Isolation of DNA from small amounts of elephant ivory ...

Forensic Science International 288 (2018) 131–139

Contents lists available at ScienceDirect

Forensic Science International journal homepage: www.elsevier.com/locate/forsciint

Isolation of DNA from small amounts of elephant ivory: Sampling the cementum with total demineralization extraction M. Winters, A. Torkelson, R. Booth, C. Mailand, Y. Hoareau, S. Tucker, S.K. Wasser* Center for Conservation Biology, Department of Biology, University of Washington, Seattle, WA, USA

A R T I C L E I N F O

Article history: Received 17 February 2018 Received in revised form 18 April 2018 Accepted 20 April 2018 Available online 27 April 2018 Keywords: DNA extraction Elephant ivory Poaching Low–template DNA Demineralization

A B S T R A C T

Genotyping ivory samples can determine the geographic origin of poached ivory as well as the legality of ivory being sold in ivory markets. We conducted a series of experiments to determine where the DNA is most concentrated in ivory samples and how best to increase DNA yield from groups of samples likely to vary in DNA concentration. We examined variation in DNA amplification success from: the layer(s) of the tusk (cementum and/or dentine) being extracted, demineralization temperature and time, and the concentration of eluates. Since demineralization of the pulverized sample produces a pellet and supernatant, we also assessed DNA amplification success from the pellet, the supernatant, their combination, as well as variation in the respective amounts used for extraction. Our results show that the outer cementum layer of the tusk contains the highest concentration of DNA and should be separated and used exclusively as the source material of ivory processed for extraction, when available. Utilizing the combined demineralized lysate improves extraction efficiency, as does increasing demineralization time to 3 or more days, conducted at 4 C. The most significant improvements occurred for low template DNA ivory samples followed by medium quality samples. Amplification success of high quality samples was not affected by these changes. Application of this optimized method to 3068 ivory samples resulted in 81.2% of samples being confirmed for both alleles at a minimum of 10 out of 16 microsatellite loci, which is our threshold for inclusion in DNA assignment analyses. © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction The illegal ivory trade has had a devastating impact on African elephant populations, with the number of poached elephants rapidly increasing since 2006 [1–3]. DNA analyses of large ivory seizures (0.5 tonnes) has already proven to be an important tool to combat illegal trade at the supply end by identifying the major poaching hotspots across Africa [2,4]. DNA analyses in conjunction with Carbon-14 analyses can also help combat illegal trade by establishing whether large scale poaching events have occurred recently, [5] as well as identifying the legality of ivory markets. The latter supports the decision reached at the 17th meeting of the Conference of the Parties for the Convention on International Trade in Endangered Species (CITES) to close ivory markets that are contributing to poaching and illegal trade. DNA amplification success is vital to all of these efforts as it affects both the cost and the number of samples that must be analyzed to acquire

* Corresponding author at: Center for Conservation Biology, University of Washington, Department of Biology, Box 351800, Seattle, WA, 98195-1800, USA. E-mail address: [email protected] (S.K. Wasser).

meaningful results. We, accordingly, conducted a series of experiments to optimize DNA amplification success of both whole and worked (carved) ivory samples. Ivory tusks are similar to teeth, but lack an enamel layer. Tusks are composed of an outer layer of cementum with a dentine core [6]. The cementum is a thin (>1–2 mm) layer that is present along the entire outer surface and is softer than the more mineralized dentine that makes up the majority of the tusk. Since the cementum layer is the most concentrated source of mitochondrial and nuclear DNA in human teeth [7–10], we assessed whether amplification success of nuclear DNA is highest when DNA is extracted from the cementum, dentine, or combined layers of the tusk. The first step in DNA extraction of bones and teeth is demineralization with ethylenediamine-tetraacetic acid (EDTA). Demineralization liberates the DNA from the sample matrix and chelates metal ions such as Mg2+ and Ca2+. Methods vary in the time and temperature of demineralization and may also include additional agents such as Proteinase K (ProK) and detergents [11,12] with temperatures ranging from 4 C [13], 37 C [14], or 55 C to 56 C [15,16] occurring over one or multiple days [17]. Typically, the EDTA supernatant is removed after demineralization

https://doi.org/10.1016/j.forsciint.2018.04.036 0379-0738/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

132

M. Winters et al. / Forensic Science International 288 (2018) 131–139

and the sample powder is then digested in a lysis buffer [11,18,19] which eliminates a source of available DNA [18]. We tested whether complete demineralization of the pellet and supernatant together would improve DNA amplification success [15,16] as this utilizes all sources of DNA. We compared the DNA amplification success from extraction of the pellet, the supernatant, or both combined in a modified complete demineralization (digestion performed after initial incubation in EDTA only). We also tested effects of demineralization at temperatures of 4 C or warmer, and for incubation periods ranging from 3 to 10 days with varying amounts of added ProK. Lastly, we tested additional methods that could be applied to low template DNA (LT-DNA; i.e. from degraded, worked or aged tusks) and whole ivory samples. These included concentrating by pooling replicates from the same sample since some filter-based concentrating methods may incur significant DNA loss [20,21], and conducting additional PCR amplifications. 2. Methods 2.1. Samples All experiments utilized samples from seized elephant ivory tusks previously genotyped in our laboratory for 16 microsatellite markers [22–24] to determine sample origin analysis [2]. Samples in these experiments were divided into three quality categories (good, medium and poor) based on prior DNA amplification success. Good samples had consistent amplification of 20–32 out of the 32 possible alleles (homozygotes counted as two alleles), medium samples had semi-consistent amplification of 8–19 alleles, and poor samples had amplification of less than 8 alleles. Extraction negatives were processed for each variable being tested in all experiments beginning at the demineralization stage. 2.2. Pre-treatment of samples One to three pieces of ivory were cut from the hollow base of the whole tusk ranging in size (3–5 cm2) and thickness (0.3–2 cm) using an electric circular saw with a fine-toothed blade before shipment. Samples were immersed in a 20% dimethyl sulfoxide/ tris/NaCl/EDTA buffer (DMSO pH 10) for at least 2 h, as per USDA protocol to reduce risk of transmitting hoof and mouth disease. Upon arrival to the laboratory, samples were rinsed under running water to remove dirt and residual buffer then air dried overnight. The thickest portion of ivory was then cut into multiple 1 cm2 pieces using a band saw. A longer cross-section (1 3 cm2) was cut for experiments separating the cementum and dentine layers

prior to extraction (see Tusk sampling source and Extended demineralization experiments). Following Mailand and Wasser [13], all ivory pieces were decontaminated by submersion in 5% Clorox1 bleach (0.3% sodium hypochlorite) given the contamination risk from tusks coming into contact with people and tusks from other elephants. This method provided decontamination without impacting DNA amplification (data not shown), contrary to earlier suggestions that DNA loss results from decontamination using 5% bleach [9,25]. Samples were submerged in 5% bleach for 30–120 s, while being wiped with gloved fingers to remove dirt, residual bone/tissue and/or any powder that may have been transferred from other samples during cutting. The pieces were next rinsed twice with sterile water and allowed to air dry for at least 24 h prior to pulverization. Roughly 0.6–0.8 g of ivory pieces were then pulverized while submerged in liquid nitrogen using a SPEX 6770 Freezer Mill (SPEX SamplePrep, Metuchen, New Jersey) with an 80 s run time at 10 cps and a 1 min cool time. Pulverization vials were allowed to return to room temperature (RT; 15 C–25 C or average of 20 C) before transferring the ivory powder to 5 mL storage tubes. The variables we examined and variations in experimental protocol are summarized in Table 1 in chronological order, whereby the best outcome was utilized in the next experiment. 2.3. DNA extraction source material To determine the optimal portion of the demineralized sample to be extracted, we compared the DNA amplification success from extracting the pellet, the EDTA supernatant, and the combined product (pellet and supernatant together) following demineralization of a 200 mg piece of pulverized ivory. We also examined how DNA amplification success varied when using 1/3 aliquots of the supernatant or combined product so that all extraction steps could still be completed in a 2 mL tube to preserve high processing throughput. Pulverized ivory from 15 samples (5 from each quality category) were used for each group of this experiment. We additionally tested whether DNA amplification success improved by doubling the amount of ProK used during cell digestion for the pellet, supernatant aliquot and whole lysate groups only. All samples were demineralized in 1.6 mL of 0.5 M (pH 7.5) EDTA and rotated at 4 C for 3 nights. The pellet group was centrifuged for 10 min at 15,600 g prior to the removal of the corresponding EDTA supernatant. One or two additions of 0.4 mg of ProK (Promega) was added to each sample, combined with 360 mL of Buffer ATL (Qiagen) for the pellet group. The supernatant and combined product groups instead had 20% sodium dodecyl

Table 1 Variables and variations in experimental protocols. Experiments are depicted in chronological order (top to bottom), each using good, medium and poor quality samples, according to the main variable being tested. The best outcome per stage (i.e. row) was used in the following experiment, but expanded with a new variable (with the exception of concentration). For example, the highest yielding DNA extraction source was the 1/3 combined lysate with 0.4 mg of ProK, eluted in 570 mL of AE. This combination was then tested in the following experiment but at additional demineralization temperatures. Tusk sampling source and demineralization time are grouped because they were tested simultaneously with a different set of samples than the former rows. Variations in the concentration of ProK were tested within the first experiment only. The volume used for extraction and elution reagent changed as we maximized the amount of lysate material and accounted for potential concentration of sample eluates. Whole = cementum and dentine cross section cut; AE = Buffer AE (Qiagen); RT = room temperature 15 C–25 C. Experiment

DNA extraction source

Tusk source(s)

Whole

Demineralization

Extraction source (s)

Time(s)

Temp

3 nights

4 C

Pellet Supernatant Combined 1 /3 Supernatant 1 /3 Combined Demineralization temperature Whole 3 nights 4 C, RT, 37 C 1/3 Combined 1 Concentration Whole 10 nights 4 C /3 Combined 1 Tusk sampling source and demineralization time Cementum and whole 3, 5 and 10 nights 4 C /3 Combined

ProK (mg)

1/3 volume Elution

0.4 and 0.8 0.4 0.4 and 0.8 0.4 and 0.8 0.4 0.4 0.4 0.4

– – – 450 mL 570 mL 570 mL 600 mL 600 mL

AE AE AE AE AE AE Water 1:10 AE


sulfate (SDS; final conc. 2%) added to maintain appropriate detergent concentrations without dramatically increasing volume. The groups with two ProK additions were incubated at 56 C for 3 h before the second addition of 0.4 mg. All samples were then incubated at 56 C overnight. After centrifuging at 15,600 g for 1 min, a 450 mL aliquot was taken from a second replicate of the supernatant to generate the 1/3 supernatant group, and 570 mL was taken from a second replicate of the combined product to generate the 1/3 combined lysate group. Heated (56 C) Buffer AL (Qiagen) and 100% ethanol was added to each sample group lysate in a 1:1:1 ratio then vortexed for 10–15 s. The samples were then purified using either a 96-well Whatman GF/F filter Microplate (SigmaAldrich) or a DNeasy Mini Spin column (Qiagen), with the DNA being eluted in two passes of heated (56 C) 100 mL Buffer AE (Qiagen) for a final eluate of 200 mL. 2.4. Demineralization temperature The same samples (additional replicates) from the previous experiment were demineralized at three temperatures, 4 C, 20 C (RT), and 37 C with rotation for 3 nights then extracted using the 1 /3 combined lysate method described in the first experiment (single addition of 0.4 mg ProK). 2.5. Concentration of replicates Using the same samples from the previous experiments, extractions were conducted in duplicate for the 1/3 combined lysate method as described (single addition of 0.4 mg ProK), but with demineralization extended to 10 nights and at 4 C only. Two 600 mL (30 mL more) aliquots were drawn from the same combined lysate but purified separately, and the second lysate was eluted with RNase-free, distilled water (from Qiagen Multiplex PCR kit) instead of Buffer AE to reduce the amount of salts present in the final eluate, which can be PCR inhibitive [26]. The two eluates were then combined and placed in a Speed Vac Concentrator (Savant) at 45 C for 20 h until dry. The concentrated samples were finally re-suspended in 100 mL of Buffer AE. 2.6. Tusk sampling source Pulverized ivory from 24 samples (8 from each quality category) were used for both groups of this experiment. Pieces were cut for both cementum (thin outer layer) and whole tusk (cementum and dentine cross section) with attempts to keep the thickness of cementum consistent for both cuts. The cementum and whole tusk pieces were extracted in duplicate using the 1/3 combined lysate method from the first experiment. Demineralization was additionally tested at five and ten nights (described below) and samples were purified using the DNeasy 96-well plate (Qiagen). One combined lysate aliquot of 600 mL was taken with heated (56 C) 1:10 diluted Buffer AE used for elution. 2.7. Extended demineralization time For each tusk in the previous experiment, powder from both the cementum and whole cross-section cuts were demineralized for 3, 5, and 10 nights at 4 C only, then extracted using the 1/3 combined lysate method just described. 2.8. Microsatellite typing and analysis Given the large number of samples across our experimental groups, we chose to measure results based on amplification success at each of three representative loci (FH126, FH40 and FH102 [22,23]) out of 16. The three loci were chosen to represent

133

the range of our allele sizes (84–264 bp) and amplification efficiencies (FH126 being most efficient and FH40 being the least), while also cost effective since these loci are able to be PCR amplified in a single multiplex reaction. These loci are also typically multiplexed together and accurately represent the first step in the 16 loci analysis for ivory seizures. PCR was conducted in 10 mL reactions containing 1X Multiple Master Mix (Qiagen), BSA (1 mg/mL), 0.25 mM each of six primers, and 2 mL of sample DNA. Following a ten minute hot-start at 95 C, the reactions cycled 40 times at 94 C for 30 s, annealed at 58 C for 1.5 min, and extended at 72 C for 1 min. A final extension was done for 40 min at 60 C. Two known positive controls and one PCR blank were amplified with every PCR plate to monitor thermocycler and electrophoresis conditions. Amplified DNA fragments were separated using capillary electrophoresis on an ABI 3730 genetic analyzer (Applied Biosystems) calibrated for a DS-30 dye matrix (FAMTM, HEXTM, NEDTM and ROXTM). A portion of the amplified product (2 mL) was added to 9 mL of Hi-Di formamide mixed with ROX400 (1.48%), and injected for 8 s at 1.6 kV. The resulting data were genotyped by two analysts using GeneMarker v 2.4.0 (SoftGenetics) with a 1000 relative fluorescent units (RFU) inclusion threshold for all alleles at all three loci. For the purposes of genotyping, a sample is confirmed as a homozygote if an allele is observed at least three times, wherein a heterozygote is confirmed when both sister alleles are seen at least twice. 2.9. Determining DNA amplification success of experimental variables For all experiments, DNA amplification success was analyzed using generalized linear models in JMP (v. 9.0) with number of alleles per PCR as the dependent measure, independent of locus, for each of the three screening loci. Each sample contributed up to 12 separate entries of the dependent variable: two PCRs for each of one or two extracts per sample, for each of three loci. Each entry was counted as a single observation with number of alleles per entry ranging from 0 to 2. This approach allowed each sample to serve as its own control, with the maximum number of alleles per PCR determined by whether each locus was homozygous (entry = 1) or heterozygous (entry = 2) for that sample, simultaneously accounting for allelic drop-out or presence of null alleles. We used the contrasts test in JMP to determine significant differences between categorical variables. 2.10. Seizure case study To determine the success rate and benefit of our method toward poor samples, we analyzed a large seizure where the tusks had been polished prior to shipment, such that no visual cementum remained. All samples were whole cut and therefore presumably reflect how worked and carved ivory would be processed. We extracted 200 samples in duplicate using the 1/3 combined lysate but employed methods shown to benefit poor quality samples including 10 nights of demineralization and concentrating four eluates into two (see step 13 of Fig. 4). We also tested the benefits of additional PCR amplifications for improving amplification success. After samples were amplified for all 16 microsatellite loci with the standard number of PCR’s (two per replicate or four total per sample), a subset of samples were chosen for additional PCR that had at least 5 confirmed loci and/or showed low heterozygosity after allele frequency analysis in Cervus 3.0.7 (Field Genetics). These samples were amplified another 3–5 times per replicate (10–14 total reactions per sample) depending on the amplification success rate and size (in bps) of the locus. The gain of confirmed loci and genotypes for each sample was compared in a step-wise format following the added information from subsequent amplifications of each replicate.

134


2.11. Application of new method to 20 large seizures We applied the 1/3 combined lysate extraction method to 20 large ( >0.5 ton) seizures, amounting to 3,068 samples in duplicate. A total of 945 samples were analyzed from whole cuts, 2123 samples analyzed from cementum cuts, and 306 samples concentrated due to poor amplification. 3. Results Sample replicates generated genotypes consistent with previous amplifications (data not shown), except in the case of allelic drop-out where only one of the two sister alleles was observed. Two samples had failed injections for a replicate during capillary electrophoresis and those injections were excluded from the analyses. None of the extraction negatives amplified. 3.1. DNA extraction source material DNA amplification success was affected by extraction source (p < 0.0001) and sample quality (p < 0.0001) for the five postdemineralization extraction sources: pellet, supernatant (EDTA after demineralization), combined product (demineralized pellet and EDTA supernatant), and 1/3 aliquots of the respective supernatant and combined product after lysis (1/3 combined lysate; Table 2). Amplification success did not significantly differ between the pellet and 1/3 supernatant (p = 0.92). However, amplification success was significantly lower for the entire supernatant compared to the pellet (p < 0.0001), 1/3 supernatant (p < 0.0001), and combined product (p < 0.0001; Table 3). The combined product had significantly higher DNA amplification success than did the pellet (p = 0.0001), supernatant (p < 0.0001), and 1/3 supernatant (p < 0.0001). The 1/3 combined lysate was comparable to the entire volume of the combined product, with no significant difference between the two sources (p = 0.61). There was also an interaction between extraction source and sample Table 2 Effect tests of experimental variables against the number of amplified alleles for all experiments. Analysis of variable interactions indicated with an ‘*’. Respective values for significance: NS = not significant; +significant at the 0.05 probability level; ++significant at the 0.01 probability level; +++significant at the 0.001 probability level. E. = extraction; T. = tusk; C = concentrated; NC = non-concentrated. Experiment

Effect test

P value

Extraction source

E. Source Quality E. source*Quality