Design of the target and phylogenetic analyses

The complete genome of the mitochondria for 334 different species (Table S1) belonging to the Class Mammalia were downloaded from GenBank (http://www.ncbi.nlm.nih.gov/nuccore) and the DNA sequences for the 16S ribosomal RNA gene were extracted using the FeatureExtract 1.2 Server (http://www.cbs.dtu.dk/services/FeatureExtract). Multiple alignment was performed by the means of MAFFT (http://mafft.cbrc.jp) using the default settings. From the resulting alignment, the sequences for the universal PCR primers, designed previously [19], were extracted together with the DNA sequence between the forward and reverse primer sites (i.e. the target region). A distance matrix was computed by pairwise comparison of the DNA sequences between the primer sites for all 334 mammal species, using MEGA v. 5.1 [20]. In order to visualize the ability to separate different mammal species, a tree (UPGMA) was created using MEGA v. 5.1. The tree was constructed based on the total number of nucleotide differences for the complete target region between the 334 mammal reference sequences. Note that only one sequence was included for each species and thus intraspecific variability was not considered (see the discussion for further comments on this).

Verification tests

A number of different analyses were performed on DNA samples of known origin and DNA samples from authentic forensic casework to test the method and its ability to be applied in routine:

1) In order to verify the specificity of the universal PCR primers, DNA from 15 different species of known origin (Table 1) was amplified and sequenced, individually with Sanger sequencing as described below. The samples were provided by The Swedish National Veterinary Institute and The Swedish Museum of Natural History [19]. 2) Twenty two-species DNA mixtures and one four-species DNA mixture (Table 2 and Table S2), with known DNA template copy number ratios, were analyzed with deep sequencing using GS 454 Junior (Roche) as described in detail below. 3) The 454 deep sequencing method was applied on five samples from authentic forensic casework. The task was to detect the minor component of assumed DNA mixtures in all samples. The first sample was taken from a case with a man accused of animal cruelty of a dog. The question was if DNA from dog could be detected in the trace sampled from the man. The remaining samples were from a case with a human corpse that had multiple bite marks of an unknown “attacker”; here the question was “which species made the bite marks?”

DNA extraction, mtDNA copy quantification and artificial mixtures

For the reference samples, DNA were extracted based on a previously described method [21], and the DNA from the two authentic case samples was extracted using a Chelex method [22]. The number of mtDNA copies in a DNA extraction was quantified using an in-house developed assay based on quantitative real time PCR. The quantification method utilized SYBR Green I chemistry using the same primer pair as for the 454 PCR amplification protocol described below (Table 3). Each PCR reaction contained 12.5 μl 2x QuantiFast SYBR Green PCR Master Mix (Qiagen, UK), 1 µl of each primer (800 nM final conc.), 8.5 μl RNase-free water and 2 μl of DNA template for a total reaction volume of 25 μl. Thermal cycling was performed in an iQ™5 multicolor Real-Time PCR Detection System (Bio-Rad) using conditions recommended in the QuantiFast protocol (Qiagen). Plasmids (GenExpress, Germany), with a known number of copies, containing the targeted DNA sequence for humans, were used as a standard reference. All samples were quantified in duplicate and the mean value was used.

Pooled DNA samples from different species were used to create artificial DNA mixtures either in a 1:1 or in a 99:1 mtDNA copy number ratio.

Sanger sequencing and 454 sequencing

For the Sanger sequencing, the targeted DNA sequence was amplified using the PCR primers, presented in Table 3, tailed with M13-adaptors. The PCR amplicons were sequenced by standard Sanger sequencing using BigDye® Direct Cycle Sequencing kit (Applied Biosystems) according to the manufacturer’s protocol. Separation, by capillary electrophoresis, was performed on an ABI3500 XL instrument using POP-7 (Applied Biosystems) and the DNA sequences were called using Sequence analysis software v. 5.4 (Applied Biosystems).

For the 454 deep sequencing, the HPLC-purified PCR primers (Biomers) were used with adaptors suitable for the 454 sequencing chemistry on the 454 GS Junior (Roche) (Table 3). Barcodes were included in the primers to aid the ability for sample multiplexing (Table S3). The DNA was amplified using an in-house developed assay as follows. Each PCR reaction contained GeneAmp 10x PCR Buffer (Applied Biosystems), 3 mM MgCl₂, 0.5 mM Nucleotide+Uracil, 2 % Glycerol, 0.016 % BSA, 1.25 U AmpliTaq Gold DNA polymerase (Applied Biosystems), 0.05 U Uracil-DNA Glycosylase (USB corporation), 800 nM of each primer and 1 μl of DNA template (1,000-10,000 mtDNA molecules), for a total reaction volume of 25 μl. Thermal cycling was performed in a GeneAmp^® PCR System 9700 (Applied Biosystems) with touch-down PCR using the following parameters: 10 min at 37 °C, 5 min at 95 °C, followed by 10 cycles of 30 s at 94 °C, 30 s at 67 °C (minus 1 °C per cycle), 30 s at 72 °C and 30 cycles of 30 s at 94 °C, 30 s at 58 °C, 30 s at 72 °C with the final extension for 10 min at 72 °C.

The PCR products were purified using AMPure (Beckman Coulter) following the manufacturer’s protocol. The PCR products were then quantified using KAPA SYPR^® FAST Bio-Rad iCycler qPCR kit (KAPABiosystems) according to the manufacturer’s protocol. Thermal cycling was performed in an iQ™5 multicolor Real-Time PCR Detection System (Bio-Rad) using conditions recommended by the manufacturer. All samples were quantified in duplicates and the final PCR product was diluted to 10⁸ molecules per µl.

For the emulsion PCR (emPCR) and sequencing on 454 GS Junior, the manufacturer’s protocols were followed with reduced volume of primers (1/2 of the suggested volume) and the addition of 20 molecules per bead compensating for the short length of the PCR amplicon.

Bioinformatics

Sff-files extracted from the 454 GS Junior instrument were converted to FASTQ-files using the web-based tool Galaxy (https://main.g2.bx.psu.edu/). Reads with low quality were removed followed by the separation of the reads originating from different PCR:s by filtering on the barcode sequences. The barcodes were further removed, together with primer sequences, from the reads using Tagcleaner (http://edwards.sdsu.edu/cgi-bin/tagcleaner/tc.cgi). The resulting reads were grouped by collapsing identical reads. Sequences with less then 20 identical reads were removed from further analysis unless otherwise stated. The remaining sequences were searched against sequences in GenBank using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi), in order to identify the origin of the unknown DNA-sequences. Data described herein is available from the Sequence Read Archive (SRA) with the accession number SRA107543.

Design of the target and phylogenetic analyses

The DNA sequences for all 334 mammal species were aligned and the primer site sequences were extracted and compared with the universal primer sequences (Table 3, modified from [19]). There were high homologies among the reference species for the majority of the nucleotide positions, both for the forward and the reverse primer sequences (Table 4). Further analysis showed that 97 % of the sequences from the reference species had primer sequences that were identical, or only with one nucleotide difference, to the universal primer.

The nucleotides in majority were as follows for the forward primer; gacgagaagaccctatggagC (5’-3’) and tccgaggtcAccccaaccTCCG (5’-3’) for the reverse primer, which are identical to the primer sequences used for the PCR amplification (Table 3). Thus, based on this phylogenetic study, the selected universal primer sequences should be able to produce PCR amplicons for further analysis for the largest proportion of mammal species.

Further analysis showed that the length of the DNA sequence between the primer sites varied (Figure 1) among the reference species, with an overall mean of 75 base pairs (SD = 3 bp).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Distribution of the length (in base pairs) of the target sequence among 334 mammal species.

https://doi.org/10.1371/journal.pone.0083761.g001

The mammal reference DNA sequences were further compared for all species in pairs, to count the number of nucleotides that differed between the sequences of different species. Figure 2 displays the distribution of the number of nucleotide differences for the target amplicon with an overall mean number of pairwise differences of 31.5 (SD = 7). For the purpose of this method it was, however, of most interest to measure the ability to separate different species. Based on the reference sequences from the 334 different mammal species, only 0.06 % of the pairwise comparisons resulted in identical reference sequences. Thus, over 99.9 % of the species could be distinguished using the present assay. Species with identical sequences are listed in Table S4. Furthermore, a tree based on the total number of pairwise differences, for the complete target sequence, is shown in Figure S1. Closely related species tend to cluster, with only a few nucleotide differences for the complete target.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Distribution of the number of pairwise differences (in base pairs) for the target sequence among 334 mammal species.

https://doi.org/10.1371/journal.pone.0083761.g002

Verification of the universal primers

The ability of the universal PCR primer pair to amplify DNA was tested with samples from a variety of different species using PCR amplification followed by Sanger sequencing. DNA sequences were successfully obtained for all 15 mammal species tested (Table 1). These sequences were confirmed to match the corresponding reference sequences in GenBank.

Species determination of mixed samples

Table 2 presents the result from experiments with artificial mixed DNA-samples of either 1:1 or 1:99 mtDNA copy number ratios. All species were successfully detected in the 1:1 mixtures, while for some of the 1:99 mixtures, only the DNA from the species in majority was observed (Table 2). For the non-human DNA mixtures, the ratio of the obtained reads was similar to the expected ratio, although large variations were seen (Table S2). The mixtures that included DNA from human resulted, however, in a decreased number of human reads, approximately with a factor of 10-20 times in comparison with the expected ratio (Table S2).

Error rates

Since the reference sequences were known, experiments were performed in order to estimate error rates based on the DNA sequences obtained from the analyses. In the first test with DNA from elk, Alces alces, a total of 11,683 reads were obtained. 10,742 of these reads (89.6 %) were identical with the reference sequence for elk, whereas the remaining 1,211 reads (10.4 %) consisted of 211 sequence variants. The mean length of these sequences was 72 bases, thus the error rate was estimated to be 0.0014 errors per base.

The error rate is known to be sequence specific, especially for sequences containing homo polymers when using 454 sequencing chemistry [18]. The human, Homo sapiens, target reference sequence contains a stretch of five adenosine bases. The second experiment, with DNA from human, resulted in 5,268 reads of which only 66 % were identical with the reference human sequence and out of the erroneous sequences 78 % consisted of a sequence variant with four adenosine bases. In total, the error rate was estimated to 0.0047 per base, thus almost five times higher than for the elk experiment. As discussed in more detail below single nucleotide errors are, however, not important for the purpose of the method described here since the complete sequence of the target amplicon is used for the search and matching.

Samples from authentic forensic casework

In the first authentic case, which consisted of a man accused of animal cruelty, DNA from both human, Homo sapiens, and the genus Canis (including the subspecies domestic dog and wolf) was detected in the sample (Figure 3 In the second case, the four different samplings from bite marks, were analyzed in one single 454 sequencing run with equal input of PCR amplicons. DNA from human, Homo sapiens, was detected in all samples and DNA identical to the reference sequence for the genus Canis was additionally found in two of the samples (Figure 3). The presence of DNA from Canis was consistent with other findings in both cases. However, a notable observation was that the presented deep sequencing method enabled detection of small proportions of the minor component (<1 %).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Results from experiments with DNA samples from authentic forensic cases using the 454 deep sequencing method.

The number of reads obtained after analyses of five samples from authentic forensic casework are presented. Case 1 was a sample taken from a human accused of animal cruelty and case 2 was four samples taken from a human corpse with multiple bite marks.

https://doi.org/10.1371/journal.pone.0083761.g003