Design of the novel Zinc-Finger TaqMan assay
In order to establish the level of sequence conservation of the Zinc-Finger gene within the elephantine taxa, we aligned the known ZFX/Y regions for both alleles and each genus using Geneious R9. For the Asian elephants (Elephas maximus), we used the previously published Sanger sequences24 (Supplementary Table S1). For the woolly mammoths (Mammuthus primigenius) and the African elephants (Loxodonta africana and Loxodonta cyclotis), due to the lack of actual Zinc-Finger sequences deposited in sequence databanks, we recovered the corresponding sequences via the mapping of published whole-genome NGS reads from known male specimens20,29 (Supplementary Table S2). This alignment shows the complete conservation of the signatures discriminating the ZFX and ZFY alleles in Asian elephants at the scale of the elephantine subfamily. Low coverage data of this region (4X) are also available for the American mastodon (Mammut americanum; Supplementary Table S2), an extinct proboscidean species, which is a quite distant relative to the elephantine taxa: their most recent common ancestor dates back to 25–30 Mya (clade Elephantimorpha30). The comparison with the elephantine sequences strongly suggests the antiquity of these allelic signatures within the proboscideans (Supplementary Fig. S1). Conversely, when added to our comparison, the overlapping ZFX/Y sequences of modern humans show several fixed divergent positions from the elephantids (Fig. 1).
Alignment of the Zinc-Finger amplicon of interest for the ZFX and ZFY alleles from humans and elephantine taxa: Loxodonta (African elephants), Elephas (Asian elephants) and Mammuthus (mammoths). The top sequence represents the elephant ZFX allele; identities are indicated by dots. Primers and MGB probes are displayed in annealing position.
We designed one pair of primers: ZF_Forward (5′-ACAAAATGGTGCATAAGGAAAAG-3′; Tm = 58.9 °C) and ZF_Reverse (5′-CTCAGCTGTCTCGTATTCACA-3′; Tm = 60.3 °C), which promote the amplification of a 74 bp long amplicon surrounding two sex-specific polymorphic sites. We chose priming sites exhibiting fixed divergent positions with human ZFX/Y sequences—specifically the final 3′ position of the forward primer—to reduce the risk of amplification of human contaminants. Based on the melting temperatures of the chosen primers, we designed two sex-specific Minor Groove Binding (MGB) fluorescent probes diverging from each other by two of their 13 nucleotides (Fig. 1): ZFX 5′-VIC/AGCCAACAAAATG/NFQ/MGB-3′ (Tm = 69.0 °C) and ZFY 5′-FAM/ATCCAGCAAAATG/NFQ/MGB-3′ (Tm = 68.8 °C), labelled with the two fluorescent dyes used by default in bi-allelic discrimination31, and manufactured by Applied Biosystems (Foster City, CA).
In vitro sensitivity experiments
To address the sensitivity of our assay, we first generated sex-specific quantitative standards: we diluted a male mammoth DNA extract (Lyakhov mammoth; Supplementary Table S4) until the point when real-time PCR reactions using this dilution as a template would only yield the amplification of one or the other sex-specific allele (or no product at all). We pooled three reactions for which only the X allele was detected in one microtube, and three other Y-positive reactions in another microtube. Each pool was purified using the minelute PCR purification kit (Qiagen, Venlo, NL) and concentrated separately in 10 µl of EBT buffer (Qiagen EB buffer supplemented with 0.05% Tween-20). We quantitated each sex-specific standard using the Qubit High Sensitivity assay kit (Invitrogen, Waltham, MA) and prepared a tenfold dilution series ranging from 1010 copies down to 10−1 copy per µl. Standard series were stored in frozen aliquots and thawed only before use.
We analyzed the sensitivity of the assay in two dimensions: (I) the sensitivity of the PCR amplification in absolute copy numbers and (II) the relative sensitivity of both X- and Y-specific allele diagnostics. We first tested the general sensitivity of the assay using a SYBR Green I approach, with 1X Sso-Advanced Supermix (Bio-Rad, Ipswich, MA) and a standard series of each allele (105 down to 10−1 each), using 6 replicates of the standards at the low end (2 × 100 and 2 × 10−1). We then evaluated the reciprocal sensitivity of each MGB probe via a TaqMan reaction using a standard series ranging from 105 down to 100 each, with three replicates for the latest. For probe-based PCR reactions, we used the dedicated TaqMan Fast Advanced Master Mix (Applied Biosystems, Foster City, CA) which contains dUTP and Uracyl-N-glycosylase (UNG) pre-treatment steps to avoid PCR contamination from carryover PCR products.
Quantitative PCR optimization and genotype analyses
We compared the behaviour of TaqMan reactions with various combinations of primer concentrations between 400 and 900 nM, final probe concentrations ranging between 200 and 600 nM, and an annealing/extension temperature gradient (55–65 °C). The best sensitivity was obtained around 60.5 °C regardless of the reagent concentrations: the Cq of the standards were retarded by up to 0.8 or 1.2 cycles when lower or higher temperatures were picked, respectively. Balanced MGB probe concentrations systematically yielded a higher response of the FAM probe over the VIC one (up to 150%), and sometimes caused a shallow crosstalk-signal artifact within the VIC detection range. Implementing uneven probe concentrations—increasing VIC by one-third and lowering FAM by as much—addressed both issues. We thus adopted the following conditions for all subsequent experiments: final reaction volumes of 15 µl with 1X of TaqMan Supermix, 800 nM of each primer, 375 nM of Y-FAM probe, 525 nM of X-VIC probe, and 1–2 µl of DNA extract.
We performed all PCR reactions on a CFX-96 real-time thermocycler (Bio-Rad, Ipswich, MA) using the following 2-step conditions: after a first denaturation of 2′ at 95 °C, we performed 40 cycles of 95.0 °C 10 s and 60.5 °C 35 s. We conducted the allelic discrimination from the qPCR output with the CFX-Manager software v3.1 (Bio-Rad, Ipswich, MA) using the following set of parameters: baseline subtracted curve fit, quantification cycle (Cq) determined via a single threshold set to 10% of average plateau fluorescence (measured in Rescaled Fluorescence Units, RFU), call of alleles on the last PCR cycle.
Specificity analyses
We investigated the level of specificity of our assay against human contaminants via straight qPCR attempts with various concentrations of control human genomic DNA (Thermofisher, cat. number 4312660): 1, 5, and 25 ng per reaction. We complemented this analysis with an in silico assessment of our assay: we used BLASTn32 to analyze the ‘nr’ collection database in GenBank, and identify which taxa shared sequence identity with at least one of our primers. Among those hits, we focused on the putative sympatric taxa of elephantids (modern and extinct) for which we aligned the available ZFX/Y fragments.
Although the risk of non-specific detection is extremely low with an MGB-TaqMan methodology31, we chose to monitor the specificity of PCR design in our case study experiments. We prepared two pools—one per case study—from all positive PCR reactions from actual specimens across an entire replicate series. We transformed these pools in double-indexed Illumina libraries33 and performed a shallow sequencing of each (in paired-end 2 * 75 bp).
Case study on elephant fecal extracts
We conducted the fecal sampling of wild elephants from November 2016 to January 2019 in Sebitoli area in the vicinity of Kibale National Park (south-western Uganda). The wildlife of this forest area, located at the north of the protected area, is studied by the Sebitoli Chimpanzee Project/Great Apes Conservation Project and the Muséum national d’Histoire naturelle (MNHN, Paris, France). Commercially logged in the 1970s, the Sebitoli forest is now composed of 70% of regenerating forests and only 14% of old-growth forest34. In areas adjacent to Kibale, human population density is high35 (circa 300 inhabitants/km2). They grow monocultures such as tea fields, eucalyptus, and banana plantations as well as crops like maize, which attract elephants and primates out of the forest. This survey is part of a project aiming at mitigating the human-wildlife conflict at the edge of the protected area in the framework of the Memorandum of Understanding SJ 445-12 between MNHN, Uganda Wildlife Authority, and Makerere University in Uganda and the MoU between UWA and GACP.
To avoid the repeated sampling of the same individuals, we collected only once when we encountered several dung boli of similar size on the same day and location. Since female elephants live in close family groups36,37—while the adult males are mostly solitary—this strategy made the sampling of male dung more likely than female ones. A quantity of 10 to 15 g of feces was stored in 70% ethanol for 24 h. After removing the supernatant, feces were placed in gauze on silica gel beads and stored at ambient temperature until processed in the laboratory. After removing the largest vegetal compounds, between 150 and 200 mg of dried feces were extracted with the Power fecal DNA Isolation Kit (MoBio, Carlsbad, CA). The DNA extraction was performed in France, at the modern lab of the ‘Plateau de Paléogénomique et Génétique Moléculaire’ (P2GM platform) from the MNHN. Total DNA yields from the extracts, as measured with a NanoDrop 2000 (ThermoFisher Scientific, Waltham, MA), ranged from 2.9 to 186.7 ng/μl (Supplementary Table S3).
To validate the assay, we used a set of 12 elephant extracts for which sex was known a priori: six male and six female specimens. We then implemented our assay in a case study that involved 91 specimens of unknown sex. Two PCR replicates per individual extract were performed, in parallel with a total of 7 PCR negative controls (NTC for ‘No Template Control’ reactions).
Case study on mammoth ancient DNA
Over the last 20 years, we have gathered several dozens of woolly mammoth samples that have been used in various paleogenetic analyses38,39. They are part of a broad comparative genomics project of diachronic specimens from Beringia which objective is to address the diversity and gene flow throughout the Late Pleistocene populations of woolly mammoths. Here we attempted to derive the genetic sex for a subset of 29 specimens using the novel assay. These samples all come from the Late Pleistocene in Siberia, and the radiocarbon-dated specimens range from 4420 up to beyond 50 ky BP (Supplementary Table S4).
DNA extractions and PCR setup of mammoth samples took place in the dedicated ‘ancient DNA cleanroom’ at the P2GM platform, which is physically isolated from the modern lab. We used a protocol previously published for DNA extraction from bone39 and extracted the specimens in 5 different series—each along with one extraction blank. We first tested six specimens of known sex (thanks to a morphological diagnosis): Lyakhov, Jarkov and Oymiakon (all males), 2001/174, Lyuba and Khroma (all females). We then implemented the assay on 23 extracts of unknown sex together with each extraction blank, several NTCs, and one absolute standard series to establish the number of template molecules for each X and Y allele available from our mammoth extracts.
Our sexing assay relies on the identification of one homozygous genotype (female) and one heterozygous genotype (male) via a bi-allelic target. In such a design, the risk of false assignation of a male to the female genotype due to allelic dropout of the Y allele is a limitation, particularly when working with templates of low DNA content40,41. We carried out all mammoth PCR reactions in triplicates, to comply with the multi-tube strategy developed to control for that risk. The implementation of a quantitative PCR framework in our sexing assay provided us with the ability to refine the estimate of the accuracy of the genotypes. Taberlet et al.41 showed that (i) the allele amplification of a bi-allelic marker behaves stochastically for very dilute samples, and (ii) for a known amount (U) of diploid genome copies in a reaction, the probability of allelic dropout can be precisely modeled (Supplementary Fig. S4). We posited that the sum of ZFX and ZFY allele copy numbers per reaction inferred via qPCR is a relevant proxy of this amount among our samples—a reasonable assumption when one considers that Zinc-Finger is a single copy nuclear gene. We derived the absolute copy number (CN) based on the Cq calculations for both Y-FAM and X-VIC between the positive specimens and the corresponding standard series. We then used this metric to estimate the probability PXX of allele dropout per reaction for a true heterozygote, based on Taberlet et al.’s model. For each specimen, the theoretical risk of wrongly being deemed a female due to allelic dropout thus translates as (PXX)n from the binomial distribution of parameters n and PXX, where n refers to the number of PCR attempts that yielded a genotype (Supplementary Table S6).
Source: Ecology - nature.com
