore, it is beneficial for researchers to combine data on samples from different sources.
Samples collected in the same type of collection tubes may also be processed differently between different consortia. For example, the TrialNet T1D samples used in this study were collected in Tempus tubes and processed using the KingFisher Purification system, an automated system that utilizes magnetic beads for RNA separation and purification. The Tempus Spin RNA isolation kit utilizes a spin column containing a silica membrane for RNA binding and elution. Because the spin columns are not optimal for retaining smaller transcripts, it is likely that these smaller transcripts are under-represented in samples extracted using the Tempus Spin RNA Isolation kit, and thus may have a different gene expression profile compared to samples processed using the KingFisher system. Our findings, however, do show that the majority of genes that are affected by processing using the Tempus Spin RNA isolation kit compared to the PAXgene blood RNA isolation kit do not significantly differ between KingFisher-processed TN-T1D samples and the Tempus system-processed control samples (Figs. 5b and 6; Set 2 genes).
Previous studies using a small number of samples (n = 3), or on a small number of genes, suggested that the choice of collection tube influenced gene expression [11, 13, 20, 21]. In this study, we demonstrate using 9 matched samples, that the collection tube and RNA extraction step affected the expression of ~900 genes. Some differences resulted from incomplete preservation of transcripts by the PAXgene or Tempus RNA-stabilizing reagent in the collection tubes, but the majority of differences are due to loss of certain mRNA transcripts when RNA was extracted using the PAXgene Blood RNA Kit.
RNA yield and quality have been reported to differ between PAXgene and Tempus samples [5, 11, 13, 16]. However, we and others observed little difference between the two systems (Table 1, [12, 14, 20]). These differences should not significantly alter relative gene expression levels, assuming genes of interest and housekeeping genes are uniformly affected. We found that the majority of genes (95%) were not significantly changed by sample processing. This is consistent with published studies showing little difference in the expression of select genes (MMP9, ARG1) in PAXgene vs. Tempus-processed RNA samples . We showed that the expression of most housekeeping genes was not significantly affected by sample processing (Fig. 3c). However, RNA18S5, B2M and TFRC did fluctuate by ~1.5 to 4-fold in Tempus vs. PAXgene samples, and could significantly influence gene expression in PAXgene vs. Tempus samples if used for data normalization. Previous work has shown that the expression of SIGLEC-7 measured in PAXgene and Tempus samples differs depending on the housekeeping gene used for normalization .
Nine hundred one genes were significantly affected by sample processing using the PAXgene vs. Tempus system. Most were downregulated in PAXgene samples and were highly enriched in ribosomal protein and lymphocyte-signature genes (Fig. 3, ). Surprisingly, the affected genes were also enriched for genes that are changed in the blood of at-risk AA+ first-degree relatives of T1D patients, recent onset [3, 4] or established T1D patients (Figs. 3d and 5). A number of ribosomal and oxidative phosphorylation pathway genes that are affected also overlapped with genes that are changed in the PBC of obese individuals and those suffering from metabolic syndrome [22, 23].
A similar study by Nikula et al. comparing PBC gene expression in PAXgene vs. Tempus samples showed differences in the expression of 443 genes . 264 of these are present in our array, and 50 overlapped with our list of differentially expressed genes (see Additional file 4). Interestingly, Nikula et al. showed an enrichment of genes involved in RNA processing and immune cell function among the changed genes, and demonstrated that the gene expression profile of PBMCs is more similar to that of blood collected in Tempus than PAXgene tubes . This supports our observation that lymphocyte-signature genes are lower in PAXgene samples. Another study showed that immune-related genes were upregulated in PHA-stimulated vs. control samples collected in Tempus, but not PAXgene tubes . These include IFNG and IL4, which are known to respond to PHA stimulation. We found that IFNG and IL4 were significantly lower in RNA prepared using the PAXgene system (see Additional file 5). Thus, PHA-induced changes seen in PAXgene-processed samples may not be as robust as those seen in Tempus-processed RNA samples.
By comparing gene expression in matching samples of whole blood lysate and RNA extracted from Tempus and PAXgene collection tubes, we asked if differences in gene expression were due to degradation of transcripts in the collection tube or loss of transcripts during the RNA extraction procedure. NanoString arrays were used to directly quantify mRNA transcripts in whole blood lysates and RNA samples without an amplification step, which may artificially distort gene expression levels [24, 25]. NanoString data were highly comparable to microarray and QPCR data and resulted in the least amount of variation among samples (Fig. 4). By comparing gene expression in whole blood lysates and total RNA of samples prepared in Tempus and PAXgene tubes, we showed that the Tempus Spin RNA Isolation Kit had little effect on gene expression. However, RNA extraction using the PAXgene Blood RNA Kit resulted in lower expression of multiple lymphocyte-signature genes and genes that reportedly are changed in the PBC of at-risk or recent-onset T1D patients (Figs. 3 and 7a-f [3, 4]). Previous studies have shown that a number of these genes including COX6C, COX7B, and RPS24 (Fig. 2) were also reduced in samples prepared using the PAXgene system vs. the QIAamp system, which lyses cells in a buffer similar to the reagent contained in Tempus tubes . Several genes (KIF20B, KIAA1841 and ZNF680) were found to be better preserved in either PAXgene or Tempus tubes (Fig. 7g-i). Since PAXgene and Tempus tubes are highly acidic and basic, respectively, it is possible that certain gene transcripts are more stable in acidic or alkaline conditions .
It is unclear why RNA extraction of PAXgene samples leads to differences in gene expression compared to Tempus samples. Both RNA extraction kits utilize a silica membrane-based column for RNA binding and purification. However, optimal binding requires high pH and salt conditions . It is possible that residual tartaric acid present in the PAXgene samples affect RNA binding to the column. This may explain the lower RNA yields previously observed in PAXgene vs. Tempus samples, but it is unclear why certain gene transcripts were more affected than others [5, 11, 13, 16].
Our findings show that ~5% of genes expressed in PBCs are significantly affected by sample processing using the Tempus versus the PAXgene system. If the specific genes of interest are known, the most accurate method of quantification is direct measurement in blood lysates using NanoString arrays. This procedure uses the same volume of blood (collected in either Tempus or PAXgene tubes), does not require RNA extraction or gene amplification and results in the most consistent gene expression profiles between samples collected in PAXgene and Tempus tubes. For microarray, RNAseq, or QPCR experiments, where purified RNA is required, samples collected and processed using the Tempus system are recommended, especially for studies of immune function and T1D. Incomplete recovery or degradation of many more gene transcripts were observed using the PAXgene than the Tempus system, and the genes affected were highly enriched in genes that are reported to be dysregulated in the PBC of at-risk, recent onset, and established T1D patients [3, 4]. These differences need to be considered in any study, including meta-analysis studies  that combine expression data from PAXgene and Tempus samples.
This work was supported by a grant from the Juvenile Diabetes Research Foundation (to C.G.F). L.Y. was supported by a grant from the Myra Reinhard Family Foundation. Peripheral blood RNA samples from established T1D patients were provided by TrialNet through the Type 1 Diabetes TrialNet GAD Study. Currently TrialNet is funded by National Institutes of Health (NIH) through the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Allergy and Infectious Diseases, and The Eunice Kennedy Shriver National Institute of Child Health and Human Development, through grant numbers: U01 DK061010, U01 DK061034, U01 DK061042, U01 DK061058, U01 DK085465, U01 DK085453, U01 DK085461, U01 DK085463, U01 DK085466, U01 DK085499, U01 DK085504, U01 DK085505, U01 DK085509, U01 DK103180, U01-DK103153, U01-DK085476, U01-DK103266 and the Juvenile Diabetes Research Foundation International (JDRF). Samples from The University of Florida were collected with support from NIH NIAID 42288.
Availability of data and materials
All microarray data are available at NCBI Gene Expression Omnibus (GEO) Database (GSE89021 and GSE89022).
LY planned and directed this work, recruited healthy controls, prepared the manuscript and figures, and analyzed the data. RF recruited healthy controls, isolated RNA, performed bioanalyzer analysis of RNA, prepared whole blood lysates, performed QPCR, analyzed the data and helped prepare the manuscript. MAA recruited and collected samples from study subjects, and contributed to the notions of data interpretation. CGF collected samples from healthy controls and was involved in the direction of this work. All authors reviewed and edited the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Approval was obtained from the Stanford University Institutional Review Board (IRB) or the University of Florida IRB. For TrialNet samples (Study TN08), IRB approval was obtained at the institution where samples were collected. All participants provided written informed consent.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Atkinson MA. The pathogenesis and natural history of type 1 diabetes. Cold Spring Harbor perspectives in medicine. 2012;2(11).Google Scholar
- Jin Y, Sharma A, Bai S, Davis C, Liu H, Hopkins D, et al. Risk of type 1 diabetes progression in islet autoantibody-positive children can be further stratified using expression patterns of multiple genes implicated in peripheral blood lymphocyte activation and function. Diabetes. 2014;63(7):2506–15.View ArticlePubMedPubMed CentralGoogle Scholar
- Reynier F, Pachot A, Paye M, Xu Q, Turrel-Davin F, Petit F, et al. Specific gene expression signature associated with development of autoimmune type-I diabetes using whole-blood microarray analysis. Genes Immun. 2010;11(3):269–78.View ArticlePubMedGoogle Scholar
- Kallionpaa H, Elo LL, Laajala E, Mykkanen J, Ricano-Ponce I, Vaarma M, et al. Innate immune activity is detected prior to seroconversion in children with HLA-conferred type 1 diabetes susceptibility. Diabetes. 2014;63(7):2402–14.View ArticlePubMedGoogle Scholar
- Duale N, Brunborg G, Ronningen KS, Briese T, Aarem J, Aas KK, et al. Human blood RNA stabilization in samples collected and transported for a large biobank. BMC Res Notes. 2012;5:510.Google Scholar
- Franken C, Remy S, Lambrechts N, Hollanders K, Den Hond E, Schoeters G. Peripheral blood collection: the first step towards gene expression profiling. Biomarkers. 2016;21(5):458–65.Google Scholar
- Hebels DG, Georgiadis P, Keun HC, Athersuch TJ, Vineis P, Vermeulen R, et al. Performance in omics analyses of blood samples in long-term storage: opportunities for the exploitation of existing biobanks in environmental health research. Environ Health Perspect. 2013;121(4):480–7.PubMedPubMed CentralGoogle Scholar
- Bair R, Paulsen K. Compositions and methods for purifying nucleic acids from stabilization reagents. Google Patents; 2006.Google Scholar
- Prezeau N, Silvy M, Gabert J, Picard C. Assessment of a new RNA stabilizing reagent (tempus blood RNA) for minimal residual disease in onco-hematology using the EAC protocol. Leuk Res. 2006;30(5):569–74.View ArticlePubMedGoogle Scholar
- Vehik K, Fiske SW, Logan CA, Agardh D, Cilio CM, Hagopian W, et al. Methods, quality control and specimen management in an international multicentre investigation of type 1 diabetes: TEDDY. Diabetes Metab Res Rev. 2013;29(7):557–67.PubMedPubMed CentralGoogle Scholar
- Meyer A, Paroni F, Gunther K, Dharmadhikari G, Ahrens W, Kelm S, et al. Evaluation of existing methods for human blood mRNA isolation and analysis for large studies. PLoS One. 2016;11(8):e0161778.View ArticlePubMedPubMed CentralGoogle Scholar
- Bayatti N, Cooper-Knock J, Bury JJ, Wyles M, Heath PR, Kirby J, et al. Comparison of blood RNA extraction methods used for gene expression profiling in amyotrophic lateral sclerosis. PLoS One. 2014;9(1):e87508.View ArticlePubMedPubMed CentralGoogle Scholar
- Nikula T, Mykkanen J, Simell O, Lahesmaa R. Genome-wide comparison of two RNA-stabilizing reagents for transcriptional profiling of peripheral blood. Transl Res. 2013;161(3):181–8.Google Scholar
- Menke A, Rex-Haffner M, Klengel T, Binder EB, Mehta D. Peripheral blood gene expression: it all boils down to the RNA collection tubes. BMC Res Notes. 2012;5:1.Google Scholar
- Asare AL, Kolchinsky SA, Gao Z, Wang R, Raddassi K, Bourcier K, et al. Differential gene expression profiles are dependent upon method of peripheral blood collection and RNA isolation. BMC Genomics. 2008;9:474.View ArticlePubMedPubMed CentralGoogle Scholar
- Matheson LA, Duong TT, Rosenberg AM, Yeung RS. Assessment of sample collection and storage methods for multicenter immunologic research in children. J Immunol Methods. 2008;339(1):82–9.View ArticlePubMedGoogle Scholar
- Palmer C, Diehn M, Alizadeh AA, Brown PO. Cell-type specific gene expression profiles of leukocytes in human peripheral blood. BMC Genomics. 2006;7:115.View ArticlePubMedPubMed CentralGoogle Scholar
- Yip L, Fuhlbrigge R, Taylor C, Creusot RJ, Nishikawa-Matsumura T, Whiting CC, et al. Inflammation and hyperglycemia mediate Deaf1 splicing in the pancreatic lymph nodes via distinct pathways during type 1 diabetes. Diabetes. 2015;64(2):604–17.View ArticlePubMedGoogle Scholar
- Zampieri M, Ciccarone F, Guastafierro T, Bacalini MG, Calabrese R, Moreno-Villanueva M, et al. Validation of suitable internal control genes for expression studies in aging. Mech Ageing Dev. 2010;131(2):89–95.View ArticlePubMedGoogle Scholar
- Hantzsch M, Tolios A, Beutner F, Nagel D, Thiery J, Teupser D, et al. Comparison of whole blood RNA preservation tubes and novel generation RNA extraction kits for analysis of mRNA and MiRNA profiles. PLoS One. 2014;9(12):e113298.View ArticlePubMedPubMed CentralGoogle Scholar
- Vartanian K, Slottke R, Johnstone T, Casale A, Planck SR, Choi D, et al. Gene expression profiling of whole blood: comparison of target preparation methods for accurate and reproducible microarray analysis. BMC Genomics. 2009;10:2.View ArticlePubMedPubMed CentralGoogle Scholar
- Ghosh S, Dent R, Harper ME, Gorman SA, Stuart JS, McPherson R. Gene expression profiling in whole blood identifies distinct biological pathways associated with obesity. BMC Med Genet. 2010;3:56.Google Scholar
- Tangen SE, Tsinajinnie D, Nunez M, Shaibi GQ, Mandarino LJ, Coletta DK. Whole blood gene expression profiles in insulin resistant Latinos with the metabolic syndrome. PLoS One. 2013;8(12):e84002.View ArticlePubMedPubMed CentralGoogle Scholar
- Nygaard V, Loland A, Holden M, Langaas M, Rue H, Liu F, et al. Effects of mRNA amplification on gene expression ratios in cDNA experiments estimated by analysis of variance. BMC Genomics. 2003;4(1):11.View ArticlePubMedPubMed CentralGoogle Scholar
- Puskas LG, Zvara A, Hackler L Jr, Van Hummelen P. RNA amplification results in reproducible microarray data with slight ratio bias. BioTechniques. 2002;32(6):1330. -4, 6, 8, 40PubMedGoogle Scholar
- Debey S, Schoenbeck U, Hellmich M, Gathof BS, Pillai R, Zander T, et al. Comparison of different isolation techniques prior 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(3):193–207.View ArticlePubMedGoogle Scholar
- Karim Z, Attmane-Elakeb A, Sibella V, Bichara M. Acid pH increases the stability of BSC1/NKCC2 mRNA in the medullary thick ascending limb. J Am Soc Nephrol. 2003;14(9):2229–36.Google Scholar
- Gjerse D, Hoang L, Hornby D. RNA purification and analysis: sample preparation, extraction, chromatography. 1st ed. Weinheim: Wiley; 2009.Google Scholar
- Peters MJ, Joehanes R, Pilling LC, Schurmann C, Conneely KN, Powell J, et al. The transcriptional landscape of age in human peripheral blood. Nat Commun. 2015;6:8570.View ArticlePubMedPubMed CentralGoogle Scholar