Protocol: A high-throughput DNA extraction system suitable for conifers

Background High throughput DNA isolation from plants is a major bottleneck for most studies requiring large sample sizes. A variety of protocols have been developed for DNA isolation from plants. However, many species, including conifers, have high contents of secondary metabolites that interfere with the extraction process or the subsequent analysis steps. Here, we describe a procedure for high-throughput DNA isolation from conifers. Results We have developed a high-throughput DNA extraction protocol for conifers using an automated liquid handler and modifying the Qiagen MagAttract Plant Kit protocol. The modifications involve change to the buffer system and improving the protocol so that it almost doubles the number of samples processed per kit, which significantly reduces the overall costs. We describe two versions of the protocol: one for medium-throughput (MTP) and another for high-throughput (HTP) DNA isolation. The HTP version works from start to end in the industry-standard 96-well format, while the MTP version provides higher DNA yields per sample processed. We have successfully used the protocol for DNA extraction and genotyping of thousands of individuals of several spruce and a pine species. Conclusion A high-throughput system for DNA extraction from conifer needles and seeds has been developed and validated. The quality of the isolated DNA was comparable with that obtained from two commonly used methods: the silica-spin column and the classic CTAB protocol. Our protocol provides a fully automatable and cost effective solution for processing large numbers of conifer samples.


Introduction
Most genetic linkage mapping, marker-assisted selection, population and conservation genetic studies require processing of a large number of samples. Cost-effective high-throughput DNA extraction is a major bottleneck for many of these applications because handling and quantification of non-liquid samples, such as plant tissues, are labour-intensive and difficult to automate.
Traditional methods of DNA extraction involve multiple time-consuming steps, including organic solvent extraction and alcohol precipitation. Although commercially available spin column-based DNA preparation kits provide higher throughput, they are relatively expensive and difficult to automate. Recently-introduced magnetic fishing protocols allow for fully-automated, flexible throughput DNA isolation from certain samples, such as blood, but these methods are less tolerant to the secondary metabolites present in conifers and other plants [1]. Polysaccharides and phenolic compounds either impede DNA extraction or inhibit enzymatic reactions in the downstream applications. As a cheaper alternative to commercial DNA extraction kits, many labs use in-house developed simplified high throughput protocols, or use crude lysate as template for PCR [2,3]. However, majority of these protocols provide little quantitative information on the DNA yields, and were designed for crop species where the concentrations of PCR inhibitors are relatively low. Conifers require a more efficient DNA purification procedure. Although some protocols have been published for higher throughput DNA extraction from conifers [4,5], they still suffer from labour-intensive tissue grinding, high cost of silica columns or inconsistent DNA quality and yield.
Here we describe a cost-effective protocol based on the Qiagen MagAttract Plant DNA kit (QIAGEN Inc., Mississauga, Ontario) for high throughput DNA extraction from conifer needles and seeds. Most steps, including DNA quantification and normalization, can be done using an automated liquid handler.

Materials
Sample collection DNA was extracted from silica-dried needles of red spruce (Picea rubens). Needles were collected in the field into plastic bags containing 5-10 g silica gel pouches as desiccant, and then stored in a freezer at -20°C upon arrival to the laboratory. With silica gel, the plant material could be kept at the ambient temperature for up to 14 days without degradation of DNA quality. The protocol was also tested and operationally used on seeds and fresh needles of red spruce and black spruce (Picea mariana), as well as needles of white spruce (Picea glauca) and eastern white pine (Pinus strobus).

Protocol
There are two versions of the protocol. Protocol A is designed for medium throughput and provides higher DNA yields per sample processed, whereas Protocol B works in the high-throughput 96-well format from the initial tissue homogenization to the purified DNA at the end. NOTE: The original MagAttract buffer system didn't work well with conifers. The yield and quality of DNA were poor when original MagAttract buffers and protocol were used. The resulting DNA concentrations were less than 1 ng/μl and no PCR products could be obtained using these DNA extracts. We were able to overcome these problems by using AP1 buffer instead of RLT for lysis and ethanol instead of RB buffer for the DNA binding step.
We recommend the fluorimetric assay (e.g. PicoGreen by Invitrogen) as the best way to determine the DNA concentration in MagAttract-processed samples. Spectrophotometric (OD 260 ) measurements in 96-well plates tend to give unstable results due to the possible sample-to-sample variation in the liquid meniscus shape which in turn leads to biased pathlength correction. The resulting DNA con-centrations were 70 ± 15 ng/μl (~7 μg DNA/150 mg needle tissue) in Picea rubens, and 49 ± 13 ng/μl (~5 μg DNA/ 150 mg needle tissue) in Picea glauca. For subsequent genotyping, working plates containing 10 ng/μl DNA dilutions were prepared using an automated liquid handler. NOTE: We have eliminated the weighing step as one of the most time-consuming operations by using a few needles instead. Depending on the foliage size and shape, the amount of the plant material should be adjusted at the beginning to keep the dry weight of sample around 3-5 mg. The resulting DNA concentrations varied from 5.6 ± 2.0 ng/μl (~1 μg DNA/3-5 mg needle tissues) for Picea rubens to 9.7 ± 6.4 ng/μl for Pinus strobus, according to the fluorimetric PicoGreen assay. For Picea rubens and Picea mariana seeds, the DNA concentrations were 28.3 ± 13.8 ng/μl (~3.5 μg DNA/seed) and 6.2 ± 1.8 ng/μl (~0.8 μg DNA/seed), respectively. Seeds were processed as they the came from the storage without additional manipulations except for the visual inspection. Seed coats were not removed. 1.0-1.2 μl of eluates were used as templates for PCR.

Comments
We validated the performance of the resulting DNA preparations by multiplex PCR assays. Amplicons of expected size were obtained. Later on more than 3000 needle and seed samples of Picea rubens, Picea glauca, and Picea mariana were processed and genotyped in our lab using more than a dozen nuclear and chloroplast microsatellites (Figure 1). Reproducible assays were possible with 4-plex reactions. DNA isolation for one 96-well plate takes approximately 3.5 hours start to finish, and the use of a robotic liquid handler further reduces the hands-on time to about 40 min.

Comparison with other DNA extraction methods
To determine the reproducibility and efficiency of DNA isolation, we have compared our Protocol A-MTP with a slightly modified protocol using Qiagen DNeasy spin columns (Qiagen), and a version of widely used CTAB method [6]. Since some variability in the efficiency of mechanical tissue disruption is unavoidable due to the natural heterogeneity of plant material, we performed parallel DNA extractions using all three protocols in 8 replicates from the same bulked homogenized sample. Twenty grams of spruce needles were ground into fine powder under liquid nitrogen using mortar and pestle. 150 ± 2 mg of ground tissue were transferred into 2 ml Eppendorf tubes along with two 5 mm stainless steel balls normally used for tissue grinding. Then the DNA isolation was performed as follows: Modified MagAttract protocol As outlined above in the section Protocol A.
Qiagen DNeasy DNA was isolated according to the manufacturer's instructions, except for increased volumes of AP1 and AP2 buffers (700 μl and 250 μl per prep, respectively). Purified DNA was eluted into 150 μl of 10 mM Tris-HCl pH 7.5.
Microsatellite genotyping of 60 spruce trees with SSRs con-taining mono-, di-, and trinucleotide repeats Figure 1 Microsatellite genotyping of 60 spruce trees with SSRs containing mono-, di-, and trinucleotide repeats. A: Gel image of fragment patterns revealed by mononucleotide chloroplast-encoded microsatellite PT26081 in Picea rubens [7]. B: Gel image of fragment patterns revealed by dinucleotide repeat (RPGSE34) and trinucleotide repeat (RPGSE03) nuclear microsatellite loci in Picea rubens. The PCR was tetraplex with two other loci being genotyped in the second channel. C: Gel image of fragment patterns revealed by trinucleotide repeat (RPGSE02) nuclear microsatellite locus in Picea glauca.
All three methods resulted in high quality DNA (Figure 2) with A260/280 values in the range 1.69-1.85. The DNA yield varied greatly among the methods (Table 1). Significant variance of DNA concentration in our protocol can be attributed to relatively high pipetting errors when handling magnetic particles.

Conclusion
The CTAB method demonstrated the best yield and good reproducibility and probably remains the method of choice where large amounts of high quality DNA are required. For research projects requiring processing of larger sample sizes, our MagAttract-based protocol can be a flexible and cost-efficient solution (around $1 per sample, compared to $4 for spin columns). Medium-through-put version of our protocol allows for processing of large sample sets while maintaining higher DNA yield. The high-throughput variation is the best solution for genotyping projects where speed is the ultimate priority. It allows for processing up to 200 samples per person per day, in comparison to other protocols where expected output would be around 40-60 samples. The protocol works from start to end in the standard SBS 96 well microplate format and is fully automatable.