Automated serial extraction of DNA and RNA from biobanked tissue specimens

Tissue collection and preparation

864 tumor-normal paired colorectal and breast frozen tissue samples (288 colorectal, 576 breast) as well as 30 samples of liver, prostate, tonsil, colon, breast, thymus, kidney, skin, uterus and lung were obtained from the frozen tissue collection at the Department of Pathology, Academic Hospital Uppsala. This study was approved by the Regional Ethical Review Board of Uppsala (2007/116) and written consent was obtained from participants. The tissues were embedded in OCT and stored at −80°C to maintain biomolecular integrity [1]. The breast and colon tumor sections contained a minimum of 50% and 40% tumor cells, respectively. The blocks were sectioned and 2 or 3 10 μm sections per specimen were collected in 2D barcoded tubes in tube racks of 96 (Micronic Roborack-96, art. no. MPW51016BC3).

Automated serial extraction of nucleic acids

We recently described a novel process for serial DNA and RNA extraction employing silica beads with differential nucleic acid binding affinities [5]. This extraction procedure was automated on a liquid handling workstation (Tecan Evo 150 MCA LiHa RoMa) equipped with wash stations for 96 and 8 tips, respectively, a twin-block heater with two different constant temperatures (EchoTherm IC22, Torrey Pines Scientific), and readers for 1D plate barcodes (Symbol MS954) and 2D tube barcodes (Ziath), respectively. Briefly, nine 96-well plates of approximately 25 mg fresh frozen tissue from 864 patient-matched tumor and normal tissues (288 colorectal, 576 breast) were collected as described. Unless otherwise stated, all liquid transfers were performed with a 200 μL fixed tip block (Tecan). At the start of each run, all reagents along with one SBS format tube rack with 96 samples were loaded on the robotic workstation and uncapped. The lysis buffer was dispensed using an 8-channel LiHa pipetting head. After addition of chaotropic lysis buffer, the samples were incubated for 15 min at 58°C, followed by incubation with DNA binding beads for 15 min. All liquid handling after the initial dispensing of lysis buffer to the tissue samples was performed using a 96-tip MCA pipetting head with tip washes between each process step.

The DNA binding beads were captured using a magnetic plate (V&P Scientific) and the supernatant transferred into a new vessel for RNA capture. Meanwhile, the DNA selective beads were washed three times in wash buffer and bound DNA eluted in TE buffer. After 15 min binding, the RNA binding beads were retrieved using a magnetic plate and washed first in DNase (ThermoScientific) containing wash buffer and thereafter washed and eluted in the same buffer composition as used for the DNA extraction [5]. The final DNA and RNA products were transferred to a 96 well Roborack barcoded storage plate (Micronic, Article No MPW51016BC3). The worktable layout is shown in Additional file 1: Figure S1.

Quality control of extracted DNA and RNA

The DNA yields from the tissue samples were assessed by measurements using a High Sensitivity dsDNA kit on a Qubit® instrument (Invitrogen). The purity of the DNA was assessed by spectrophotometry (OD 260:280 ratio) using a Nanodrop instrument (Thermo Scientific). The integrity of DNA was assessed by separation in a 0.7% agarose gel (Sigma Aldrich) and staining with SYBR Safe (Invitrogen). The integrity of selected RNA samples was assessed using an RNA 6000 Pico Assay on an Agilent 2100 Bioanalyzer instrument (Agilent). The samples were diluted 1:10 or 1:20 in RNase free water and denatured for 2 min at 70°C before separation. The 28S/18S ribosomal RNA ratio and RNA integrity (RIN) scores were computed using the Agilent Technologies 2100 Expert software package.

Performance evaluation in genomic analyses

The extracted DNA was used in several downstream applications including PCR-based STR analysis, deep sequencing on an Illumina platform following a HaloPlex selection (Agilent), Sanger sequencing and gene copy number analyses.

PCR-based STR analysis was performed on 238 colorectal DNA samples (119 tumor/normal pairs). Briefly, 24 STR markers in regions showing loss of heterozygosity in cancer were amplified using a touchdown PCR protocol. PCR amplification was carried out using 2.5 ng of genomic DNA as template. The primers were each conjugated to one of the 3 fluorophores FAM, NED, or VIC (Sigma-Aldrich, Applied Biosystems). PCR was performed in 10 μL reactions containing 1 × PCR buffer (67 mM Tris–HCl, pH 8.8, 6.7 mM MgCl₂, 16.6 mM NH₄SO₄, 10 mM 2-mercaptoethanol), 1 mM dNTPs, 1 μM forward and 1 μM reverse primers, 6% DMSO, 2 mM ATP, 0.25 U Platinum Taq (Invitrogen) and 2.5 ng DNA. Reactions were carried out in 96-well ABI 2720 thermocyclers using a touchdown PCR protocol (1 cycle of 96°C for 2 min; 3 cycles of 96°C for 10 sec, 64°C for 10 sec, 70°C for 30 sec; 3 cycles of 96°C for 10 sec, 61°C for 10 sec, 70°C for 30 sec; 3 cycles of 96°C for10 sec, 58°C for 10 sec, 70°C for 30 sec; 41 cycles of 96°C for 10 sec, 57°C for 10 sec, 70°C for 30 sec; 1 cycle of 70°C for 5 min). Fluorescently labeled PCR products were analyzed by fragment analysis in a capillary sequencing instrument (ABI PRISM 3730xl) using ROX500 (Applied Biosystems) as size standard followed by allele identification using GeneMapper Software v4.1 (Applied Biosystems).

Haloplex target enrichment for second-generation sequencing (Agilent) of 540 genes potentially implicated in colorectal cancer was performed on 400–800 ng DNA from 192 colorectal samples (96 tumor/normal pairs) according to the manufacturer’s instructions [7]. The enriched and barcoded targets were then deep sequenced on an Illumina next generation sequencing platform (Illumina) [8]. Sanger sequencing of the PCR products amplified for mutation validation was carried out by an initial touchdown PCR protocol as described above, using the 192 samples previously deep sequenced on an Illumina platform as DNA template. Following this, 18 μL reactions were prepared containing 20 ng PCR product template and 4 pmol M13 primer (Biomers). The sequence reactions were delineated at Uppsala Genome Center on an ABI PRISM 3730xl sequencing apparatus (Applied Biosystems).

Gene copy number analyses of 70 of the colon cancer samples were performed using Genome Wide SNP6 microarrays (Affymetrix), according to the manufacturer’s instructions.

Automated biomolecule extraction from tissues

The serial DNA/RNA extraction method described in [5] was successfully implemented in a fully automated fashion on a Tecan Freedom Evo workstation for parallel extraction of 96 samples or 1–96 samples using 8-channel liquid handling. A tissue biobank workflow for integration of the extraction process is outlined in Figure 1. By omitting the grinding and shaking step of the original procedure, the number of transfers and the amount of hardware required in the robotic platform were reduced while maintaining yields and biomolecule integrity (data not shown). The total run time for extraction of 96 tissue samples was 1 h 40 min or 3 h 20 min when DNA alone or DNA and RNA, respectively, was recovered. The dropout rate (no DNA in eluate) for the initial 576 samples was ~ 2%, and the dropouts were likely caused by tissue clogging of the narrow bore fixed pipette tips. The remaining 288 samples were extracted using a wider bore 96 tip block (Tecan, art. no. 10290619), and there were no sample dropouts in this set.

Serial recovery of high quality DNA and RNA from tissue specimens

The mean DNA yield of all samples was 3.2 μg (SD = 0.08, n = 576) as measured by a Nanodrop instrument and 0.9 μg (SD = 0.17) as measured by Qubit. The mean 260:280 ratio, used as a measure of protein contamination, was 1.66. The median 260:230 ratio, used as a secondary measurement of purity and indicating the presence of salts and other contaminants, was 1.57. The integrity of four randomly selected DNA samples is shown in Figure 2, where long strands of gDNA can be seen on a 1% agarose gel stained with SYBR Safe. Yields and purity (assessed spectrophometrically) of breast and colon DNA samples are presented in Table 1.

Storage stability of extracted RNA

The integrity of eleven randomly chosen RNA samples were measured using an RNA 6000 Pico assay. The RNA quality was found to be acceptable for use in most downstream processes, with an average RIN value across 11 samples of 6.85 at time zero. To assess storage stability, the samples were stored at −70°C and assessed at 3 months (average RIN = 6.69), 6 months (average RIN = 6.86) and 9 months (average RIN = 6.45), demonstrating stability under these conditions even with several freeze-thaw cycles. Peak profiles from each time point are shown in Figure 3. We have previously demonstrated the successful use of RNA from this extraction chemistry in RT-PCR [5].

Conventional and next-generation sequencing analyses of genomic DNA

The extracted DNA was first used in PCR-based short tandem repeat (STR) analysis to compare genomic loci between tumor/normal matched samples to ensure that they were correctly paired. This allowed the detection of loss of heterozygosity in chromosomally unstable samples, as well as revealing that two of the paired samples were mismatched (Additional file 2: Figure S2).

Next, 400 – 800 ng DNA (as measured by Qubit) from 96 pairs of matched tumor and normal colorectal tissue (192 samples in total) was used in a Haloplex target enrichment and barcoding protocol (Agilent) and sequenced on an Illumina platform to identify driver mutations and key pathways involved in colorectal cancer (CrC) (Mathot et al., in progress). Plots of the fraction of regions of interest covered by at least a certain read depth, illustrating the quality of the data produced by the NGS technology for both the normal and tumor samples are shown in Figure 4. The mean read depth was approximately 500 for the normal samples and 1000 for the tumor samples.

Identification of genomic aberrations using Genome Wide SNP6 arrays (Affymetrix) was also carried out on 70 of the CrC samples. The mean sample quality (QC) value as calculated by Nexus Copy Number™ software from Biodiscovery (measuring the probe to probe variance) was 0.19 (SD = 0.05). A QC value in the range 0.15 - 0.20 is considered high quality, with higher quality samples approaching 0.2.