Our experience with using the AcroPrep™ Advance 96-well filter plates for purification of DNA samples prior to next generation sequencing (NGS), in comparison with traditional organic extraction and ethanol precipitation methods, was that:
- Handling was simplified.
- Overall processing time was reduced.
- More samples could be processed simultaneously.
- More consistent results were obtained.
Introduction
As costs for NGS continue to decrease, new possibilities exist for genomic analysis. Where there is a need to analyze hundreds of individuals, significant time and expense can be incurred for purifying individual DNAs and then producing bar-coded libraries from each. Several years ago, because of the costs and technology for producing bar-coded libraries, we used bulked segregants for comparing single nucleotide polymorphism frequencies in samples based on phenotype and gender (1, 2). New systems for production of bar-coded libraries have been developed which are technically simpler, less expensive, and faster, therefore allowing NGS of individuals rather than the need for “bulking” of DNAs. One impediment remaining however, is the time, expense, effort, and reliability of production of DNAs sufficiently pure for NGS library generation.
We experimented with various high-throughput methods for purification including 96-well ion-exchange capture and paramagnetic bead capture, but both methods were highly unreliable or problematic because the samples contained high molecular weight DNA which either did not pass through the resin or could not be easily eluted. The most reliable method was by organic extraction which is time consuming and employs hazardous chemicals. Since the centrifugation steps are in individual microfuge tubes, they must be processed in batches of 10-24 samples at a time with many pipetting steps. Often, a number of sample extractions fail to produce sufficient quantity after the ethanol precipitation and must be repeated. Here we examine a purification method employing 96-well filter plates as an alternate to organic extraction. In our hands, the filter plate-based method proved cost effective, provided consistent results, and allowed processing of 96 samples in less than two hours.
MATERIALS AND METHODS
Preparation of initial DNA extracts
Crude DNA lysates were prepared by the method of Bailes et al. (3). Briefly, 10 µL of whole chicken blood was pipetted from a 28-gauge lancet puncture of a wing vein and triturated into 400 µL of cold 64 mM sucrose, 20 mM Tris-Cl pH 7.5, 10 mM MgCl2, 0.5% nonionic detergent in a 96 well assay block on ice. After all samples were collected, the block was sealed with a silicone mat lid, inverted several times, then centrifuged at 1000 × g for five min. The plate was inverted to discard the supernatant. The nuclear pellets were resuspended by trituration into TEN+Pronase (10 mM Tris-Cl pH 8.0, 1 mM EDTA, 10 mM NaCl, 100 μg/mL Pronase E). The sealed block was incubated at 37°C for 60 min with shaking at an approximate 45-degree angle, then incubated in a 65°C water bath for 15 min. Assay blocks were stored at -20°C in a non-frost-free freezer. At the conclusion of the chicken work, crude DNA samples from birds in particular phenotypic groups were identified. For these samples, 100 µL of the crude DNA lysate was mixed with 100 µL TEN+Pronase, then incubated at 37°C for 60 min with shaking at an approximate 45-degree angle. The block was then incubated at 65°C in a water bath for 10 min.
Purification by organic extraction
The pronase-treated, heat-inactivated DNA solution was successively extracted with phenol/chloroform/isoamyl alcohol (50:48:2), then phenol/isoamyl alcohol (24:1). The final aqueous solution was mixed with 20 µL 3 M NaOAc pH 5.3, then 600 µL 95% -20°C ethanol, centrifuged at 10000 × g for 15 min, decanted, pellet rinsed with -20°C 70% ethanol, centrifuged, decanted, and dried by centrifugal evaporation. The pellet was redissolved overnight at 4°C in 30 µL Te (10 mM Tris Cl, 0.1 mM EDTA, pH 7.5).
Purification by centrifugal filtration
Stacks each consisting of two filter plates and an assay block were prepared for centrifugation consisting, from top to bottom, of an AcroPrep™ Advance 96-well filter plate with 3.0 µm glass fiber/1.2 µm Supor™ membrane (1.2 µm filter plate), an AcroPrep Advance filter plate for ultrafiltration with Omega™ 100K MWCO membrane (100K MWCO filter plate), and a 0.5 mL assay block used to collect the final flow through. The stacking of the plates in this fashion allowed clarification of the pronase-treated, heat-inactivated DNA solution and retention of the DNA on top of the ultrafiltration membrane in a single centrifugation step. Care was taken to avoid touching the bottom surface of the plate to prevent contamination of samples. The pronase-treated, heat-inactivated DNA solutions were pipetted into the corresponding wells. The stack was centrifuged at 1500 × g for 10 min at 5°C. The counterbalance was a stack of used filter plates with water added to the top plate for proper balance. After centrifugation, the stack was visually inspected to be certain nearly all liquid has passed through the filter plates. If not, the centrifugation was repeated. The flow-through in the assay block was discarded. Te (200 µL) was added to each well in the top plate, and the stack centrifuged as above. The Te rinse was performed a total of two times. The top 1.2 µm filter plate was discarded and 100 µL of Te was added to the wells of the 100K MWCO filter plate sitting on the emptied waste collection block. The 100K MWCO filter plate was covered with a lid and the stack was placed on a platform orbital shaker at 500 rpm for 10 min at room temperature to resuspend the DNA from the MWCO filter. The resuspended DNA was transferred to a 96-well raised-lip PCR plate. The PCR plate was sealed with a 96-well silicone mat lid and stored frozen at -20°C until ready to make the libraries.
Library construction purified
DNAs were quantified by Hoechst fluorescence on a GloMaxu-Multi Junior instrument (Promega, Madison, WI). Bar-coded library construction employed 4 µL of each purified DNA using the RIPTIDEu kit (iGenomX, South San Francisco, CA) with a 3:1 ratio of the AT-rich primers to GC-rich primers. Pooled bar-coded libraries were sequenced on Illumina HiSeq Xu or NovaSequ platforms by Admera Health Systems (South Plainfield, NJ) or Novogene (Davis, CA). DNA data was demultiplexed and reads quantified using fgbio 0.8.1 (http://fulcrumgenomics. github.io/fgbio/) in a Docker container.
Results and discussion
In three different experiments, chickens were phenotyped for particular traits and individuals identified from the extremes of the phenotype. DNAs from those individuals were either purified individually by organic extraction (organic extractions 1 and 2), or in a 96-sample batch by the Cytiva filtration method (centrifugal filtration). For the organic extractions, 12 of 48 and 15 of 48 samples (organic extraction 1 and 2, respectively) required new extractions as the first extraction failed to yield a sufficient quantity (>12.5 ng/µL). With the Cytiva filtration method, none of the 96 samples had less than 24 ng/µL. Evaluation of the NGS yield (total gigabases) for organic extraction 1 or 2 vs Cytiva filtration method (Figure 1) reveals that NGS yield is not a function of the total amount of input DNA (see linear trend lines). Specifically note that samples with the lowest total yield were seen in organic extraction 2 which had the highest range of total input. Rather, the variation may relate to quality of individual DNAs or impurities.
Fig 1. Total gigabases of Illumina sequence data yield is plotted versus total input DNA for bar-coded library synthesis. Left and middle panels are for two different sets of 48 samples purified by organic extraction. Right panel is for 96 samples purified by the Cytiva filtration method. Dotted lines in each panel represent a linear trend-line.
As there were 7-8 samples with low NGS yield from the Cytiva filtration method, we analyzed the purified DNA from seven low yield samples relative to two with high NGS yields. The DNAs were resolved by agarose gel and the lane fluorescence graphed for comparison (Fig 2). The scan data show that all of the DNAs had the same distribution of DNA size range with the peak around 1500-2000 base pairs. All seven of the low-yield samples had a lower total fluorescence than the two high-yield samples (marked G01 and F09), indicating less total DNA even though all samples were matched for quantity by Hoechst 33258 fluorescence (GloMax Jr; Promega Corp; Madison, WI).
However, the NGS yield did not seem to be proportional to the amount of DNA. For example, samples E11 and F10 appear to have the third and fourth highest fluorescence profile below G01 and F09 but had the approximate same NGS yield as sample E10 which had the lowest overall fluorescence profile. Therefore, the low yield for some samples may be affected by DNA quantity but is more likely to relate to either DNA quality, or random technical errors (e.g., pipetting mistakes with multichannel pipettors) with some samples.
Fig 2. Gel analysis of selected DNAs to compare samples with different total bases from NGS libraries. Samples are listed in the legend along with the total gigabases of NGS data in parentheses. For each sample 200 ng of DNA was electrophoresed in a 0.7% agarose gel stained with ethidium bromide and scanned (λex = 532 nm and λem = 600 nm). Fluorescence was quantified for each lane using ImageJ and graph rendered in MicroSoft Excel. Dotted lines were low total yield and solid lines were high total yield. Molecular weights (base pairs) were based on an Axygen 1 kb DNA molecular weight marker.
In our experience, the Cytiva filtration method provided advantages over the organic extraction method. Whereas the organic extractions were often done in sets of 12 or 18 samples with each set requiring 4-5 h, in our experience the Cytiva filtration method allowed for purification of 96 samples in under 2 h. For organic extraction, each sample had to be carefully transferred individually after each extraction. The Cytiva filtration method on the other hand, allowed the use of multi-channel pipettors to load the samples, add the rinse solution, and recover the cleaned DNA.
We experienced a significant failure rate with the samples obtained by organic extraction which could have resulted from loss of DNA during the extraction, during ethanol precipitation, or during vacuum drying. Using the Cytiva filtration method, we did not experience any failures and recovered sufficient DNA from all 96 samples for the library production.
Conclusions
Our experience with using the AcroPrep Advance 96-well filter plates for purification of DNA samples prior to NGS sequencing, in comparison with traditional organic extraction and ethanol precipitation methods, was that:
- Handling was simplified.
- Overall processing time was reduced.
- More samples could be processed simultaneously.
- More consistent results were obtained.
References
- Parveen A, Jackson C, Dey S, Tarrant KJ, Anthony NB, Rhoads DD. Identification and validation of quantitative trait loci for ascites in broilers using whole genome resequencing. BMC Genet 2020, 21:54.
- Dey S, Parveen A, Tarrant KJ, Li cknack T, Kong B, Anthony NB, Rhoads DD: Whole Genome Resequencing Identifies the CPQ Gene as a Determinant of Ascites Syndrome in Broilers. PLoS One 2018, 13(1):e0189544.
- Bailes S, Devers J, Kirby JD, Rhoads D: An inexpensive, simple protocol for DNA isolation from blood for high-throughput genotyping by polymerase chain reaction or restriction endonuclease digestion. Poult Sci 2007, 86:102-106.
*Note added by the University of Arkansas: Nothing in this case study should be taken as an endorsement of any product or representation of results that may be obtained.