Page 90 - Read Online

Longhi et al. Microbiome Res Rep 2024;3:4 https://dx.doi.org/10.20517/mrr.2023.02 Page 5 of 15

[27]
filtering tables reported below) . Retained reads were taxonomically classified through the
[28]
METAnnotatorX2 pipeline using a set of databases of reference genomes whose taxonomy was previously
[29]
validated to maximize the accuracy of homology-based taxonomic classification of reads . For those
samples which, after the filtering step, reached a very low number of reads (e.g., gastric biopsies with ~200
reads), we verified that read depth was sufficient for comparisons. For this purpose, we considered the
samples with the higher number of reads, such as sputum, saliva, and skin-treated samples (average of 191
high-quality reads). We reanalyzed them utilizing a subset of 200 reads, which returned comparable
microbial profiles [Supplementary Table 1].

RESULTS AND DISCUSSION
Evaluation of bacterial and human abundance in biological samples with high content of eukaryotic
DNA
To first estimate the effectiveness of the best approach for the bacterial DNA extraction aimed at reducing
the carryover of human host DNA, we tested different amounts of saponin (2.5%, 1%, 0.1% and
0.0125% wt/vol) on a saliva sample collected from a healthy human donor. Saliva was chosen as an optimal
test case due to the consistently high percentage of human DNA (~90%) determined by shotgun
metagenomic sequencing . Treatment with saponin aimed to induce lysis of eukaryotic cells, followed by
[8]
adding DNase to the DNA extract to remove eukaryotic DNA before bacterial DNA extraction with a
commercial kit supported by preliminary mechanical bead-beating lysis. Specifically, to assess the bacteria-
human DNA ratio, qPCR was performed on the total DNA obtained using specific PCR primers targeting
the bacterial 16S rRNA gene. In contrast, the human DNA level present in the saliva sample was determined
by absolute quantification of the β-globulin gene . In addition, this analysis was performed on DNA
[25]
extracted from the same saliva sample that did not undergo the additional depletion process as a reference
control to rule out if this saponin-based protocol can represent a potential cause of missed/additional
bacterial detection. As a negative control, we used a sterile water sample treated with the same depletion
protocol to ensure the extraction reagents did not contain contaminations. All qPCR analyses were
performed on biological and technical triplicates.

Across all saponin amounts used, 2.5% wt/vol of detergent was found to successfully deplete most of the
human DNA [Figure 1A]. Indeed, in a saliva sample treated with 2.5% wt/vol of saponin, the human DNA
decreased to a concentration of 8.01 gene copy number/mL, while for the untreated sample, the host’s DNA
recovered was around 71.3 gene copy number/mL [Figure 1A]. Conversely, host DNA abundance in the
same saliva samples treated with 1%, 0.1%, and 0.025% wt/vol of saponin remained relatively stable at a
concentration of 20-30 gene copy number/mL [Figure 1A]. Furthermore, the 2.5% wt/vol saponin-treated
sample showed a 5.17 × 10 gene copy number/mL bacterial content compared to the untreated saliva in
8
which bacterial DNA had a concentration of 2.43 × 10 gene copy number/mL [Figure 1B]. These findings
9
showed that saponin lyses human cells by reducing the eukaryotic DNA content within the sample, but high
amounts of such detergent (e.g., 2.5% wt/vol) also acted on bacterial cells, decreasing in part their
abundance. In addition, by comparing the saliva sample treated with the lowest concentration of saponin
(0.025% wt/vol) to its untreated counterpart, we noticed a reduction in bacterial DNA, probably due to the
extraction protocol, but which does not appear to be statistically significant. Considering these preliminary
results, detailed study-specific investigation of such a range of different concentrations of saponin on the
retrieved DNA is mandatory. Thus, given the impact of saponin on the bacterial content, we suggest
preliminarily testing the biological matrix according to the purpose of the study to assess the optimal
quantity of saponin to be used. Specifically, treatment with saponin is risky if applied to biological samples
originally low in bacterial content because the latter could be totally lost. On the other hand, this data
showed that depletion protocol may be useful if applied to samples with a high content of bacteria to reduce
the contamination of host-derived DNA.