To copy a target DNA fragment with Polymerase chain reaction (PCR), typically 1-150 ng Template DNA is used in the reaction (see e.g. 5Primer HotMaster Taq manual). We use the Cytochrome c oxidase subunit I (COI) barcoding gene to identify hundreds of mostly insect specimens in bulk samples, using DNA metabarcoding. However, when amplifying hundreds of specimens in the same reaction, DNA template concentration suddenly becomes a concern! At least a few target DNA fragments from each individual specimen from the bulk sample should be present in PCR, so all individuals can be detected. While we started our first PCRs with 100 ng DNA per PCR reaction, we gradually reduced the amount of template DNA to 50, 25 and even 12.5 ng, to reduce the effect of PCR inhibiting substances present in the DNA extractions. If you use less DNA in your PCR, also less inhibitors are present!
Additionally, the PCR reaction volume can be increased to reduce PCR inhibition, but is more expensive and complicated to pipet in the lab. We successfully amplified 12.5 ng of problematic samples using 250 µl PCR mix per reaction. While we prepared volumes of 250 ul, the samples were divided equally onto 5 PCR stripes prior to PCR and pooled again afterwards.
The easiest and most convenient solution to combat PCR inhibition would just be to use less template DNA in a standard 50 µl PCR reaction volume. But would you start missing specimens in your bulk sample? If you use so little template, there might not be a single COI template DNA fragment present for certain specimens, simply due to stochastic effects.
In the following, I will try to estimate the amount of ng DNA needed in a PCR reaction to amplify the COI barcoding gene of a single insect specimen. Based on these results we will additionally discuss the amount of DNA needed for metabarcoding insect bulk specimens, wich are affected by additional problems.
Minimum amount of DNA needed for insect barcoding (singe specimen)
Most organisms are diploid, and thus most genes are present with 2 copies in each cell (ignoring gamets, and multi-copy genes here). However, the COI barcoding gene we are interested in lies within the mitochondria, which are present in each cell (again ignoring exceptions, like red blood cells in vertebrates) hundreds to thousands of times and contains one genome copies in each individual mitochondria (bacteria can contain several copies of their genome, but we will calculate with n=1 here). For simplicity, lets assume each cell contains 1.000 mitochondrial genomes (there are many papers reporting mitochondrial copy numbers of < 1.000 MT per cell, but publications are from a wide range of tissues and non insect taxa).
I have downloaded genome sizes for 925 insect taxa from genomesize.com (February 2016). Genome size is given as C-values, wich is the amount of DNA (in picograms, pg) contained within a haploid nucleus. As our DNA extractions contain DNA from mostly diploid cells, we have to double all C-values to get the actual genome sizes. The downloaded dataset contained mostly genome sizes from Coleoptera, Diptera, Hymenoptera and Odonata. The mean genome size is 2.465 pg, but the median of 1.04 pg does represent the common genome sizes probably better, as the mean is strongly biased by Orthoptera with comparatively large genomes.
Thus, when extracting DNA from a single insect specimen, 1.04 pg DNA (= one “complete” diploid genome) contains ~1.000 mitochondrial genomes. Thus 1.04 pg DNA should be plenty, for amplifying the mitochondrial COI gene, but is not enough for targeting nuclear genes!
1 picogram (pg) converts to 0.001 nanogram (ng), thus 1.04 pg DNA is 0.00104 ng DNA. Thus, 1 ng DNA which is on the lower end of template DNA used in PCR translates into ~ 961.000 mitochondrial genomes which can be potentially used to amplify the COI gene.
Implications for DNA metabarcoding of bulk samples
One nanogram of DNA roughly contains 1.000.000 mitochondrial genomes. In a bulk insect sample with e.g. 1.000 specimens that could correspond to 1.000 mitochondrial genomes per specimen, if they all have similar biomass. Unfortunately, insect biomass in complete environmental samples often varies 3 – 4 magnitudes, which translates directly into amount of DNA contributed and amount of sequences recovered (see Elbrecht and Leese, 2015 PlosONE, Figure 2). Depending, on amount of taxa collected and extracted in the bulk sample, 1 ng DNA can be already dangerously low if individuals in the sample varies in biomass.
Thus 10 ng should probably be considered as the minimum amount DNA used in PCR for DNA metabarcoding bulk samples, giving a 10x “buffer” of template DNA based on the discussion above. Of cause, these are all estimates, and samples with less DNA might work just fine. However, this issue of insufficient template DNA amount is rarely addressed or tested prior to sequencing. Further DNA quality can additionally decrease amplification success (if the DNA his highly degraded, DNA concentration is not representative of amplifiable template DNA). Unlike sequencing depth, wich can to some degree be tested using rarefaction curves, it’s difficult to estimate if sufficient template was used in PCR as specimens just “disappear” is they were not amplified.
Maybe we will test the calculations made here with some experimental data at some point. Amount of template DNA might just be another little thing to look out for when designing metabarcoding and mitogenomics projects. Also ways to remove PCR inhibitors as well as test alternative DNA extraction protocols might be something to consider. We did use DNA salt extraction + RNAse digestion + Qiagen min elute reaction cleanup kit on ground bulk insect samples, and still experienced PCR inhibition for some samples when using 12.5 ng template DNA. Lot’s of things to test and think about!