Identification of blood genomic biomarkers of efficacy or toxicity in small preclinical species, especially rodents, offers great potential for translation to the clinic. However, the sample collection, RNA isolation procedures, and microarray processing methods need to be optimized to small amount of whole blood in order to make such research activity practical in animal models. The present study describes a practical and efficient workflow using a QSI isolation method coupled with NuGEN Ovation Whole Blood Solution for blood transcriptomic profiling from as little as 25 μL whole blood, a volume easily obtained from a standard tail vein bleeding procedure. A comparison with the standard PAXgene system demonstrated that the QSI method can produce similar yields of total RNA per volume of rat whole blood. The total RNA samples produced from the QSI and the standard PAXgene methods have an average RIN score of 9, well above the RNA quality generally preferred for microarray-based profiling (RIN > 8) [¹⁰,²⁰,²¹].

It is possible to modify the standard PAXgene method to isolate RNA from small amounts of whole blood. The method by Krawiec et al. [²¹] yielded an average of 40 ng total RNA per μL whole blood with a RIN score of 7.7 from 50 μL mouse whole blood. The method by Robison et al. [²²] yielded an average of 3.6 ng total RNA per μL whole blood with a RIN score of 9.3 from 70 μL whole human blood collected via fingerstick. The low yield from this later study is likely a reflection of the difference between human and rodent blood. In general, the RNA yield and quality were comparable to the QSI method reported here. However, both modified PAXgene methods are manual and require extra steps to achieve sufficient yield. In contrast, the QSI method can be fully automated. Using the Qiagen Automated BioRobot 3000 RNeasy-96 RNA isolation protocol, only 90 min of hands-on time is needed to isolate RNA from 96 whole blood samples. The QSI method is also cheaper than the PAXgene method. The total cost for 96 samples using the automated QSI method is a quarter of the cost of the standard PAXgene method and half the cost of the modified PAXgene method.

Commercial RNA isolation and stabilization kits that are designed specifically for laboratory animals have also been recently developed from several resources. The ZR Whole Blood Total RNA Kit (Zymo Research, Orange, CA, USA), the Mouse RiboPure Blood RNA Isolation Kit (Ambion/Applied Biosystems, Austin, TX, USA), and the RNeasy Protect Animal Blood System (Qiagen, Valencia, CA, USA) are designed for blood volumes of 100–500 μL. Whether these kits can be further scaled down to 25 μL whole blood remains to be determined.

Of note, neither the PAXgene system, nor the QSI method removes globin mRNA during the RNA isolation process. To reduce the artifacts associated with highly abundant globin mRNA transcripts, the NuGEN Ovation Whole Blood Solution procedure was used downstream of the total RNA isolation from rat whole blood. Results with human blood samples suggested that globin RNA amplification can be reduced with the NuGEN procedure [¹⁷]. However, in our experience with rat whole blood, there was no significant reduction of globin peak in cDNA samples when compared to cRNA samples prepared with the standard Affymetrix protocol. Similar results were also noted by the manufacturer (personal communication). Rat blood has twice the amount of reticulocytes compared to human blood [²³]. The difference in globin reduction performance between human vs. rat blood is likely a reflection of the high proportion of reticulocytes in rat blood. The NuGEN procedure generates cDNA targets as compared to the cRNA targets prepared with the standard Affymetrix protocol. Despite the pronounced globin peak in cDNA targets from rat blood, the procedure is less prone to non-specific cross-hybridization with globin transcripts as a result of higher fidelity of DNA-DNA hybridization compared to RNA-DNA hybridization [²⁴]. In addition to the reduced interference with globin mRNA, the NuGEN procedure requires a relatively small amount (20–50 ng) of total RNA for transcriptomic profiling. The QSI method described here can easily yield over 1000 ng total RNA from 25 uL blood, an amount sufficient for several microarray experiments using the NuGEN procedure.

A comparison of RT-PCR and microarray results showed good concordance of gene expression changes for a panel of 9 transcripts selected to represent low, mid, and high abundance genes. There are differences in the absolute fold change measured by RT-PCR and microarray. However, the trend (ie., up- or down-regulation) was the same across all abundance levels. The concordance in trend, but not in absolute value, is a typical observation between RT-PCR and any type of microarray platform regardless of sample resources [²⁵,²⁶]. Although this is only a small sample size relative to the whole genome, these results, taken together with the prototypical LPS-induced pathway changes in the blood transcriptome, suggested that the NuGEN procedure can provide an accurate representation of transcripts across different abundance levels.

The microarray performance metrics showed that the array quality using RNA prepared from the PAXgene method was slightly superior to that from RNA isolated with the QSI method. However, the QC parameters for all samples were within the acceptable range recommended by the manufacturer and the general microarray community [²⁷–³⁰].

We used LPS-induced acute immune response in rats as a model system to address if the slight difference in QC parameters between the two methods affected the representation of gene transcripts and biological interpretation. LPS is a component of the bacterial cell wall of Gram negative bacteria and activates a complex of pattern recognition proteins in a variety of mammalian cell types, especially macrophages [³¹]. The interaction with the protein complex results in activation of a downstream signaling cascade that eventually leads to the production of pro-inflammatory cytokines, as well as the recruitment of inflammatory cells. In human whole blood, LPS induces transcriptomic changes characterized by increased expression of genes associated with the defense response to pathogens, such as cytokines, chemokines, and acute-phase transcription factors, and decreased expression of genes associated with lymphocytes and ribosomal proteins [³²]. In the present study, the rat blood transcriptomic profiles revealed a pattern consistent with the underlying LPS-induced acute immune response, regardless of whether the blood total RNA was isolated with the PAXgene or the QSI method.

Male Sprague-Dawley rats [Crl:CD^®(SD)IGS BR] weighing approximately 250 g were obtained from Charles River Laboratories, Inc, Portage, MI, USA. The animals were permitted non-certified Rodent Chow and water ad libitum. Rats were housed two or three per cage for two days after receipt to aid in acclimation.

Thereafter, rats were single housed in ventilated, stainless steel, wire bottom hanging cages equipped with feeders and an automatic watering system. The animals were administered a single intravenous injection of vehicle (saline) or LPS (5 mg/kg, Sigma-Aldrich # L2880) and euthanized 2, 6, or 24 hrs post dosing (3 rats/time point/group). Blood samples were collected via the abdominal vaudal vena cava at necropsy for RNA isolation (Figure 6). For the PAXgene method, 2.5 mL of blood were collected into PAXgene Blood RNA Tubes according to the manufacturer’s instructions (BD Biosciences, San Jose, CA). After incubation for 2 hrs at room temperature, the PAXgene tubes were stored at −20 °C until analysis. For the QSI method, approximately 25 μL, 50 μL, 100 μL, 200 μL, and 500 μL of blood were added to 1 mL or 2 mL (for 500 μL blood volume only) of QIAzol (QIAGEN, Valencia, CA). Upon addition of the blood, the tubes were inverted twice and incubated at room temperature for up to 3 hrs before storage at −80°C for future analysis. Experiments were conducted in accordance with the Guiding Principles in the Use of Animals in Toxicology [³³] and approved by the local Institutional Animal Care and Use Committee.

RNA isolation from samples collected in PAXgene Blood RNA Tubes was performed according to the manufacturer’s specifications (BD Biosciences). For the QSI method, total RNA was isolated from 750 μL of blood-QIAzol lysate following the Qiagen Automated BioRobot 3000 RNeasy-96 RNA isolation protocol (QIAGEN). After manual centrifugation of a 96-well plate containing QIAzol lysate/chloroform mixture, the Qiagen BioRobot 3000 automatically transfers the upper, aqueous phase to a 96-well RNeasy plate for automated RNA purification. Nucleic acid concentration was determined by O.D. 260 nm (NanoDrop ND-1000, Thermo Scientific, Wilmington, DE, USA). The RNA integrity was evaluated and an RNA Integrity Number (RIN) was generated using an Agilent 2100 Bioanalyzer and its accompanying software (Agilent Technologies, Foster City, CA, USA).

All blood RNA samples were processed to generate microarray hybridization targets using the Ovation Whole Blood Solution (NuGEN Technologies, San Carlos, CA, USA), a module containing three NuGEN products. Briefly, 50 ng of DNAase-treated RNA aliquots were amplified using the Ovation RNA Amplification System V2 coupled with the Ovation WB Reagent following the manufacturer’s instructions (NuGEN). Approximately 4.4 μg of each amplified single strand cDNA sample was then biotinylated and fragmented using the FL-Ovation cDNA Biotin Module V2. The resulting cDNA was hybridized to an Affymetrix Rat Genome 230 2.0 Array (Affymetrix, Santa Clara, CA, USA) using the NuGEN-modified hybridization protocol, with 2 min of heat denaturation of the hybridization cocktail at 99 °C and a minimum of 16 hrs of hybridization at 45 °C using an Affymetrix Hybridization Oven 640. The array was subsequently washed and stained with streptavidin-phycoerythrin (Molecular Probes) on the Affymetrix Gene-Chip Fluidics Workstation 400 following the NuGEN-recommended protocol EukGE-WS2v4_450. The array was then scanned using an Affymetrix GeneChip Scanner 3000.

The following primers and probes were selected from the Applied Biosystems Assays-on-Demand gene expression products: Tcf7 (Rn_00493446_m1), Map2k6 (Rn_00586764_m1), Nrp1 (Rn_00595457_m1), Alas2 (Rn_00566201_m1), Flad1 (Rn_01512487_m1), Canx (Rn_00596877_m1), S100a9 (Rn0585879_m1), Tnf (Rn_99999017_m1), Il1b (Rn_01514151_m1), and 18S (Hs_99999901_s1). Expression levels of each gene were quantified by real-time RT-PCR with an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Carlsbad, CA, USA). Briefly, 200 ng of total RNA was reverse transcribed using the iScript^TM cDNA Synthesis Kit according to the manufacturer’s instructions (Bio-Rad Laboratories). Real-time PCR analyses were then performed in a total volume of 10 μL containing 1:15-diluted synthesized cDNA using Taqman^® Gene Expression Assays (Applied Biosystems). The PCR cycling profile was as follows: 10 min at 95 °C, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The relative fold changes were calculated using the delta delta Ct method and normalized to 18S ribosomal RNA, the endogenous control gene.

All non-microarray statistical analyses were conducted using JMP 7.0.1 statistical software from SAS (Cary, North Carolina). Comparisons were made using the two-tailed t-test with a significance value of α = 0.05. For microarray analysis, the scanned image and intensity files were imported into Rosetta Resolver gene expression analysis software version 7.2 (Rosetta Inpharmatics, Seattle, WA). Individual gene expression ratios were built for each treatment animal versus the averaged vehicle controls at the corresponding conditions using the Rosetta Resolver error model [³⁴]. Genes and experimental treatments were grouped for visualization by agglomerative hierarchical clustering using the Pearson correlation as described in each figure. For pathway analysis, the average gene expression ratios were calculated as the in silico pool of treated animals versus the corresponding vehicle controls using the Rosetta Resolver error model. Genes with a fold change over 2 and a p-value < 0.01 were mapped to pathways and biological functions using Ingenuity Pathway Analysis (Ingenuity Systems, www.ingenuity.com).