Page 47 - Read Online

Page 12 of 17 Wang et al. Microbiome Res Rep 2024;3:39 https://dx.doi.org/10.20517/mrr.2024.21

[37]
genomes of diverse origins to establish a prokaryotic AMP database, AMPSphere , which consists of
863,498 predicted AMPs. To evaluate the impacts of a metaproteomic workflow for AMP identification, we
re-searched both DDA and DIA datasets using the AMPSphere database concatenated with all identified gut
microbial peptide sequences. Consistent with total and small protein identification, DIA-PASEF showed a
higher AMP identification (200~300 AMPs, excluding the FASP-3K group) compared to DDA-PASEF
methods (50~150 AMPs) [Figure 3E and Supplementary Figure 7]. When looking at the relative abundance
of AMPs within each sample, it appears that differential centrifugation decreases the abundance of AMPs
[Figure 3F]. Evaluation of the sample-wise Pearson’s correlation of the quantified AMP intensities indicates
higher correlations intra-groups compared to inter-groups in both DDA and DIA datasets [Supplementary
Figure 2E and F]. The findings suggest that NC samples with digestion methods of either in-solution and
FASP-10kDa followed by MS analysis with DIA-PASEF mode resulted in high AMP diversity observed at a
high relative abundance.

Over-representation of Muribaculum in mouse fecal metaproteome with differential centrifugation
We next evaluated whether different sample preparation methods and MS data acquisition mode impact the
taxonomic profiles using metaproteomics. By using a threshold of a minimum of three distinctive peptides
for confident taxon identification, this study identified 12 phyla, 13 classes, 21 orders, 23 families, 62 genera,
and 96 species in the DDA-PASEF dataset, and 17 phyla, 22 classes, 19 orders, 32 families, 70 genera, and 96
species in the DIA-PASEF dataset. Here, we selected the phylum and genus levels for the evaluation. No
obvious difference in the abundance distribution of abundant taxa was observed between DDA and DIA
datasets at both phylum and genus levels [Figure 4]. The major differences were observed between DC and
NC, which is in agreement with previous studies of both human and mouse microbiomes [30,40] . A higher
relative abundance of Chordata or Mus (host) was observed in NC groups compared to DC groups, in
particular at the genus level [Figure 4A-D]. Interestingly, we observed a difference in Mus (host) relative
abundance between different protein digestion methods in DDA-PASEF data, but not in DIA-PASEF data
[Figure 4C and D, Supplementary Figure 8]. On the contrary, the microbial compositions were highly
consistent in both DDA-PASEF and DIA-PASEF datasets.

Metaproteomics analysis demonstrated that Bacillota (previous Firmicutes) and Bacteroidota (previous
Bacteroidetes) were the two predominant phyla in mouse feces, but the relative abundances of these two
phyla were dramatically altered by the use of differential centrifugation during sample pre-processing.
Marked higher relative abundances of Bacteroidota and lower abundances of Bacillota were observed in DC
samples compared to NC in both DDA and DIA datasets. This is in agreement with previous studies on
mouse metaproteomes . However, the direct opposite of changes was reported in human microbiomes
[40]
where lower levels of Bacteroidota and higher levels of Baccilota were obtained when samples were prepared
using differential centrifugation . This difference might be due to the known marked different bacterial
[30]
genus/species compositions of mice and human microbiomes. As shown in Figure 4E and F, Muribaculum
was the genus mainly driving the elevation of Bacteroidota in DC samples, which represent ~70% of the
microbial abundance in the samples. Muribaculum species are known to be dominant in mouse gut
microbiota, but were only recently well characterized and demonstrated to have very high host preference
(prevalence of 67% in mice compared to 7% in human gut) . In contrast, in human gut microbiota, the
[41]
genus Bacteroides is usually the most abundant in Bacteroidota . In this study, we found that the relative
[42]
abundance of Bacteroides was lower in DC compared to NC, which is in agreement with the observations in
human gut microbiota. It is unknown why Muribaculum displayed different responses to DC compared to
their neighboring genera within the sample phylum Bacteroidota, but this observation suggests that sample
preparation methods need to be optimized for microbiomes of different origins and the microbial species of
interest for a particular study.